Unwired Planet International Ltd v Huawei Technologies Co. Ltd & Anor

UnwiredPlanet have a worldwide patent portfolio which includes numerous patents which are declared essential to various telecommunications standards (2G GSM, 3G UMTS, and 4G LTE). Most of the relevant portfolio was acquired from Ericsson. UnwiredPlanet’s business is licensing those patents to companies who make and sell telecommunications equipment such as mobile phones and infrastructure. This action began in March 2014 when UnwiredPlanet sued Huawei, Samsung and Google for infringement of six UK patents from their portfolio. Five were claimed to be SEPs (see below). UnwiredPlanet contended their patents were infringed and (so far as relevant) essential.

The dispute was managed to consist of a series of trials, docketed to me. First would be five “technical” trials relating to the validity and infringement/essentiality of the six patents (two patents are divisionals). These were called trials A to E and were to run from October 2015 to July 2016. The patent in trial A was patent EP (UK) 2 229 744 which is for an invention concerning poll triggers, the patents in trial B were divisionals EP (UK) 2 119 287 and EP (UK) 2 485 514 which are for an invention related to selfconfiguring networks, the patent in trial C was EP (UK) 1 230 818 which relates to inter-RAT handover, the patent in trial D was EP (UK) 1 105 991 which related to Hadamard codes, and the patent in trial E was EP (UK) 0 989 712 which relates to network messaging. The 712 patent was not said to be a SEP.

After the five technical trials would be a non-technical trial in autumn 2016. The nontechnical trial was to address all the competition law issues as well as the FRAND issues (see below), injunctive relief (if any) and, by a later direction, damages for past infringements. Finally, and only if necessary, there might be a further trial to deal with some outstanding questions including pass through licences.

This judgment arises from the non-technical trial and relates to the patents which are or are said to be Standards Essential Patents or SEPs. Part of the process of standardisation involves holders of patents which are essential to an international telecommunications standard declaring them as essential to the relevant standards body. In this case that body is the European Telecommunications Standards Institute (ETSI). The ETSI IPR Policy requires that a patentee declaring patents as essential to a standard commits to licensing those patents on FRAND terms. FRAND stands for Fair Reasonable And Non-Discriminatory. The judgment mostly concerns FRAND.

After proceedings began, in April 2014, UnwiredPlanet made an open offer to the defendants to license its entire global portfolio (SEPs and non-SEPs). The defendants denied infringement/essentiality and contended the patents were invalid, counterclaiming for revocation. So, they said, no licence was needed. They also contended that UnwiredPlanet’s offer was not FRAND. In addition, Huawei and Samsung raised defences and counterclaims based on breaches of competition law. This involved both arguments about Art 101 of the Treaty on the Functioning of the European Union (TFEU) relating to the Master Sale Agreement (MSA) whereby UnwiredPlanet acquired patents from Ericsson and arguments about Art 102 TFEU concerning abuse of dominant position. The allegation that the offer was not FRAND was pleaded as a breach of competition law. These allegations were the subject of counterclaims against other companies in what was then the UnwiredPlanet group (the ninth and tenth defendants) as well as against Ericsson, who were joined as the eleventh

Sometimes in this case the terms 2G, 3G and 4G are used to refer to different standards and sometimes GSM, UMTS (or WCDMA) and LTE respectively. They are not the same but the distinction rarely matters. In this judgment I have tried to use the terms which reflect the way the argument and evidence went in any given context but it is impossible to be consistent. A complication is multimode handsets. A 4G/LTE handset will usually be able to work on earlier standards (2G/GSM and 3G/UMTS). It is therefore “multimode”. There can be exceptions and so calling a handset 4G or LTE can be ambiguous since it probably refers to a multimode device but might not. Again it is impossible to be consistent.

In June 2015 as a result of directions from the court which are considered further below, each side made certain open offers of licensing terms. UnwiredPlanet’s June 2015 proposals included offers of a worldwide SEP portfolio licence, a UK SEP portfolio licence (the UK portfolio consists of more patents than just the five SEPs in suit) and per-patent licences for any SEPs the licensee chose. The details do not matter at this stage but one point to note is that the royalties claimed for per-patent licences or a UK portfolio were higher than the global rate on offer. The rates all scaled by reference to the same global rate proposals as in 2014, i.e. a global rate of 0.2% for LTE and 0.1% for GSM/UMTS.

Huawei’s June 2015 proposal was for a per-patent licence limited to the UK SEPs in suit. The rates for all five SEPs together were 0.034% for LTE, 0.015% for UMTS and zero for GSM.

In the summer of 2015 and before trial A, Google settled as regards the SEPs. From then on they would only have played a role in the fifth technical trial (E) since that related to the implementation patent. By about April 2016 three technical trials had been completed and the parties agreed to postpone any further technical trials indefinitely. By that stage UnwiredPlanet had won two and lost one of the technical trials. Two of UnwiredPlanet’s patents had been found to contain claims which were valid and were essential to the relevant standards while the other two patents were held invalid. The results of all three technical trials are under appeal to the Court of Appeal. Also in about April 2016, the claimant company and the tenth party (UnwiredPlanet

In the summer of 2016 Samsung settled with UnwiredPlanet and Ericsson. As a result of the settlement, proceedings against Samsung ended and, with the court’s leave, Samsung’s competition law counterclaim was discontinued. This included discontinuance against the ninth party. Huawei, also with leave, then discontinued significant parts of their counterclaim, including all their counterclaims against

On 1^st August 2016 each side made new offers. UnwiredPlanet’s offers were on the same terms as before but with lower rates. UnwiredPlanet obviously felt able to offer lower rates at that stage at least in part because the royalty floor provisions in the MSA had been removed. The global SEP portfolio rate for 4G/LTE in this offer was 0.13%. The corresponding rates for GSM /UMTS were 0.065%.

For a UK SEP portfolio licence UnwiredPlanet’s August 2016 proposals are:

Huawei’s 1^st August offer was on the same UK only per-patent basis as before. The rates proposed were:

On 11^th October 2016, about two weeks before the trial, Huawei made a new licensing proposal. This amended the per-patent royalties on offer and also proposed a licence under the whole of UnwiredPlanet’s UK SEP portfolio. The UK portfolio rates were: i) for LTE: infrastructure 0.061%; mobile devices 0.059%; ii) for UMTS: infrastructure 0.046%; mobile devices 0.046%;

The August 2016 UnwiredPlanet proposals and October 2016 Huawei proposals represent the parties’ positions going into the trial, subject to a point referred to in this judgment as “hard-edged non-discrimination” arising from the settlement with Samsung.

By the opening neither side had actually set out complete terms of a licence they were prepared to enter into. Up to trial each side’s licensing proposals had been made in outline terms rather than based on a fully worked out licence. That was sensible up to a point since the parties took the view that they were so far apart on the main terms that spending time on the details was not worthwhile. Nevertheless in order to try to ensure

Simply stated the main dispute to be resolved is about whether and to what extent various terms on offer are or would be FRAND. One key battleground is the value of UnwiredPlanet’s patents which is reflected in the royalty rate. However that is not the only issue. There is a major dispute about the proper scope of any licence. The case also involves important questions of whether the April or July 2014 offers were FRAND and/or whether they amounted to an abuse of a dominant position by UnwiredPlanet contrary to Art 102 TFEU.

Depending on the outcome of the main dispute the question of an injunction to restrain patent infringement may arise together with the issue of whether Huawei have a defence to a claim for an injunction under competition law.

Also depending on the outcome of the main dispute, there may need to be a decision on appropriate reasonable royalty rates to be used in a calculation of back damages for the infringements committed by Huawei in relation to the two SEPs which have been found to be valid and infringed as well as the other three SEPs in issue in the technical trials (including the one found invalid in Trial B since they are subject to appeal).

In opening it emerged that there was a fundamental disagreement about the scope of the main dispute.

Huawei contended that the procedural position which had been reached meant that as a matter of principle a UK SEP portfolio licence was the inevitable and mandatory outcome. This necessarily also meant that an injunction would never be granted even if, contrary to Huawei’s case, UnwiredPlanet were entitled to seek an injunction and Huawei had no defence to an injunction under competition law. That was because the court was going to settle a UK portfolio licence.

This had not been apparent to me either from the written openings or the evidence and it came as a surprise to UnwiredPlanet, whose clear preference was for a global licence rather than a UK portfolio licence. Huawei argued that the result it contended for followed from a combination of three steps: first, Huawei had stated that it no longer intended to maintain that UnwiredPlanet was obliged to offer per-patent licences, second, therefore the only thing on offer from Huawei was a UK SEP portfolio offer and Huawei had undertaken to accept whatever royalty rate the court set for that licence, and third, a licence of that scope was one of UnwiredPlanet’s offers. Consequently, Huawei submitted, a licence of that scope must be the outcome. This was so even though it was obvious that what UnwiredPlanet really wanted was a global licence and even though global rates and the FRAND status of global offers were at the heart of the dispute. I will need to examine this issue in more detail below.

Aside from the clear dispute about the value of UnwiredPlanet’s patents and the concomitant FRAND royalty, the parties’ submissions as to the outcome of this nontechnical trial are as follows:

It is convenient to divide the issues I have to decide into three broad topics. The first is FRAND. This involves working out what FRAND is and the principles which apply to it. Then I need to resolve what royalty rates are FRAND in order to determine whether any of UnwiredPlanet’s offers were FRAND and if not, what would be FRAND? After that I need to resolve arguments about any other disputed terms in a FRAND licence. This first broad task represents the bulk of the work. The second broad topic is competition law, resolving Huawei’s case that UnwiredPlanet have abused their dominant position. Finally there are remedies – injunction, damages and declarations.

One of the challenges in trying this case was confidentiality. The arguments, evidence and disclosure documents included a large amount of material in which confidentiality was claimed. Some of the claims were from parties or companies who had been parties (Ericsson and Samsung) but some of the confidential material was confidential to third parties such as licensees. The legal representatives of all parties were privy to all the material but some aspects, e.g. material relating to Samsung or Ericsson, was maintained as confidential from Huawei or UnwiredPlanet staff. Attempting to determine the confidential status of material during the hearing would have been impossible, so the trial was conducted accepting many of the claims to confidence for the time being. While wide claims to confidentiality had been made before trial, they were reduced considerably at and during the hearing. Thus much of the trial took place

I made it clear that I wished to hold as much of case in public as possible. During the hearing there were occasional slips and parties who claimed their confidentiality had been breached wrote to the representatives. From my perspective the solicitors and counsel for all parties before me did an excellent job in tricky circumstances to conduct an open trial and pay due respect to claims to confidentiality. This judgment includes material about which some claims to confidence are maintained. I indicated that when the confidential draft judgment was distributed to the parties (UnwiredPlanet and Huawei) I was prepared to include the lawyers for Samsung and Ericsson. The lawyers for Samsung and Ericsson would not be entitled to discuss the draft with their clients but they would be able to make submissions about confidentiality. A public judgment, or public version of the judgment, could then be produced.

UnwiredPlanet called the following fact witnesses:

Mr Saru has been UnwiredPlanet's Vice President of Standards of Licensing since early 2013, and has been the Managing Director since February 2014. He joined UnwiredPlanet shortly before the MSA with Ericsson closed, and he was responsible for making the necessary declarations to ETSI. His evidence related to the licence negotiations between UnwiredPlanet and Huawei, including UnwiredPlanet's approach to determining what was FRAND, and specifically the MNPA (see below).

Mr Robbins was, until 1 July 2015, Executive Vice President and General Manager of the Intellectual Property Division of UnwiredPlanet. His evidence related to the relationship between UnwiredPlanet and Ericsson, including the MSA, and UnwiredPlanet's approach to FRAND and the offers it made in 2014, and the Lenovo licence.

Mr Ware is the Chief Executive Office of PanOptis Equity Holdings LLC. He founded the PanOptis Group in September 2003. On 30 June 2016, PanOptis acquired UnwiredPlanet LLC and UnwiredPlanet International Limited (and so the UnwiredPlanet portfolio of patents) from UnwiredPlanet Inc. His evidence related to PanOptis's business model and focus on long-term relationships with licensees. In particular, his evidence related to the negotiation and circumstances of the UnwiredPlanet/Samsung licence.

UnwiredPlanet also served Civil Evidence Act Notices in respect of: (i) a recording of the “UnwiredPlanet Management Webcast and Conference Call” on 6 April 2016 and a copy of an accompanying presentation entitled “UnwiredPlanet Corporate Transformation”; (ii) a Schedule 14A Form filed by UnwiredPlanet with the United States Securities and Exchange Commission on 20 April 2016; (iii) witness statements from Boris Teksler, UnwiredPlanet’s former Chief Executive Officer; (iv) witness statements from Gustav Brismark, Head of Region IPR and Licensing at Ericsson; and (v) three ETSI FAQs dated 22 December 2008, 10 July 2014 and 26 March 2015.

Huawei’s fact witnesses were Xuxin Cheng, Emil Zhang (also known as Xiaowu Zhang), and Chaobin Yang.

Mr Cheng has been the Deputy Director of Huawei's IP Department since February 2008, and Vice President for IP Licensing & Transactions. He gave his evidence through an interpreter. His evidence related to Huawei's position on FRAND, whether any steps have been taken to avoid infringing UnwiredPlanet's patents, and Huawei's conduct in licence negotiations.

Mr Zhang is the deputy director of Huawei's IP litigation department and has day-today conduct of this action. He gave his evidence in English with the interpreter standing by to help if need be. His evidence related to Huawei's conduct in the dispute, including the pre-action correspondence between Huawei and UnwiredPlanet and Huawei's level of sophistication in conducting litigation, and the geographical extent of Huawei's activities.

Mr Yang is the Vice President and Senior Marketing Officer for Huawei's Wireless Network product line. He has worked at Huawei's Shanghai R&D centre since 1998. He gave his evidence through an interpreter. His evidence related to the significance of certain releases of technical specifications in the context of LTE deployment since 2013. This was relevant to the relative value of patents that are essential to the earlier releases relative to those that are essential to later releases.

All six of Mr Saru, Mr Robbins and Mr Ware for UnwiredPlanet and Mr Cheng, Mr Zhang and Mr Yang for Huawei were good witnesses. UnwiredPlanet suggested that Mr Zhang did his best to defend Huawei’s conduct but at times was somewhat argumentative when answering the questions. That is not a useful observation because in my judgment the equivalent point could be said of all fact witnesses, including Mr Saru, Mr Robbins and Mr Ware. None of the fact witnesses called by either party was neutral but in my judgment all of them tried to explain their position as fairly as they could. If it is necessary I will deal with any detailed point or criticism about something a witness said in context.

At this stage I must pay tribute to the interpreter, Ms Langtree. Her work was of great assistance both to the witnesses and the court. I am very grateful to Ms Langtree.

There were further witness statements in the trial bundles from a number of witnesses. The statements were from Jay Shim and Tae Hyung Kim for Samsung; Christina Petersson, Andreas Iwerback, Jon Lawrence and John Han for Ericsson. UnwiredPlanet also had statements from Annabel Thomas, Timothy Ellis, Andrew Wynn, and Andrew Sharples.

The expert witnesses called by UnwiredPlanet were Dr David Cooper, Mr Mark Bezant and Dr Gunnar Niels.

Dr Cooper is a Consulting Engineer at Hillebrand Consulting Engineers GmbH. In 1978, he obtained a degree in Mathematics from Imperial College, London. He was subsequently employed by several companies in software development roles, and in 1987, he joined Coherent Research (a communication and engineering consultancy) as a software manager.

From 1994 to 1998, Dr Cooper was employed by NEC, during which time he represented the company within ETSI in the development of the GSM and UMTS Standards. In 1998 he was appointed the Vice-Chairman of the 3GPP standards committee SA1, which was responsible for the definition of UMTS services. From 1999 to 2008, Dr Cooper worked for Panasonic, representing them in the context of standardisation at 3GPP. In 2001 he earned his PhD in electronics and engineering at Surrey University. Since 2008, Dr Cooper has been a consultant at Hillebrand Consulting Engineers GmbH and has been retained as an expert witness on multiple occasions. He appeared for UnwiredPlanet before me in Trial A.

Dr Cooper’s evidence related to certain UnwiredPlanet patents and whether they were truly essential. He also addressed the parties’ rival approaches for identifying and counting truly essential patents.

Mr Bezant is the head of Economic and Financial Consulting Practice for Europe and a Senior Managing Director of FTI Consulting Inc. He is a fellow of the Institute of Chartered Accountants in England and Wales and the founding and current chair of the Valuation Special Interest Group of the Institute of Chartered Accountants in England and Wales. He has acted as an expert witness on numerous occasions, including over 60 IP cases and licensing disputes. His evidence related to FRAND rates and the relevance of various comparable licences.

Dr Niels is a partner at Oxera Consulting LLP. He holds a PhD in Economics from Erasmus University, Rotterdam and is a professional economist of 20 years' experience in the field of competition policy, covering mergers, restrictive agreements, abuse of dominance and damages. From 1995 to 1999, when he joined Oxera, he worked at Mexico's Federal Competition Commission. He is currently a non-governmental adviser to the International Competition Network Working Group on Unilateral Conduct. He has acted as an expert witness in competition and commercial disputes in a number of jurisdictions. Dr Niels’ evidence related to the impact of the FRAND obligation and the potential for hold-out on the behaviour of licensors and licensees.

All three experts called by UnwiredPlanet were good witnesses, trying to help the court. Huawei did not criticise these witnesses but they did point out a very odd aspect of Mr Bezant’s evidence. In cross-examination Mr Bezant repeatedly stated that he was not attempting to value the UnwiredPlanet portfolio but only seeking to express a view on whether UnwiredPlanet’s various offers were or were not FRAND. One aspect of this made sense in that Mr Bezant was presenting a range of licence rates with a view to explaining simply that a given rate fell within the range. Huawei sought to suggest that this had something to do with the procedural argument about global rate setting and suggested that in that sense Mr Bezant was the same as Mr Lasinski (Huawei’s expert) in that he too was only concerned with what Huawei called “the binary question of whether UnwiredPlanet’s offers were or were not FRAND?”. However, I do not accept these suggestions. The point is just as odd in the context of setting a UK only rate as setting a global rate and Huawei positively asked the court to set a UK only rate. In the end whatever Mr Bezant’s purpose, or for that matter Mr Lasinski’s, they both gave evidence which is relevant to the question of the value of the UnwiredPlanet portfolio. I suspect that what was really going on was that Mr Bezant was purporting to avoid answering the question of what in his opinion would be the correct FRAND rate for UnwiredPlanet.

The expert witnesses called by Huawei were Dr Apostolos "Paul" Kakaes, Mr Michael Lasinski and Professor Damien Neven.

Dr Kakaes is consultant at Cosmos Communications Consulting Corporation. He holds a B.S. and M.S. in Applied Mathematics and Electrical Engineering from the University of Colorado. In 1988, he was awarded a PhD in Electrical Engineering from the Polytechnic Institute of New York. From 1987 to 1994, he worked in the Department of Electrical Engineering at George Washington University, Washington DC, teaching graduate courses in the area of communication engineering.

In 1995, Dr Kakaes left George Washington University and went to work full time for Cosmos Communications Consulting Corporation, a private communications engineering consulting firm specialising in mobile communications, which he formed some years earlier. He has acted as an expert witness in a number of trials and arbitrations since 2006. His evidence related to Huawei's patent analysis (HPA) and UnwiredPlanet's MNPA.

Mr Lasinski is a Managing Director and the Chief Executive Officer of 284 Partners LLC, a professional services firm focussed on IP valuation, litigation consulting, IP strategy and IP transactional services. He holds a BSc in Electrical Engineering and an MBA from the University of Michigan, is a Certified Public accountant licensed in the state of Illinois, is certified in Financial Forensics by the American Institute of Certified Public Accountants and is a Certified Licensing Professional originated by the Licensing Executives Society (United States and Canada) (LES), one of North America's largest IP licensing trade organisations. His consulting practice has focussed on the financial aspects of IP since 1995. He is a past president of LES. He has acted as an expert witness in tax proceedings in the US and in international arbitrations, and provided IP valuations for taxpayers and the IRS. His evidence related to whether certain of UnwiredPlanet's licence offers were fair and reasonable, in particular by reference to various comparable licences.

Professor Neven is Professor of Economics at The Graduate Institute, Geneva and a senior Consultant at Compass Lexecon, having previously been a senior Academic Consultant to Charles River Associates. He holds a doctorate in economics from Nuffield College, Oxford, and has taught at INSEAD, the College of Europe and the University of Lausanne. From 2006 to 2011, he was Chief Competition Economist at the European Commission and has previously acted as an expert witness in the General Court and the Court of Justice. His evidence related to the relevance of the FRAND obligation in assessing market dominance, and the potential for hold-out on the behaviour of licensors and licensees.

UnwiredPlanet rightly made no criticism of Prof Neven, who was a good witness. They suggested his evidence was rather theoretical. There is something in that but it applies to Dr Niels too, albeit with less force. UnwiredPlanet did criticise Dr Kakaes and Mr Lasinski. The criticisms of Dr Kakaes and some of the criticism of Mr Lasinski relate to the Ericsson-Huawei arbitration and are best addressed in that context.

UnwiredPlanet also submitted that Mr Lasinski seriously lacked objectivity, stepped outside the ambit of his expertise and was prepared to speculate and did not approach the joint experts’ statement appropriately.

The challenge to objectivity relied on four alleged inconsistencies, which were that: Mr Lasinski was inconsistent in placing reliance on the words of a licence when it suited him but then relying on context in other cases; he was inconsistent in rejecting all pre2013 licences because they all involved hold-up but then relying on the 2009 EricssonHuawei licence when it suited him; his approach to the two UnwiredPlanet licences with Lenovo and Samsung was inconsistent, relying on Samsung when it suited him but not on Lenovo when it did not; and he was inconsistent in his approach to using a figure for UnwiredPlanet’s share of Ericsson’s patent portfolio, having been happy to use a share of 10% at the beginning but then changing to a lower figure once UnwiredPlanet’s claimed rate dropped from 0.2% to 0.13% because the 10% share would have supported UnwiredPlanet.

Five points were relied on to illustrate the submission that Mr Lasinski exceeded the ambit of his expertise and speculated. They are: that he passed judgment on whether terms in the Lenovo agreement were purely cosmetic; that he ignored Mr Ware’s evidence about the UnwiredPlanet-Samsung licence; that he decided for himself to accept or reject certain evidence about […] in the Ericsson-Samsung 2014 licence; that he was not an expert on the technical issues underpinning the two rival portfolio comparison methodologies but decided for himself to use the HPA but not the MNPA; and that he was prepared to express the view that there had been hold-up leading to a licence but never addressed hold-out.

The point on the joint experts’ statement was that he approached the exercise as one of identifying where his evidence was on a topic rather than by engaging with the issues raised by the other experts.

In terms of the specific points taken by UnwiredPlanet, the first three points on objectivity and the first three points on ambit are best addressed in context. Having heard Mr Lasinski, the suggestion, if made, that I should reject his evidence wholesale would be unwarranted. Mr Lasinski gave his oral evidence fairly, however many of these criticisms arise from Mr Lasinski’s approach to his written evidence in this case.

The parties exchanged evidence of French law from Prof Fauvarque-Cosson for UnwiredPlanet and Prof Libchaber for Huawei. Prof Fauvarque-Cosson has been a Professor of Law at the Université Panthéon-Assas (Paris II) since 2002, while Prof Libchaber has been a Professor of Law at the Paris I University (Panthéon-Sorbonne) since 2001. Neither witness was cross-examined and neither side challenged the qualifications of these witnesses to cover the matters they addressed in their reports.

During the trial I decided that it would be useful if the evidence of Prof Neven and Dr Niels started with a short period of concurrent evidence (CPR 35PD section 11). This was in order to address certain general questions which I would otherwise have asked the experts during their oral evidence. They were not points addressed to either witness in particular and it seemed to me that the fairest way of dealing with it was to ask them both, hence a concurrent evidence session. Prof Neven was scheduled to be crossexamined first, followed by Dr Niels so the concurrent session took place just before

What follows is a summary of the factual background. Although some detailed points are addressed later, it is convenient to set out the overall history from start to finish here.

The business which became UnwiredPlanet was founded in 1994 as Libris Inc. Its purpose was to develop technology concerned with how mobile devices (phones) could access the internet. In 1996 Libris changed its name to UnwiredPlanet Inc. and launched its first commercial product, called up.link, which was a mobile network system. In 1998 UnwiredPlanet was a member of the WAP Forum. At the time WAP was an early approach to mobile internet access. The WAP forum included Ericsson, Motorola and Nokia. Between 1999 and 2001 what had been UnwiredPlanet was now called Openwave Systems Inc. In 2002 Openwave was one of 200 companies to found the Open Mobile Alliance. This group included the WAP Forum companies. The general aim of both groups was similar, to try and come up with global standardised protocols for mobile internet and to lobby standard setting organisations. Openwave continued to develop mobile internet technology but by the time Mr Robbins joined in 2011 the company’s success was uncertain. That was because of a shift in the balance of power away from Openwave’s customers (carriers) and towards device makers like

The history of Ericsson’s business up to 2011 does not matter. By 2011 Ericsson was a major technology developer in telecommunications and a participant in standard setting. It had a handset business via a joint venture with Sony and an infrastructure business. Ericsson also had a major licensing business earning revenue from its patents. Ericsson had licensed a wide range of companies including Samsung. In 2009 Ericsson and Huawei had signed a telecommunications patent licence which was still in force in 2011.

In October 2011, Ericsson announced that they were going to leave the handset business, selling their shares in the joint venture Sony-Ericsson to Sony. The transaction completed in February 2012. Since then Ericsson has remained active in infrastructure. Its largest competitor is Huawei. At the same time, Ericsson started thinking about selling some patents. Ericsson’s motives for this come up later. Part of this exercise involved identifying organisations which might take some patents from Ericsson with a view to licensing them to the industry. The patents would include SEPs. At some point before June 2012, Ericsson identified UnwiredPlanet as a possible candidate.

By early June 2012 the discussions between Ericsson and UnwiredPlanet were underway with the name […]. In July and August 2012, UnwiredPlanet and Ericsson engaged in extensive discussion concerning the composition of the “[…] portfolio” of patents to be transferred. Ericsson were in control of the process. The process included swaps, in that UnwiredPlanet could ask for a patent which was earmarked to be transferred to be swapped out and replaced by another. During the negotiations possible royalty rates were discussed. The evidence is clear that UnwiredPlanet did not see Ericsson’s actual royalty rates, because the licences were confidential. UnwiredPlanet made its own assessment based on whatever material was available, including public statements by Ericsson and others.

On 10th January 2013 the MSA was executed. One of the parties is an entity called Cluster LLC but the detailed corporate arrangements do not matter. Pursuant to the agreement 2,185 patents and applications were transferred to UnwiredPlanet from Ericsson via Cluster LLC. In numerical terms this represented about [Z%] of Ericsson’s relevant portfolio. […]

The MSA included, in clause 3.2(a), three “tiers” for the revenue split between UnwiredPlanet and Ericsson, being 20:80 in UnwiredPlanet’s favour between $0m and $100m, 50:50 from $100m to $500m and 70:30 in Ericsson’s favour above $500m. The MSA also contained royalty floor provisions in clauses […]. The clauses are quite complicated. They include a floor rate referred to as the ARR. The ARR is […] for 3G and […] for 4G. […] That is why the ARR operates as a royalty floor.

After the MSA had been executed, UnwiredPlanet began to formulate a strategy for approaching potential licensees under its new portfolio of Ericsson-derived patents and in particular SEPs. UnwiredPlanet’s initial aim was to contact and commence negotiations with various manufacturers they had identified with a view to closing three deals by the end of the year. The initial list of manufacturers did not include Huawei but by April 2013 UnwiredPlanet had identified Huawei as a company to be contacted by the end of June. UnwiredPlanet decided to offer a flat rate of $1 per LTE multimode handset. Taking a handset sale price as $200, that would be 0.5%. At that time Ericsson’s publicly stated expectation for a rate for their 4G/LTE patents was 1.5%.

Before entering into the MSA and thereby transferring patents to UnwiredPlanet, […]. The transfer process was a bit more complicated than this but for this purpose that does not matter. Ericsson ranked its patents in tiers and selections of patents were chosen in each tier. […] So purely on a numerical basis based on total numbers and declared numbers there is a ratio of about [Z%] between UnwiredPlanet’s and Ericsson’s patent portfolios.

UnwiredPlanet’s contacts with Samsung started in October 2012. Under the cover of a Non-Disclosure Agreement (NDA) claim charts were provided on 17th December 2012 along with some other information about litigation brought by UnwiredPlanet against Apple and Google in Nevada USA. The first meeting took place in May 2013 and a second meeting took place between UnwiredPlanet and Samsung in August 2013. UnwiredPlanet’s position is that since the start in October 2012, Samsung were holding out.

By August 2013, according to Mr Robbins, UnwiredPlanet had contacted 27 manufacturers and was in “active conversations” with 14, whilst the remaining 13 had

“refused to engage with us at all”. In oral evidence, Mr Robbins stated that there was not “much market interest in even discussing rates” and that he had hardly managed to “speak about economics with anybody”, it being “largely technology discussions”. Before me UnwiredPlanet contend that this shows that the licensees in general were holding out. Huawei contend that in fact the rates UnwiredPlanet were demanding were just too high. There is evidence that […] were prepared to at least contemplate deals at […] per LTE unit rather than the $1 being proposed by UnwiredPlanet. There was a debate about how much the ARR played a part here. Mr Robbins downplayed its significance but in my judgment from the beginning the ARR was an important factor driving UnwiredPlanet’s strategy.

In June 2013, UnwiredPlanet decided to approach Huawei. The approach was about a possible purchase by Huawei of Ericsson-derived infrastructure patents. UnwiredPlanet discussed this with its advisors Evercore and the approach occurred on 2 July 2013. Correspondence ensued and by 22nd August 2013 Huawei had informed Evercore that it was not interested in acquiring Ericsson patents.

A point which UnwiredPlanet emphasise in these proceedings is that the EricssonHuawei 2009 licence had expired at the end of 2012 and that, as a result of the MSA, by 2013 certain Ericsson SEPs were now held by UnwiredPlanet. The significance of this point is that while I accept Huawei’s case that this first approach from UnwiredPlanet to Huawei via Evercore concerned a purchase and not licensing (contrary to suggestions from UnwiredPlanet), nevertheless, as UnwiredPlanet submit, by 2013 Huawei ought to have known that they would need a licence from UnwiredPlanet to continue to use SEPs they had formerly licensed from Ericsson. There is no evidence Huawei considered this point at the time at all and I doubt they did. In crossexamination Mr Zhang made the point that since UnwiredPlanet were trying to sell the patents it had acquired from Ericsson, then from Huawei’s point of view it was not clear the patents would remain with UnwiredPlanet. They might be sold on elsewhere. That is true but it does not take away the force in UnwiredPlanet’s point that after early 2013 Huawei knew all they needed to know to appreciate that certain SEPs which they had formerly licensed were now held by a different company and, if and to the extent a licence was required, it would have to come from UnwiredPlanet or its successors.

On 13th September 2013, Mr Saru wrote letters to Mr Guo Ping and Mr Ren Zhenfei, who are two Board members of Huawei, the latter being the founder and CEO, suggesting that the two companies should “sit down and have an extended discussion” at some point in October 2013 with a view ultimately to concluding a licence. No reply was received to these letters. Huawei justified this on the basis that the letters were not copied to Huawei’s IP or Licensing departments. Mr Zhang explained that they were not brought to the attention of those departments. No chasing letters were sent by UnwiredPlanet until 25th November 2013 (see below).

By October 2013 further meetings with Samsung had taken place. Mr Robbins said he feared that Samsung had no intention of ever engaging in sensible commercial negotiations. By now UnwiredPlanet were concluding that the underlying thesis, that the strength and depth of the portfolio would be so compelling that licensing deals could be reach without litigation in many cases, was wrong. On 23rd October 2013, Unwired

On 25th November 2013 UnwiredPlanet contacted Huawei’s IP department. Huawei responded very promptly. There was a brief delay during December 2013. On 13th January 2014 Huawei asked UnwiredPlanet for claim charts. On 16th January 2014, UnwiredPlanet agreed to produce charts under an NDA and included draft terms. On 29 January 2014 Huawei proposed different NDA terms. On the same day, UnwiredPlanet replied stating “thank you for your message. We will consider this”. Mr Saru passed Huawei’s draft NDA to “our contract lawyer” for consideration. There was no further contact until 10^th March when the litigation started.

UnwiredPlanet’s contacts with Lenovo had started in May 2013 with a licensing proposal. In August 2013 Evercore had discussions with Lenovo about Lenovo acquiring part of UnwiredPlanet’s patent portfolio. Negotiations continued from then on. The contract was finally agreed in March 2014. The details of the contract are discussed below as a comparable. Lenovo paid $100m to UnwiredPlanet and so under the MSA from that time only the second tier arrangements apply. Under the contract Lenovo acquired 21 families from UnwiredPlanet, […]. UnwiredPlanet also acquired some further patents from Ericsson to add to its portfolio. After the Lenovo transaction was complete UnwiredPlanet claimed to own 30 SEP families.

In order to facilitate litigation arrangements had to be made which required Ericsson’s consent. This was embodied in an agreement dated 27th February 2014. On 5th March 2014, the Board of UnwiredPlanet approved the European litigation.

On 10th March 2014 Huawei received an email from Mr Saru stating that UnwiredPlanet had decided “to proceed with enforcement in Europe” by suing Huawei for patent infringement in the UK and Germany. Mr Saru says in his witness evidence that his email was to “confirm” that they were suing Huawei in Germany and the UK, but this was the first Huawei had heard of being sued. The previous contact between UnwiredPlanet and Huawei had been about the terms the NDA, whereby Huawei was waiting to receive comments from UnwiredPlanet.

On the same day, UnwiredPlanet issued patent infringement proceedings in the UK and Germany against Huawei, Samsung and Google, and against HTC in Germany. There is an issue about the nature of the relief claimed by UnwiredPlanet which I will address in the competition law section below.

The April 2014 offer followed. In June 2014, UnwiredPlanet completed an NDA with Huawei. Discussions continued thereafter. A debate between UnwiredPlanet and Huawei arose about without prejudice privilege. On 30 July 2014, UnwiredPlanet made a without prejudice licensing proposal which was later repeated in open correspondence and is referred to as the July 2014 offer. Without prejudice negotiations continued after that.

In September 2014, UnwiredPlanet commenced working on a method of counting and classifying SEPs to use in licensing negotiations. It is called the “modified numeric proportionality approach” (MNPA). It is addressed at length below.

The litigation continued with various case management and strike out hearings in 2014/2015. In June 2015, following directions from the court, the parties made open offers. In February 2016 UnwiredPlanet and Huawei exchanged open correspondence concerning their lack of progress in concluding a FRAND licence. Each accused the other of the blame for the lack of progress.

Unknown to Huawei, UnwiredPlanet and PanOptis had been negotiating and on 6 April 2016 the acquisition by PanOptis was announced. On 10 August 2016, UnwiredPlanet’s solicitors disclosed the 2016 UnwiredPlanet-Samsung Licence. This licence is addressed in detail below. […]

What is FRAND and what principles apply to it?

The point of FRAND in standard setting is fairly easy to understand. Standards exist so that different manufacturers can produce equipment which is interoperable with the result that the manufacturers compete with one another. So the phone makers compete in the market for phones and the public can select a phone from any supplier and be sure (for example) that if it is a 4G phone, it will work with any 4G network. As a society we want the best, most up to date technology to be incorporated into the latest standards and that will involve incorporating patented inventions. While the inventor must be entitled to a fair return for the use of their invention, in order for the standard to permit interoperability the inventor must not be able to prevent others from using the patented invention incorporated in the standard as long as implementers take an appropriate licence and pay a fair royalty. In this way a balance is struck, in the public interest, between the inventor and the implementers. The appropriate licence is one which is fair, reasonable and non-discriminatory. That way a standard can safely incorporate the invention claimed in a patent without giving the inventor or his successors in title unwarranted power over those who implement the standard. Thus the public interest is served because telecommunication standards can be set using the best and most up-to-date technical expedients available and the inventor’s private interest is served because the FRAND undertaking ensures they or their successors will obtain a fair reward for their invention.

Telecommunications standards worldwide are formulated and set by SSOs (Standards Setting Organisations). In Europe the relevant SSO is ETSI. SSOs require the holders of patents which are essential to the standards to give an undertaking to license on FRAND terms if they wish to participate in standard setting.

In ETSI this process is based on its Directives which include Rules of Procedure and a Guide to IPRs. Within the Directives, as an annex to the Guide to IPRs, is an IPR Policy. The policy and the rules have been adjusted over time but none of the issues before me turn on any differences. Article 4.1 of the ETSI IPR Policy requires members of ETSI to inform ETSI of "ESSENTIAL IPRs" in a timely fashion. ESSENTIAL and IPR are defined terms (article 15). A patent which would inevitably be infringed by operating in accordance with a standard is an example of an ESSENTIAL IPR. By definition a SEP is an ESSENTIAL IPR. Once an ESSENTIAL IPR has been declared by its owner to ETSI, the owner will be requested by ETSI (Article 6.1 of the ETSI IPR Policy) to give an irrevocable undertaking in writing that it is prepared to grant irrevocable licences on FRAND terms. In its form applicable to the 2014 offers the relevant policy is dated 20^th March 2013. Article 6.1 in that form is:

“6.1

When an ESSENTIAL IPR relating to a particular STANDARD or TECHNICAL SPECIFICATION is brought to the attention of ETSI, the Director-General of ETSI shall immediately request the owner to give within three months an irrevocable undertaking in writing that it is prepared to grant irrevocable licences on fair, reasonable and non-discriminatory (“FRAND”) terms and conditions under such IPR to at least the following extent:

●

MANUFACTURE, including the right to make or have made customized components and sub-systems to the licensee's own design for use in MANUFACTURE;

●

sell, lease, or otherwise dispose of EQUIPMENT so

MANUFACTURED;

●

repair, use, or operate EQUIPMENT; and

●

use METHODS.

The above undertaking may be made subject to the condition that those who seek licences agree to reciprocate.”

Whether a declarant is an ETSI member or not does not matter in that undertaking. Under the rules anyone declaring a patent to ETSI as essential IPR is asked to give a FRAND undertaking. A question is whether the commitment to ETSI is enforceable by a third party putative licensee such as Huawei in these proceedings. This is addressed in the next section.

Other relevant terms in the ETSI IPR Policy are:

A notable feature which is not in the ETSI IPR Policy is any obligation on ETSI to check whether declared patents are in fact essential. The only mechanisms which exist to decide these questions are court proceedings or arbitration.

The history and purpose of FRAND

ETSI was established in 1988 and the FRAND undertaking was developed by it with the close involvement of the European Commission, which reflects the importance of FRAND from the point of view of competition policy. The first public formulation by the Commission of a specific requirement for FRAND terms in the context of IP and standardisation was in a paper “Communication on Intellectual Property Rights and Standardisation” on 27^th October 1992 (COM (92) 445 final at 4.3.3). Nevertheless the concept has its origin in US anti-trust law (see “A Brief History of FRAND: Analyzing Current Debates in Standard Setting and Antitrust Through a Historical Lens” Jorge Contreras 80 Antitrust Law Journal 39 (2015)). In the US it is sometimes called “RAND” rather than FRAND but there is no material difference between the two. The idea of FRAND licences for patents essential to standards is not unique to ETSI. It is something other international SSOs require as well (e.g. the IEEE and ITU referred to by US District Judge James Robart in Microsoft v Motorola Case C10-1823JLR, 2013 US Dist LEXIS 60233 (W.D. Wash. April 25 2013 at page 3)).

Moreover although ETSI is a European organisation, it is truly international in scope. The ETSI FRAND undertaking has been considered in courts all over the world including the German courts (see further below), US International Trade Commission (see decisions between InterDigital v Nokia and ZTE dated 13^th June 2014 Inv No. 337-TA-868 and 27^th April 2015 Inv. No. 337-TA-613), the IP High Court of Japan (see the Apple v Samsung Case No 2013 [Ne] 10043 dated 16^th May 2014) and the

This international scope arises because the standards supported by the ETSI FRAND undertaking are and are intended to be interoperable globally both from the point of view of implementers and also users (see e.g. Art 6.2 and Mr Cheng’s evidence above). A benefit of standardisation is that implementers will be able to make and sell products compliant with the standards all over the world. Another benefit is that users will be able to use the same equipment globally.

The underlying purpose of the FRAND undertaking is to secure a proper reward for innovation whilst avoiding “hold up”, i.e. the ability of the owner of a SEP to hold implementers to ransom by reason of the incorporation of the invention into the standard by declining to grant them a licence at all or only granting one on unfair, unreasonable or discriminatory terms. The idea is to strike a fair balance. This way of describing the purpose of FRAND is not in dispute and can be seen in numerous sources. It was put in the following way in the Huawei v InterDigital case in China:

“For good faith users who are willing to pay reasonable royalties, holders of standards-essential patents should not directly refuse to grant licenses. On the one hand, it is necessary to ensure that patentees can obtain sufficient returns from their technical innovations. On the other hand, holders of standards-essential patents should be prevented from charging exorbitant royalty rates or attaching unreasonable terms by leveraging their powerful position forged by the standards. The core of the FRAND obligations lies in the determination of reasonable and non-discriminatory royalties or royalty rates.”

[section IV 2^nd paragraph (p56 of the translation)]

I agree with the Guangdong High People’s Court’s succinct summary of the purpose of FRAND.

The same principles have been recognised in other courts internationally, see:

However eliminating hold up value is not the only consideration to take into account. The authorities and the economics literature have identified a countervailing factor called “reverse hold up” or “hold out”. The idea is that an unscrupulous licensee may use their economic strength to avoid paying anything to a patentee, unduly dragging out the process of licence negotiation, thereby putting the patentee to additional cost and forcing it to accept a lower royalty rate than is fair. The possibility of delaying tactics from a licensee is recognised in Huawei v ZTE (CJEU) at paragraph 71 and also paragraphs 37-38 (referring to the referring court).

In a paper of which Prof Neven was an author (“Injunctions for Standard-Essential

When talking about FRAND economists refer to the idea that the FRAND rate represents the rate which would be agreed “ex ante”, in other words before the patented invention is adopted into the standard. This is another way of saying that the rate seeks to eliminate hold up and to that extent is uncontroversial. In the concurrent evidence session Prof Neven explained that he did not regard FRAND as a scheme which meant the patentee could not appropriate some of the value that is associated with the inclusion of his technology into the standard and the value of the products that are using those standards. Dr Niels agreed with that. Neither side disputed this and to the extent it is a matter for the economists, I accept their evidence. The economists’ opinions show that it is not necessary to deprive the patentee of its fair share of those two sources of value in order to eliminate hold up and fulfil the purpose of FRAND. To that extent I may be differing from certain parts of the decisions in Innovatio IP Ventures and Ericsson v D-Link in the US but it is not necessary to look into that any further since neither side before me took the point.

Enforceability of the ETSI FRAND undertaking and French law

Although it is common ground in this case that UnwiredPlanet is bound in law to license on FRAND terms, the questions of its enforceability and the legal basis for it are not purely academic for two reasons, first because a correct identification of the legal basis may inform decisions about the undertaking’s scope and effect, and second because the degree of uncertainty around its enforceability plays a part in the argument about abuse of dominance.

My judgment on the strike-out in April 2015 ([2015] EWHC 1029 (Pat) para 29) noted that FRAND could be considered in three relevant legal contexts: (1) compliance with the FRAND commitment as a matter of contract, (2) compliance with competition law and (3) the grant or refusal of injunctions (“equitable refusability”). At this stage I am concerned with the first context.

Article 12 of the ETSI IPR Policy provides that it is governed by French law and the IPR declaration forms also refer to French law. They provide that the construction, validity and performance of the undertaking is governed by French law (see the extract from the forms set out later in this judgment). The parties exchanged reports of French law experts.

UnwiredPlanet referred to the judgment of the Court of Appeal from the strike-out judgment ([2016] EWCA Civ 489) in which it was held that:

“38 […] UP LLC and UP were required to give FRAND undertakings and they each did so shortly after the SEPs in issue were transferred to them. It is true that UP is not a member of ETSI but it is just as constrained by the FRAND undertaking it has given as it would be if it were such a member and the judge was right to hold that, as a practical matter, any third party may require UP to grant it a licence under the SEPs on FRAND terms.”

No doubt this clear decision is the reason why neither side actually disputes that as a practical matter the FRAND undertaking is binding on UnwiredPlanet and enforceable by Huawei. However the opinion of Huawei’s French law expert, Prof Libchaber, is that in French law those conclusions are not correct and, as is said in his second report (paragraph 10 footnote 7) the Court of Appeal did not explore the legal basis for its conclusion and French law matters were not in issue on the appeal. Given that the French law matters were argued before me (in writing) and for the reasons already explained, it is necessary to address the issue.

It is not in dispute that ETSI as an association was formed under the Loi du 1er juillet 1901 relative au contrat d'association (the French law of 1 July 1901 and the decree of 16 August 1901), Article 1 of which states: (in translation): “An association is a contract by which two or more persons bring together, in a permanent manner, their knowledge or their activities for a non-profit purpose. It is ruled, as to its validity, by the general principles of law applicable to contracts and obligations”.

Prof Libchaber explained that as an association which results from a contract, the legal principles governing contracts are applicable and he pointed out that Article 1165 of the old French Civil Code provides that agreements only produce effects between the contracting parties. It is only as an exception that they can benefit a third party and that is in the case provided for in Article 1121. I will address Article 1121 and third party rights below.

Prof Libchaber noted that paragraph 1.4 of the ETSI Guide states that the IPR Policy defines the rights and obligations of ETSI as an Institute for its members and states that while non-members have certain rights under the policy they (non-members) do not have obligations. Accordingly his opinion is that non-members of ETSI such as UnwiredPlanet are not subject to any legal obligations arising from French contract law under the policy. At this stage the analysis is not concerned with the effect of making a declaration to ETSI of SEPs and the French law expert relied on by UnwiredPlanet, Prof Fauvarque-Cosson, did not disagree with Prof Libchaber so far.

Prof Fauvarque-Cosson’s view is that making a declaration under Art 6.1 of the ETSI IPR Policy gives rise to a binding contract between ETSI and the declarant and this is so regardless of whether the declarant is a member of ETSI or not. Prof Libchaber does not agree. The dispute has a number of dimensions which are best taken separately although they interact. The first is whether making an ETSI declaration forms a contract at all. The second dimension relates to Art 1121 of the old French Civil Code and the French law doctrine of stipulation pour autrui, in other words the law whereby third parties may acquire enforceable legal rights against one party to a contract. The third dimension relates to the nature of any obligation which does arise, Prof Libchaber expressing the view that it is not the usual practice for French courts to compel a party to enter into a contract. Finally the fourth dimension concerns the efficacy of a unilateral commitment under Article 1000-1 of the old French Civil Code.

The French Civil Code referred to in the previous paragraphs is called the “old” French Civil Code because on 1^st October 2016 a new Civil Code relating to contract law came into force in France. The new Civil Code is not retrospective and so it is not applicable to UnwiredPlanet. Prof Fauvarque-Cosson referred to both and expressed the view that the changes did not make any difference to the result for any of the matters in debate although in some cases the law was clearer under the new Civil Code. Prof Libchaber explained that the new Civil Code did not apply to UnwiredPlanet but he did not identify any specific aspect in which the difference mattered. In this section I will refer to the two codes as the old Civil Code and the new Civil Code.

Does making ETSI declaration form a contract at all

In the relevant declaration a patentee declares to ETSI which patents it contends are essential to the standards. In Prof Fauvarque-Cosson’s opinion it is the making of this declaration to ETSI by a non-member which leads to that non-member having a legally enforceable obligation to license on FRAND terms. Since UnwiredPlanet is not a member of ETSI, this is crucial.

Prof Fauvarque-Cosson explained that Article 1108 of the old Civil Code provided that the conditions necessary for the formation of a contract were:

“1.

the consent of the party assuming the obligation

The Professor explained that under the new Civil Code the concept of a “cause” is no longer stated as a condition governing the formation of contracts and is not relevant to the issues arising in this case. Under Article 1128 of the new Civil Code the corresponding conditions are: 1. the consent of the parties; 2. their capacity to contract; 3. content which is lawful and certain.

The Professor explained that under French law a contract is formed once the parties agree on its essential elements. This can be by means of a sufficiently precise offer and acceptance. Although the old Civil Code did not contain detailed provisions expressly dealing with formation, the new Civil Code does and the rules generally enshrine existing case law. Once an offer has been made it must be accepted and the acceptance does not require any particular formality. It can be express or implied unless the offer specifies that acceptance must be express. Prof Libchaber did not disagree with any of this.

Prof Fauvarque-Cosson analysed the position in the following way:

Point (v) will be addressed below. At this stage I am concerned with steps (i) to (iv) which in Prof Fauvarque-Cosson’s opinion create a binding contract between the IPR holder and ETSI.

The question has three aspects: (a) whether making the declaration imposes any sufficient obligations on ETSI to form a contract, (b) whether the declaration forms lack sufficient clarity to impose legal obligations on the declarant, and (c) whether in truth the declaration is really just a way of giving information to ETSI and the market.

whether making the declaration imposes any sufficient obligations on ETSI to form a contract

Prof Libchaber explained that a contract must create legal consequences for both parties. He referred to the new Civil Code in Article 1101 which provides that a contract is (in translation) “a meeting of wills between two or more parties designed to create, modify, transmit or extinguish obligations” (i.e “Le contrat est un accord de volontés entre deux ou plusieurs personnes destiné à créer, modifier, transmettre ou éteindre des obligations”). Prof Fauvarque-Cosson did not disagree with this nor was there any suggestion that the principle was any less applicable to cases governed by the old Civil Code. The idea underlying this is not difficult to grasp and is akin to the English law concept that for a promisor’s promise to be enforceable as a contract, consideration must move from the promisee.

Prof Libchaber’s view was that no contract was formed by the steps (i) to (iv) because even if a declaration form which is completed and sent by a patentee to ETSI creates obligations for the patentee (perhaps because they are an existing member of ETSI), it does not have any corresponding obligatory consequences for ETSI.

It is true that the ETSI IPR Policy itself does not purport to impose anything by way of an obligation on ETSI in return for receiving an essentiality declaration form from a patentee but Prof Fauvarque-Cosson explained that there is a step which ETSI will take after receiving a declaration, that is to include or maintain the declared essential patent on the ETSI database of essential IPR. Prof Libchaber’s view is that that is not sufficient to support the theory that making a declaration triggers the formation of a contract under French law because the entry of the patent in the database is not the counterpart to a declaration but rather a tool designed to assist interested parties in finding out which patents have been declared essential and are said to be available to license. His view is that it is artificial to analyse this as if it were a contractual advantage, and he doubted that a declarant would be able to compel ETSI to enter the declaration into the IPR database if this procedural step were not taken. He also pointed out that while the registration procedure may be implicit from the Policy, it appears nowhere in the Licensing Forms and so while the registration of the patent in the ETSI database could be seen as an obligation on ETSI where the declaration is made by an ETSI member, there will be no such obligation where the declaration is made by a nonmember.

Prof Libchaber is correct that neither the IPR Policy itself nor the declaration form refers to the ETSI IPR database however the database is described in Article 3.1.2 of the ETSI Guide to IPRs. To that extent steps (i) to (iv) are too narrowly stated but the difference does not matter. These are the provisions in the ETSI Directives to which the IPR Policy is an annex. Article 3.1.2 explains that the database allows online access to declarations such as those in issue in this case. The third sentence of the Article states as follows:

“Unless otherwise specified, all IPRs contained [in the ETSI IPR Database] have been notified to ETSI, with an undertaking from the owner to grant licenses according to the terms and conditions of Clause 6.1 of Annex 6 of the ETSI Rules of Procedure (the

ETSI IPR Policy).”

[ETSI Guide on IPRs 19 September 2013; ETSI Directives Version 36, June 2016]

This is a clear statement of ETSI’s registration policy. It shows that Prof FauvarqueCosson is correct to say that after receiving a declaration ETSI will include or maintain the declared essential patent(s) on the ETSI IPR database. Moreover, as the statement explains, unless otherwise stated the owners of the IPRs on the database will have given a FRAND undertaking. Article 3.1.2 is an express and public statement of what ETSI will do when it receives declarations. In my judgment the fact it is not mentioned in the declaration form is irrelevant. On its face this article applies whether the declarant is a member of ETSI or not and it would make little sense otherwise. I am not persuaded by Prof Libchaber’s view that at best this might be applicable to members rather than non-members.

If a contract is formed from the four steps described by Prof Fauvarque-Cosson (adjusted to include reference to Article 3.1.2) then one can see that an obligation with far reaching legal consequences for the declarant (to grant patent licences to all comers on FRAND terms) is an obligation of much more weight and significance than the apparently minor counterpart obligation on ETSI (if it exists) to enter the declaration on the IPR Database. However English contract law would not weigh up the relative values of the promise and the consideration for it and neither Professor suggested French law would do that either. Prof Libchaber’s view is that a requirement to put the declaration on the database is not “sufficient” to support the contract but that was not because of its inconsequential nature, it is because of the Professor’s view that it is not the counterpart of the declaration but rather a tool to assist interested parties. I accept that placing the declarations on the public database is a tool to assist interested parties but I do not accept that this is not a “counterpart” to the declaration. If Prof Libchaber is just referring to the fact that the database is not mentioned in the declaration form then I have dealt with that point.

Considering the issue as a matter of substance, the placing of the declaration on the database is indeed the counterpart to the declaration because the purpose of the declaration from ETSI’s point of view is to provide information which will be useful to give notice to standard setters and implementers. That information involves two important facts: first the existence of a patent which is, or at least which the owner contends is, essential to the standards, and second the fact that the owner is prepared to grant licences on FRAND terms. A declaration to ETSI which is not made public on the database is much less useful. Furthermore publication on ETSI’s own database of the assertion of essentiality and of the undertaking to enter into FRAND licences has a value to a patentee seeking to license its rights and obtain fair return for the use of its invention.

The terms of Article 3.1.2 are quite clear and do not leave room for ETSI to choose not to place a regular declaration on the database. In the Article ETSI is offering to declarants to behave in a particular way if a proper declaration is provided. It seems to me therefore that a person who provided a properly completed declaration form to ETSI would be entitled to expect ETSI to do what it says it will do in Article 3.1.2. In that case ETSI is obliged by Article 3.1.2 to place the declared IPR on the database in a manner which shows that an undertaking has been given, i.e. not to “otherwise specify”. Given the statements about French law both in the form and in the ETSI Directives, declarants are entitled to expect these statements to represent legally binding statements

whether the declaration forms lack sufficient clarity to impose legal obligations on the declarant

Prof Libchaber’s view is that the licensing forms cannot be regarded as an offer because they lack the necessary clarity and so do not have the necessary quality of certainty to amount to an offer capable of acceptance under French law. This is because they allow the declarant various options to choose from. The options are described by Prof Libchaber as follows:

“In this regard I note that the declarant may choose to make the declaration: (i) in respect of either IPR relating to contributions made to the standard setting process by the declarant or any of its IPR; (ii) in relation to specific ETSI standards / specifications or all of them; and (iii) subject to a condition of reciprocity. It is also open to the declarant to inform ETSI that it does not wish to make its IPR available for licensing on FRAND terms at all.”

Prof Libchaber is correct that these options exist but in my judgment their existence does not amount to a lack of clarity which could lead to a contract based on them being unenforceable. Each option is presented and it will be clear which one the declarant has chosen. The fact that options exist does not introduce a lack of clarity.

Dealing with the individual options, the first is concerned with which rights the declaration relates to if a general declaration is made. Either choice has a clear effect – either the declaration will relate to IPRs contained in technical contributions made by the declarant or its affiliates or else it will relate to all its IPRs. There is nothing uncertain in either option but in any case a general declaration is not the only option available. A declarant can list the IPRs to which the declaration relates and in such a case there is no lack of clarity at all.

The second point is about the standards to which the IPR is declared to be essential. Declaring IPRs as essential to all ETSI standards is not unclear, it is simply wide. And while a wide declaration like that may not help an implementer decide if they need a licence in particular circumstances, the database does not warrant that the IPRs are in fact essential in any case. The implementer will always have to form their own view about that (or have the matter resolved in court). I suppose a declarant could define the standards in an ambiguous way but regardless of how realistic the possibility is, it does not undermine the clarity or enforceability of the process itself. Assuming the declarant identifies the relevant standards sufficiently then the declaration will be clear and the undertaking could be enforced.

Similarly the third point, on reciprocity, does not introduce a lack of clarity. A declarant may state that they will only offer FRAND licences on condition that those who seek licences reciprocate. Whether they do or do not, either option is clear.

Finally the declarant can refuse to give the FRAND undertaking altogether. Again this is not unclear. If they do refuse it means that ETSI will know that there are patents alleged to be essential for which no FRAND undertaking is available. This is vital information and the ETSI Directives have a number of provisions dealing with the possibility (Article 8 of the ETSI IPR Policy). Broadly in that case ETSI will try and ensure that whatever the relevant technology is, it is removed from the standards.

I am not persuaded by this aspect of Prof Libchaber’s evidence. I can quite see that French law, like English law, imposes a requirement of certainty before a contract can be enforceable. That makes sense. However none of the grounds identified give rise to any material lack of certainty about the obligations which would be imposed on the declarant.

whether in truth the declaration is really just a way of giving information to ETSI and the market

Although it was not strongly advanced as a separate legal ground for rejecting the enforceability of the FRAND undertaking in the declaration, Prof Libchaber made the point in paragraph 17 and 18 of his third report that really a declarant is simply “informing” ETSI, and indirectly interested parties, that it is prepared to grant licences in accordance with Clause 6.1 of the Policy and that this is not consistent with the view that a binding contractual relationship is formed. The Professor put the word “informing” in quotation marks because that language comes from the declaration form.

There is also a general declaration but the use of “informs” is the same. The language has to be seen in the context of the declaration form as a whole. The next part, in which the declarant makes or refuses to make an irrevocable FRAND licensing declaration, is important:

“IPR LICENSING DECLARATION

In accordance with Clause 6.1 of the ETSI IPR Policy the Declarant and/or its AFFILIATES hereby irrevocably declares the following (check one box only, and subordinate box, where applicable):

 To the extent that the IPR(s) disclosed in the attached IPR Information Statement Annex are or become, and remain ESSENTIAL in respect of the ETSI Work Item, STANDARD and/or TECHNICAL SPECIFICATION identified in the attached IPR Information Statement Annex, the Declarant and/or its AFFILIATES are prepared to grant irrevocable licences under this/these IPR(s) on terms and conditions which are in accordance with Clause 6.1 of the ETSI IPR Policy.

 This irrevocable undertaking is made subject to the condition that those who seek licences agree to reciprocate (check box if applicable).

 The Declarant and/or its AFFILIATES are not prepared to make the above IPR Licensing Declaration (reasons may be explained in writing in the attached IPR Licensing Declaration Annex).

The construction, validity and performance of this IPR information statement and licensing declaration shall be governed by the laws of France.

SIGNATURE

By signing this IPR Information Statement and Licensing Declaration form, you represent that you have the authority to bind the Declarant and/or its AFFILIATES to representations and commitments provided in this form.

[signature etc] ___________”

These passages are from the form to be used when specific patents are identified. The general form is not materially different. The language is clear and at face value reads as something which would be understood as being intended to have binding legal force (for example note the reference to the laws of France and the reference to having authority to bind the Declarant). The fact that in the earlier part of the form the declarant is “informing” ETSI about particular IPRs makes sense since ETSI needs to be informed about what IPRs are the subject of the declaration, not least so that they can be identified on the database. I reject Prof Libchaber’s idea that all this amounts to is a means of providing information to the market rather than a formal commitment capable of binding the declarant. On the contrary the text as a whole is plainly written in such a way that the person signing it understands they are making a formal legal commitment on behalf of the declarant.

Art 1121 of the old French Civil Code and stipulation pour autrui

Prof Fauvarque-Cosson explained that French law includes a doctrine of “stipulation pour autrui” (an agreement for the benefit of third parties). While the concept is derived from Art 1121 of the old Civil Code, she explained that the provisions of the new Civil Code, which codify existing law, state the principles more clearly, as follows (in translation):

The term translated as “stipulator” in this extract may also be translated as “promisee”. Prof Fauvarque-Cosson explained the provisions on revocation but nothing turns on them. Prof Libchaber did not disagree with Prof Fauvarque-Cosson that these provisions restate the previous law without substantive change.

Prof Fauvarque-Cosson’s view is that the FRAND undertaking given by a declarant has the effect of engaging this doctrine under French law. Her view is as follows:

Prof Libchaber does not agree with this. He explained that the doctrine was not commonplace in French law because the conditions were rarely satisfied and there were two classic situations in which it applies – insurance (e.g. where a person takes out an insurance policy on their life, in which someone else is the beneficiary) and shipping (e.g. where a sender contracts with a carrier for the benefit of a recipient of goods). Prof Libchaber’s particular objections were these. First he did not agree that a contract between the declarant and ETSI was formed at all. I agree that such a contract is a necessary element for the doctrine to apply but I have already found that such a contract is formed. Second he did not consider the doctrine applied because ETSI imposes no requirement as to what the terms of any contract between the declarant and ETSI were. In other words, declarants can choose to declare IPRs to ETSI but not give a FRAND undertaking or do so conditionally. Again I have addressed this already. Prof Libchaber is right that a declarant can choose not to give the FRAND undertaking or to attach conditions to it. However the fact that the declarant has options does not seem to me to undermine the existence of a legally enforceable contract at least in the relevant case, in which the declarant gives an unconditional FRAND undertaking. Prof Libchaber’s view was that the doctrine could only apply if ETSI was imposing a particular obligation on the declarant instead of permitting them to choose between several options, but I do not see why that follows. An insurer could provide an insured with different options such as levels of cover but that would not undermine the contract between the two once it was formed on the basis of whatever option the insured had chosen nor would it undermine the fact that the contract was for the benefit of a third party.

Prof Libchaber’s next objection was as follows:

“If Professor Fauvarque-Cosson’s analysis is correct, the same Licensing Form would be considered as: (i) the offer made by

I am not persuaded by this. I cannot see any reason why a blank form is incapable of being an offer made by ETSI nor any reason why a properly completed form cannot be an acceptance of that offer, indicating which of the pre-defined options the form shows ETSI is prepared to offer, the declarant has chosen to accept. As for the terms of future contracts, the form makes an unambiguous reference to Clause 6.1 of the ETSI IPR Policy. That policy expressly provides that a future licence will be available on FRAND terms. The issue of what FRAND terms are is addressed below. At this stage the key point is that (as I have found below) it is possible for a court to adjudicate whether a licence is or is not FRAND. Whether terms are FRAND is an objective matter for a given set of circumstances and therefore whether a FRAND undertaking has been complied with can be determined as a matter of law. In that sense the undertaking is legally enforceable. Knowing that the licence will be on FRAND terms is all the parties need to know.

I sympathise with Prof Libchaber’s observation that the doctrine is not commonplace in French law. However that is no reason not to apply it to the relatively new problem of how to facilitate the setting of standards in rapidly developing areas of technology and balance the public benefit of access to the latest and best technology for these standards with the appropriate incentives for inventors and investors in technological innovation. The FRAND undertaking sought by ETSI when a patentee declares its patents as essential to an ETSI standard is an undertaking given in terms to confer a benefit on third parties. I accept Prof Fauvarque-Cosson’s analysis and find that the doctrine of “stipulation pour autrui” applies to the FRAND undertaking and renders it enforceable by third parties.

The nature of any obligation which does arise

This point is related to the previous one but is not the same. Prof Libchaber expressed the view that it is not the usual practice for French courts to compel a party to enter into a contract and that therefore the limit of the terms of Clause 6.1 as interpreted by a French court would be to impose an obligation owed by the declarant to third party implementers to make a good faith attempt to negotiate a FRAND licence. So he explained this is another reason why “stipulation pour autrui” does not apply, because for it to apply the right must be capable of specific performance, and on this analysis it is not. He explained that to form a licence contract French law would require the parties to agree on both the IPR covered by the agreement and thle royalties to be payable. If the royalty was not agreed, there can be no contract and a French court would not set the applicable price. French law (Art L613-8 §5 of the Intellectual Property Code) requires an intellectual property licence to be in writing otherwise it is a nullity and so without a specific form, there can be no contract. These points were in Prof Libchaber’s final report to which Prof Fauvarque-Cosson did not respond.

These are significant issues for which I have considerable sympathy. An idea which has been canvassed and was in the pleadings at one stage in these proceedings although

Neither side before me suggests that the FRAND undertaking is specifically enforceable in the sense I have described. For my part I doubt that the FRAND undertaking can be specifically enforced in such a way that either party could legally be compelled to enter into a contract against their will. Certainly the implementer could not be so compelled and I doubt the patentee could be either. However a proper analysis of the full legal situation needs to have regard to the intellectual property rights which the FRAND undertaking relates to as well as the contractual position. It also needs to take into account competition law.

I do not believe it is necessary in order for the FRAND undertaking to be legally effective, for it to be true that the undertaking is specifically enforceable in such a way that the IPR holder could be compelled to enter into a contract against their will. In other words, even if a patentee cannot be compelled to enter into a contract by specific performance of the FRAND undertaking, that undertaking can still have substantive legal effect. As mentioned already FRAND is an objective standard. Courts concerned with patent cases in a number of countries around the world have set FRAND rates and this court will do so too. If a patentee refuses to enter into a licence which a court has determined is FRAND then, subject to the Vringo problem which I will consider below, a court can and in my judgment should normally refuse to grant relief for patent infringement. The converse applies to an implementer who refuses to accept a FRAND licence. In that case the normal relief for patent infringement should normally follow. Thus there is no need for contract law to go as far as creating a power to compel parties to enter into FRAND licences against their will because patent law already has the tools available to give legal effect to the FRAND undertaking.

Therefore, while I recognise the force in Prof Libchaber’s concern, I do not accept that the nature of the FRAND obligation itself, which is examined at length below, means that the doctrine of “stipulation pour autrui” cannot apply to it.

The efficacy of a unilateral commitment under Article 1100-1 of the French Civil Code.

As a fall back Prof Fauvarque-Cosson stated that even if no contract was formed between the IPR holder and ETSI, the FRAND undertaking could be regarded as constituting an “acte juridique unilatéral”, meaning a manifestation of a will intended to produce legal effects and thereby enforceable in French law under Article 1100-1 of the Civil Code which codifies previous law. Prof Libchaber did not agree but since I have already decided the issue on the basis above there is no need to resolve that.

Standing back I recognise that the enforceability of the FRAND undertaking in French law is not a clear cut question. Prof Libchaber stated that there remains widespread uncertainty about the issue of whether the doctrine of “stipulation pour autrui” can be

Can there be more than one set of FRAND terms?

A question which needs to be grappled with in order to resolve this dispute is whether there can be two sets of rival licence terms which are both FRAND. To put the matter another way, the question is whether there can be a FRAND range rather than just a FRAND rate. This is not a theoretical problem, it has real practical significance in this case.

From the point of view of economists, the FRAND royalty rate is the rate which the parties in a given set of circumstances would converge upon and agree to. The circumstances are idealised in various ways – for example there is no hold up and no hold out. The FRAND royalty rate arrived at in that way may not be the same as the rate actually decided upon by a court or an arbitrator and Prof Neven strikingly contemplated that the court’s judgment on a FRAND rate would inevitably be incorrect at least to some degree since it would always be an imperfect approximation to the “true” FRAND rate for a given circumstance since it would inevitably be based on limited evidence. This illustrates an important aspect of the economists’ approach to FRAND. For a given set of circumstances there is only one FRAND rate and, by parity of reasoning therefore, only one FRAND set of licence terms.

In Vringo v ZTE (both [2013] EWHC 1591 (Pat) and [2015] EWHC 214 (Pat)) and in earlier judgments in these proceedings I considered what happens if each side in a patent dispute makes a FRAND offer. As those judgments indicated, it may be that competition law, the contractual basis of the ETSI FRAND undertaking and the English court’s equitable discretion which relates to injunctions deal with these problems in different ways. This problem (the Vringo problem), in which offers presented by each party differ but are both FRAND, necessarily presupposes that different terms can both be FRAND. If that is possible then competition law and the contractual FRAND undertaking may be satisfied but the problem then may have to be resolved by the grant or refusal of an injunction. In Vringo I described as “international coercion” the effect which might arise if a court granted an injunction in its territory on the basis that a putative licensee had no licence when the reason the licensee had no licence was because the only terms on offer were a global licence which the licensee did not want.

The question is what to do once a court has decided that a given patent is valid, essential and infringed by an implementer and in which both the patentee and the implementer have offered licence terms but those terms differ. That is this case. Even if the dispute is only about a royalty rate then the problem still arises because if there can be a range of FRAND rates then asking if a rate is FRAND does not provide the court with a basis for resolving the dispute. If there can be a FRAND range then in order to adjudicate that a particular rate is the “right” rate in the circumstances either there needs to be some further principle to apply aside from FRAND or the parties would have to agree to accept whatever rate the court chooses in the exercise of its discretion. The equitable discretion relating to the injunction does not solve this problem. All that can be achieved in effect is enforcement of a determination made by the court as to what licence terms are acceptable but that does not indicate which set of terms should be accepted.

Before me the parties’ cases were diametrically opposed:

If, on the other hand, there is only one set of terms (including the rate and all other terms) which are truly FRAND in a given set of circumstances then a different problem arises. At first sight it seems uncontroversial to state that the criterion which defines whether SEP licensing conduct does or does not amount to an abuse of dominant position in EU competition law (Art 102 TFEU) is whether the licence terms are FRAND as defined by the ETSI undertaking. However if only one set of terms in given circumstances can truly be FRAND and if FRAND also represents the line between abusive and non-abusive conduct then every agreed licence in the entire industry is at a serious risk of being contrary to competition law and open to being unwound. Simply as an illustration of the potential significance of this point, consider the impact of Huawei’s submission that the global nature of a licence means it cannot be FRAND because it amounts to illegal bundling. All the numerous comparable licences in evidence are global.

Although I deal with competition law below, I can address this issue now. Focussing on rates, Art 102 TFEU only condemns excessive pricing. Both economists agreed that for a royalty rate to amount to excessive pricing it would have to be substantially more than FRAND. How much more that is will depend on all the circumstances but whatever it is, the point demonstrates that a royalty rate can be at least somewhat higher than the true FRAND rate and still not be contrary to competition law. Therefore, as a matter of principle, the boundary between what is and is not a true FRAND rate as defined by the ETSI undertaking is not and cannot be necessarily co-extensive with competition law. In other words the premise which I described above as at first sight uncontroversial, is not correct. Competition law considerations may well indicate why a rate is not FRAND but in general and as a matter of principle, for competition law to be engaged, it will be necessary but not sufficient for a rate not to be the true FRAND rate. Again, the logic applicable to royalty rates is applicable to any licence term.

In saying this I have well in mind that the CJEU in Huawei v ZTE appears to equate an obligation to make a FRAND offer with compliance with Art 102 TFEU. That is addressed below.

One possible objection to the idea that there is only one set of true FRAND terms in a given set of circumstances is that it might create legal uncertainty by allowing a party who had to agree licence terms to later contend that the agreed terms were not FRAND because they differed from the sole “true” FRAND terms. In my judgment the answer to this objection is that such an agreement has different effects from the point of view of competition law and the ETSI undertaking. In terms of competition law, the fact that terms have been agreed does not and cannot mean that they avoid the scrutiny or effect of that law. So, if the agreed terms are so far from FRAND as to contravene competition law, then that would no doubt make the terms unenforceable and have the other consequences of a breach of competition law. However in terms of the ETSI FRAND undertaking, there is no reason why the undertaking should entitle either party subsequently to challenge agreed terms as being non-FRAND absent competition law considerations. If parties agree licence terms then their rights and obligations under the ETSI FRAND undertaking will be discharged and replaced by their contractual rights under the licence. Thus having only one set of true FRAND terms for a given situation would not create a practical problem for operators based on competition law nor would it undermine the purpose and efficacy of the ETSI FRAND undertaking.

Accordingly the concept that there exists only a single set of FRAND terms for a given situation is workable. It will promote certainty and will enhance the normative aspect of FRAND. It would make the enforcement of the ETSI FRAND undertaking conceptually straightforward. If there is only one set of true FRAND terms for a given situation then a court will be able to hold parties to their obligations arising from the FRAND undertaking. Both parties would be entitled to insist on FRAND terms and neither would be entitled to insist on anything other than FRAND terms. By definition the FRAND terms are the terms which are fair, reasonable and non-discriminatory. They are the terms which a truly willing licensor and truly willing licensee would agree upon in the relevant negotiation in the relevant circumstances absent irrelevant factors such as hold up and hold out.

The fact that the evidence of putative comparable licences might show a range of rates and other terms as having been agreed between other parties does not falsify the idea that for a given situation there is only one set of true FRAND terms. Each real licence was arrived at between particular parties in particular circumstances which may or may not be good evidence about what would be FRAND in the case in issue. Furthermore the fact that the terms of a given comparable licence, objectively speaking, may not represent the true FRAND terms for the circumstances in which they were agreed does not mean those contracts would all be vulnerable to being unwound, for the reasons already addressed.

The concept of a single set of FRAND terms also eliminates the Vringo problem. That is a significant virtue. If more than one set of terms can be FRAND then the Vringo problem of rival FRAND offers cannot be solved in a fair way. I do not accept either party’s submission about what a court should do if presented with rival terms both of which are FRAND. I reject UnwiredPlanet’s submission that the patentee should win

The wider interpretation is preferable for another reason too. It is more consonant with the purpose of the FRAND undertaking itself. An obligation focussed only on making FRAND offers (my emphasis) is unrealistic since a process of fair negotiation will usually involve some compromise between the parties’ rival offers. If the ETSI undertaking demands that offers made by a patentee must themselves consist of FRAND terms, then that would condemn patentees to always end up with negotiated rates below a FRAND rate. Therefore it makes much more sense to interpret the ETSI FRAND obligation as applicable primarily to the finally agreed terms rather than to the offers. In other words, it is an obligation to enter into FRAND licences. The same logic also applies to implementers: an obligation on implementers to make FRAND offers as opposed to enter into FRAND licences would have them paying rates higher than the FRAND rate.

I have referred to an obligation on implementers because I believe the ETSI FRAND undertaking does impose duties on them too. Although the ETSI FRAND undertaking is an obligation imposed on the patentee, I agree with UnwiredPlanet that it also has the effect of creating an obligation applicable to the implementer, as follows. Although some implementers are themselves ETSI members, the ETSI FRAND undertaking must work in the same way whether the implementer is a member of ETSI or not. The implementer, as an implementer, owes no contractual duties to ETSI at all and the implementer has no duty to ETSI to offer FRAND terms to a patentee. However the logic of the FRAND undertaking means that an implementer must negotiate fairly if it wishes to take advantage of the constraint which the patentee’s FRAND undertaking places on the patentee’s rights. Just as an implementer is entitled to demand FRAND terms in a licence from a patentee subject to the ETSI FRAND undertaking, so a patentee is entitled to demand FRAND terms in the same licence. In other words, an implementer who does not negotiate fairly is not a willing licensee and may ultimately be subject to an injunction.

I also reject Huawei’s proposed solution to the Vringo problem that the implementer’s offer is necessarily the one which must be accepted. All the reasons which apply to the patentee’s terms apply just as much to the implementer’s terms and it is not correct to say that the FRAND undertaking is simply for the benefit of implementers. Its purpose is to strike a balance between the respective rights of patentees and implementers.

The considerations above illustrate that FRAND needs not only be a description of a set of licence terms but is also apt to describe the process by which a set of terms are agreed. In argument Huawei’s counsel used the expression “a FRAND approach” to describe how negotiations ought to proceed. That is accurate and it bears on the parties’ obligations arising as a result of the FRAND undertaking. Prof Neven said in his second report that:

“…a FRAND commitment is not simply a commitment to abide by the terms of a court-determined FRAND licence and / or FRAND rate, but requires the SEP holder to behave in particular ways (and for instance to make (or at least attempt to make) offers capable of being FRAND).”

(para 41)

I agree with the sentiment expressed by Prof Neven. Both patentees and implementers should take a FRAND approach to the negotiation of a licence under a SEP or SEP portfolio governed by a FRAND undertaking. The patentee is obliged by contract to take a FRAND approach to the negotiation and to grant a licence on FRAND terms. The implementer must take a FRAND approach to the negotiation and accept a licence on FRAND terms if it wishes to take advantage of the constraint on the patentee’s rights imposed by the FRAND undertaking. A FRAND approach to negotiation does not mean that parties cannot negotiate in good faith and a FRAND approach will allow for starting offers which leave room for negotiation. The fact an opening offered rate is higher than the true FRAND rate does not mean of itself that a patentee has failed to take a FRAND approach any more than the converse could be said about an implementer. On the other hand, making extreme offers and taking an intransigent approach which prejudice fair, reasonable and non-discriminatory negotiation is not a FRAND approach.

In my judgment the economist’s view of FRAND terms as a single thing for a given set of circumstances is also applicable to the question of whether terms are FRAND within the meaning of the ETSI FRAND undertaking. I find that for a given set of circumstances there will only be one set of FRAND terms and only one FRAND rate.

Therefore the solution to the problem of parties presenting rival FRAND terms to the court is simple enough. The court has to decide what terms would be FRAND in the given circumstances and can grant a declaration to that effect. Only one set of terms will be compliant with the FRAND undertaking.

A patentee who refuses to accept those terms would be in breach of its FRAND undertaking. Even if a court cannot go as far as directly enforcing the FRAND undertaking by compelling a patentee to make an offer in those terms (see the section on French law), I think an English court would at least refuse to grant a patentee an injunction if it refused to accept FRAND terms. That would be a proper exercise of the court’s equitable jurisdiction to grant or refuse an injunction.

A defendant who had already been found to infringe a valid patent cannot be compelled to accept an offer of a licence but a defendant with no licence, who had refused to accept terms on offer which had been found to be FRAND, would not be entitled to the protection from injunctions provided for by the patentee’s FRAND undertaking. An injunction would follow and to grant it would be a proper exercise of the court’s equitable jurisdiction. The only coercion in that case would be to enter into a licence on FRAND terms. It would apply to both sides with equal force.

Before finally concluding on this issue I will return to the factor which concerned me most. That is the diversity of terms in the real agreements in the industry. Does a conclusion that only one set of terms is FRAND mean that most or all of these agreements are not FRAND? I have answered this already but given the importance of the point I will repeat the answer I have arrived at in a different way. For concluded agreements between patentees subject to an ETSI FRAND undertaking and implementers, the importance of the FRAND undertaking will be historic. The process aspect of FRAND was important in requiring both sides to approach the negotiations appropriately and the requirement that a royalty rate had to be FRAND would be something to be prayed in aid during the negotiations. However once the agreement has been reached the contract must be the thing which governs the rights and obligations of the two parties with respect to each other while it is in force. Competition law must leave latitude to the parties to agree and cannot draw the line between acceptable and unacceptable contract terms in the same place as the line between whether a term is ETSI FRAND or not.

Can the court set a FRAND rate or other FRAND terms?

Having now heard this trial I remain of the view that the court cannot craft a set of FRAND terms out of thin air. That is what I described in Vringo as a Copyright Tribunal type exercise although it may be noted that even in the Copyright Tribunal there must be a licensing scheme in order for the Tribunal’s jurisdiction to be engaged. However courts all over the world have now set FRAND rates. I am sure the English court can do that as well. I decided at the CMC in March 2015 that the court could declare that a given set of terms were FRAND and also, within that framework, can decide that certain terms need to be adjusted in order to make a set of terms FRAND. I am sure this applies to a royalty rate. After all, arriving at a FRAND royalty rate is not different conceptually from assessing what a reasonable royalty would be in a patent damages enquiry albeit the particular factors applicable in setting a FRAND royalty for a licence to be FRAND and their application may differ from assessing damages. So the court’s jurisdiction is not restricted to the binary question of assessing a given set of terms but extends to deciding between rival proposals and coming to a conclusion different from either side’s case on such a proposal.

How to assess what is FRAND

There was no real dispute of principle about how to work out what is and is not FRAND. The question is what would be fair, reasonable and non-discriminatory. Asking what a willing licensor and a willing licensee in the relevant circumstances acting without holding out or holding up would agree upon is likely to help decide that question. The evidence of the parties themselves will be relevant, including evidence of how negotiations work in practice in the industry. To the extent they are available other licences may be deployed as comparables. Just as comparables may be useful in a damages enquiry when considering a reasonable royalty and may be useful in determining the terms of a licence of right or in a Copyright Tribunal, so comparables may be useful in deciding what is FRAND. As always judgments will have to be made about how closely comparable any given licence is to the relevant circumstances in issue. The relevance of comparables is that they are evidence of what real parties in real negotiations have agreed upon. But like any real situation many factors may have been in play which make the licence less relevant. The negotiations may have involved

The decisions of other courts, assuming they are not binding authorities, may be useful as persuasive precedents. A point arises in this case about a licence which was the product of an arbitration. A licence agreement settled in an arbitration is more like terms set by a court than it is like a licence produced by negotiation and agreement. Huawei submitted that such a licence would be evidence of what a party was actually paying and as such was relevant. Aside from certain aspects of non-discrimination which I will address separately, I do not accept that evidence of what a party is paying as a result of a binding arbitration will carry much weight. If the licence is the product of an arbitration then the paying party has no choice. A further difficulty with the particular licence in question is that the arbitral award has not been produced. So although we know what the licence terms are, we do not know what the reasoning was which led to them. As a persuasive authority an arbitrated licence without the arbitral award is not much use. There were a few references in the evidence to the way the arbitrators decided the case but without seeing the award itself I will not place weight on that.

In relation to comparables generally Huawei submit that the approach to be followed is that set out by Lloyd LJ in Smith Kline & French Laboratories Ltds (Cimetidine) Patents [1990] RPC 203 as follows:

“The object of the comparability exercise, in this as in any other branch of the law, is to find the closest possible parallel. If there is an exact parallel, there is no point in looking any further. If there are slight differences, an allowance may be made. But once you have found your comparables, whether one or more, which enable you to arrive at the appropriate figure, it would surely be erroneous to modify that figure by reference to other cases which are not truly comparable at all, so as to bring the case into line with a predetermined range. This was, with great respect, the mistake which the hearing officer made.”

Huawei argue that Mr Lasinski’s approach (which was to select the two or three “best” comparables and rely on those) accords with these principles. On the other hand, Huawei criticise Mr Bezant’s approach of including many more licences. I do not accept that criticism. In my judgment, if a group of comparables are at least potentially as relevant as each other and are not the same, it is not right to elevate a small subset above the others. That is also not what Lloyd LJ in Cimetidine said one must do; instead, he said that, assuming there is no exact parallel, once true comparables have been determined one should be careful not to dilute them by reference to other cases which are not truly comparable at all. Mr Bezant’s general approach does not do this.

If a group of good comparables corroborate one another then no doubt that is a factor to take into account but equally if apparently good comparables, when properly understood, contain different rates that is also relevant too.

Huawei also submit that the comparables selected should include some, or ideally all, of three criteria: (a) the licensor is UnwiredPlanet or Ericsson, (b) the licensee is Huawei, or a similarly situated company such as Samsung and (c) the licence is recent. I agree with (a) and subject to what “recent” means I agree with (c). However I am not convinced that (b), the identity of the licensee, should be a strong factor in determining

In argument the rates on which both sides’ submissions were based were derived from global rates for the whole SEP portfolio albeit that Huawei’s case is that the licence should be a UK licence. This worked because both sides agreed that the correct way to arrive at a UK portfolio rate was by starting from a global rate in effect as a benchmark and then adjusting upwards. Conceptually the approach was common ground although the level of the correct adjustment was in dispute. This approach, in which a rate is determined as a benchmark and then adjustments made as appropriate, is a useful way of determining what a FRAND rate or rates should be. Arriving at a benchmark FRAND rate is a neutral way of making appropriate findings. In this case it caters for the parties’ rival cases about what the territorial scope of the licence should be.

A hard-edged non-discrimination aspect of FRAND

Some arguments were addressed to the non-discrimination (“ND”) aspect of FRAND as opposed to the “FR” aspect of FRAND as if they were distinct. However it is not that simple. Most of the time the concepts of non-discrimination, reasonableness and fairness relate to one another. In that sense it is useful to characterise a royalty rate as FRAND rather than try to distinguish between something which is merely fair and reasonable as opposed to fair, reasonable and non-discriminatory. The argument about non-discrimination treated it as a concept which would apply to reduce a royalty rate even if that rate was otherwise “FR”. For want of a better expression, I will distinguish between a “hard-edged” and a “general” non-discrimination obligation. The general non-discrimination obligation is the aspect of non-discrimination which I have mentioned already. It is part of an overall assessment of the inter-related concepts making up FRAND by which one can derive a royalty rate applicable as a benchmark. This rate is non-discriminatory because it is a measure of the intrinsic value of the portfolio being licensed but it does not depend on the licensee. The hard-edged nondiscrimination obligation, to the extent it exists, is a distinct factor capable of applying to reduce a royalty rate (or adjust any licence term in any way) which would otherwise have been regarded as FRAND. This will take into account the nature of the particular licensee seeking to rely on it.

Concepts used to derive a FRAND rate with telecoms standards

The FRAND royalty rate reflects an assessment of the value of the licensor’s patent portfolio under licence. Two approaches to this have been taken both in this case and in other cases. One approach (referred to as “top down”) starts with a number representing what the appropriate total aggregate royalty burden should be for a given standard (call it T). One can take a view about what the total royalty burden for all the intellectual property relating to the standardised telecommunications technology in a handset should be and indeed various companies have made public statements about this. Starting from this figure T one can then share out the royalty across all licensors in proportion to the value of each licensor’s patent portfolio based on assessing that value as a share (call it S) of the total relevant patent portfolio essential to that standard. The FRAND rate is the product of the two (TxS).

The other approach is to use comparable licences. These are licences which have already been entered into. The most directly comparable licences will be licences the patentee has already entered into for the portfolio in question. There are two in this case, the UnwiredPlanet-Lenovo 2014 licence and the UnwiredPlanet-Samsung 2016 licence. One might assume directly comparable licences would represent the best evidence of the value of the portfolio in issue. However the rates in these two licences are very different from each other and each side contends that one of them is not a useful comparable at all. I will return to those licences below.

Given that at one time Ericsson, Samsung and Huawei were all parties to these proceedings and they are all major telecommunications companies, a large number of further patent licences were in disclosure. (The disclosure had been managed to keep the numbers under control but it was still substantial.) The comparison with third party licences is indirect and the relationship between those licences and the value of the portfolio in issue will depend on the evidence. In order to use them a view has to be formed about the relative value of the portfolios licensed in them as against UnwiredPlanet’s portfolio. Since the relevant UnwiredPlanet patents all came from Ericsson, the Ericsson licences at one time included all the SEPs in issue. That alone makes Ericsson licences relevant. So if the rate for Ericsson’s portfolio is E and the relative value of UnwiredPlanet’s portfolio to Ericsson’s portfolio is R, the UnwiredPlanet rate is ExR.

The factors S and R are measures of the value of a licensor’s patent portfolio relative to the industry as a whole and to another licensor. It is clear that in negotiating licences in this field the parties seek to make an assessment of this value. Tools for doing this were sometimes called portfolio strength metrics. One might think that in order to do this it would be necessary to examine the value of the contribution made to the standard by the invention claimed in each patent. Obviously as a portfolio increases in size the burden of that increases too but one of the reasons this trial was docketed to the same judge who heard the technical trials was because I would be familiar with some of the patents chosen by UnwiredPlanet to litigate, which one might expect would be the good ones. However the exercises conducted by both UnwiredPlanet and Huawei for this trial, subject to a point on Ericsson, have been based on categorising and counting patents. The techniques treat all patents in a given category as of equal value.

There was ample evidence before me that apart from Ericsson (see below), parties negotiating SEP licences in fact use methods which are based on patent counting. That is evidence which supports a finding that a FRAND approach to assessing a royalty rate is to engage in some kind of patent counting. Indeed when one thinks about it some sort of patent counting is the only practical approach at least for a portfolio of any size. Trying to evaluate the importance of individual inventions becomes disproportionate very quickly.

It may be that other technology standards are different but I am not surprised that patent counting is the approach taken for GSM, UMTS and LTE telecommunications standards. Each standard defines a system with a large number of different parts all of which have to interact with each other. The interactions and interdependencies are complicated. To make a coherent system which works at all, let alone one which delivers the performance demanded of these systems, is difficult and demands insight and creativity on the part of the engineers involved. It is unsurprising that many inventions (and therefore many patents and SEPs) will be involved. Short of the disproportionate task of evaluating every single patent thoroughly in order to compare each one with all the others, one can only ever hope to analyse SEPs in broad categories and it is not meaningful to attempt to weigh the value of individual patents within these categories against one another.

I suppose in some cases it may be possible to identify a patent as an exceptional sort of keystone invention underpinning the entire technical approach on which a standard is based but that is not this case. There was unchallenged evidence that UnwiredPlanet’s patents made an “average” contribution to the standards. I am satisfied that none of the UnwiredPlanet patents are in the exceptional keystone category.

The evidence is that Ericsson sought to deploy a different technique in licensing negotiations based on evaluating a party’s technical contributions to the standard setting process as a way of valuing their portfolio and Mr Lasinski used this method for

“unpacking” Ericsson’s licences (see below). Using it as a technique to address Ericsson’s licences is logical since it is an Ericsson technique, and Mr Bezant and Mr Lasinski were in agreement that it made more sense to use a metric of strength for unpacking which was available to the parties negotiating a licence at the time rather than one which was not available. However the Ericsson technique has problems if applied more generally and neither side suggested that it should be. For one thing it is already at one remove from the legal rights, which derive from patents not technical contributions. Also the technique cannot handle a portfolio of patents acquired after the standards were set – e.g. the UnwiredPlanet portfolio. Ericsson have been closely involved in the standard setting process and that may be why they like this method, I do not know. In any case the fact that Ericsson advanced arguments on this basis during negotiations does not mean it is accepted as a method by the counterparty.

The patent counting approach works in the following way. Starting from a portfolio of declared SEPs the first task is to derive a number representing the Relevant SEPs.

“Relevant SEPs” is my term, coined after the trial had finished and intended to avoid language used in the case which can be confusing such as “truly essential patents” or “deemed”. Both sides’ approaches require making an assessment of the Relevant SEPs somehow. The parties do not agree how it should be done but one way or another a number is produced. Armed with that information it is possible to scale one company’s rates relative to another to derive the factor R or to find the share of the total and derive S.

A significant dimension to the task to evaluating comparable licences is the fact that many patent licences in this industry have terms which make the comparison difficult. The two major problems are that they may be based on a lump sum rather than a running royalty and they may be cross-licences with a balancing figure which may be a rate or

It was common ground as between Mr Bezant and Mr Lasinski that to address the lump sum and the cross-licence problem a process of “unpacking” has to be carried out. The unpacking can derive a notional royalty rate from a lump sum by treating the lump sum as the net present value of an income stream from running royalties analysed using a discounted cash flow based on some appropriate estimate of sales figures. This introduces uncertainties since it needs to have some estimate of sales figures which may be historic and/or future estimates and which in any case may not represent what the negotiators thought when they negotiated the licence.

The unpacking of a cross-licence can resolve two one-way royalty rates from a single balancing figure based on the notion that the single figure represents the effect of balancing the value in royalty terms of each party’s patent portfolio. If the balancing figure is a lump sum then unpacking will involve net present value assessments for each party with the attendant uncertainties. In any event there also needs to be some means for assessing the relative value of each party’s portfolio unless one has a figure for one or other party directly. To achieve this takes one back to the Ericsson contributions technique and/or counting patents. For a cross-licence between A and B, if A has 100 Relevant SEPs and B has 200 then the ratio is 1:2 and that allows one mathematically to derive figures for the underlying one way rates. Inevitably this introduces yet more uncertainty.

Cross-licences will generally be entered into by companies who are both implementers and licensors at the same time. There is a risk that the rates agreed in a cross-licence understate the inherent value of the rights being licensed because the revenue the parties earn from licensing itself will be much less than it would be if the licence was not a cross-licence but it is impossible to evaluate how significant a difference this might make.

For a given comparable licence to which at least one party to the litigation was a party, an issue arose at the case management stage as to whether additional disclosure should be given of the documents and circumstances surrounding the negotiation of that licence in order to enhance its utility as a comparable. I refused that disclosure across the board because I did not think conducting a post mortem examination of the negotiations would be proportionate. Although in theory it might have obviated the need for unpacking, having now heard the trial I am sure that this disclosure would not have helped with unpacking and would just have generated further argument without advancing the issues. Although unpacking involves significant uncertainties, in fact the arguments about unpacking itself were quite minor. As Huawei pointed out in closing about [a] licence:

“the details of the unpacking process make little significant difference to the implied […] rates: the rates ascertained by Mr Bezant, Mr Lasinski or Dr Leonard were all in a relatively close range ([…]), whichever unpacking method was used {U1/6/1}. As Mr Lasinski explained, the portfolio strength metrics employed by the three valuation experts have a minimal effect on the effective rates they derive from the […] Licence because […] that this factor swamps almost everything else.” [closing paragraph 77]

A number of points emerge from this. The parties’ experts (Dr Leonard was to have been Samsung’s expert) had analysed this licence in order to derive an implied […] royalty rate charged by […] as licensor and accepted by […] as licensee for [the] portfolio of SEPs. The parties have come to figures for the effective implied royalty rate for [… ]portfolio using very different techniques but the answers all came to a number which Huawei characterise as a relatively close range. The only aspect of the submission I do not accept is the qualification “relatively”. In my judgment bearing in mind all the uncertainties and assumptions which go into these unpacking exercises, the spread of these figures is remarkably close. The spread is about ±20% around the midpoint ([…]).

In the particular example of the [X-Y] licence another factor relevant to the unpacking exercise was a key assumption which had to be made about a different aspect of the agreement. This is about […].

Absent the figures from Dr Leonard (which make no difference) this range at {U1/6/1} was put to Mr Bezant in cross-examination who explained (T11 p81-82) that:

“The differences are not that great on the unpacking, in the context of the exercise. […] And, indeed, given the uncertainty of unpacking cross licences. It's not as if one number is strong and another it is weak. They are -- they are all somewhat fragile when you're unpacking a cross licence.” and

“But I'm just signalling that when I say not sensitive that's partly the numbers don't move very much, but it's also partly a recognition that the numbers themselves are inherently uncertain.”

Part of the point Mr Bezant was making was that the rival patent counting methods do not have a major impact on unpacking. That is because in the unpacking process the experts use the methodologies consistently. One also needs to bear in mind that the numbers themselves are inherently uncertain. I accept Mr Bezant’s evidence about that (which was not in dispute). There is an exception on unpacking methodology which relates to the two […] licences but for reasons addressed below, in the end that does not matter.

There are further complications in the analyses which I have not mentioned so far such as handling the effect of discounts, royalty floors, royalty cap, release periods and passthrough licences. Mr Bezant and Mr Lasinski did not agree how best to deal with these but I am not satisfied it is necessary to delve into the minutiae to the degree necessary to resolve those arguments. The relevant point is that they are all sources of uncertainty rather than automatic reasons for completely ignoring a licence. That was Mr Bezant’s approach and if it is necessary to resolve a debate on that score at a methodological level then I prefer Mr Bezant’s approach to Mr Lasinski’s.

In order to determine what the values E, R and S are in this case two tasks need to be performed. To determine S (UnwiredPlanet’s share of the total) and R (the relative value of UnwiredPlanet’s portfolio to Ericsson’s) it is necessary to count Relevant SEPs. To determine E (the rate for Ericsson’s portfolio) it is necessary to consider the comparable licences. At the same time the UnwiredPlanet comparables and other evidence on rates can be addressed.

Relevant SEPs – shares and ratios

An area of dispute which makes a major difference to the final royalty rate is how to count Relevant SEPs. As I have explained, R and S are ratios which come from dividing UnwiredPlanet’s number of Relevant SEPs by a number for another licensor such as Ericsson or the industry as a whole.

For Huawei’s case both the numerators and the denominators in these ratios are derived using the same patent counting technique called the Huawei Patent Analysis (HPA).

It is common ground that some kind of appropriate methodology is needed beyond simply adding up patents on the register or the ETSI database. One needs to cater for the different jurisdictions, divisionals and other things. Some of this can be dealt with by focussing on families rather than patents (but that is not perfect either) however a very significant reason why one cannot just count up declared patent families is recognition of the problem of over declaration. There was no dispute this exists. The debate is as to its extent.

The over declaration problem is the following. Very many more patents are declared to be essential than in fact are essential. This can be for many reasons. For ETSI members Art 4.1 of the IPR Policy requires members to declare essential patents in a timely fashion and creates an incentive to err on the safe side and so, if in doubt, declare. Also determining essentiality for certain is not easy. The technology can be difficult and the patents and the standards can be hard to interpret. Patent claims are also amended over time and in a single family the different national patents will vary in scope around the world; standards themselves can also vary over time. Keeping track of all this would be time consuming and costly, and if reasonable royalty rates can be agreed without determining essentiality for certain, it is a disproportionate task. Notably also no-one tries to take account of validity. Various studies have been done on overdeclaration and rates of over-declaration quoted in the literature. Each side criticises the other’s counting techniques and specifics over over-declaration are addressed in the sections below dealing with the alleged flaws in the techniques. It is just too difficult.

Nevertheless it must also be recognised that the fact that rates are negotiated by counting patents creates a perverse incentive to declare as many patents as possible, making over-declaration worse.

A further point is that UnwiredPlanet’s approach only uses patent counting for the denominators. The numerators, in other words the numbers representing UnwiredPlanet’s own Relevant SEPs for a given type of technology, are the result of a detailed assessment of the individual patent families. UnwiredPlanet say that is the appropriate thing to do and it is inappropriate to take the approach advanced by Huawei by using the same counting technique for both numerator and denominator. Huawei say the opposite and UnwiredPlanet’s approach is inconsistent whereas their approach is the correct thing to do. I will deal with that at the same time as other criticisms, below.

The parties are very close on the numerators and far apart on the denominators. In other words, at least superficially, they are close on the number of Relevant SEPs in UnwiredPlanet’s portfolio. In any event they are far apart on the number of Relevant SEPs in other companies’ patent portfolios or as a whole.

For example each side contends UnwiredPlanet have 6 relevant LTE SEP families for handsets. They arrive at this number in different ways but they both arrive at 6. The complete set of relevant numbers for UnwiredPlanet’s patents are shown in these tables:

Huawei’s case:

UnwiredPlanet’s case:

RAN stands for Radio Access Network. It is a major part of the infrastructure of these systems. There is another kind of infrastructure which relates to the core network but it is common ground that this is a totally different market. Note that in the Huawei table the total column is not a simple sum of the numbers for handsets plus RAN infrastructure because one patent can cover both.

Two detailed assessments have been made of UnwiredPlanet’s patents. First, as part of their licensing efforts Mr Saru explained that they (UnwiredPlanet) carried out their own detailed assessment of the patents in their LTE portfolio (Mr Saru I para 51). There were 19 LTE families to start with and UnwiredPlanet decided they held 9 of what they called the True LTE families. Second, in these proceedings Dr Cooper carried out a detailed assessment of some UnwiredPlanet patents. For LTE the patents assessed were 7 of the 9 (because the other 2 had been litigated in trials A and B and found to be essential). The detailed assessment Dr Cooper carried out was the same as he carried out on certain Samsung and Huawei patents which come up below in the context of the MNPA. I accept Dr Cooper’s assessment. As for the two litigated patents, Huawei submitted the Trial B patent should have been classed as optional

For 2G/ GSM and 3G/UMTS the position is more complicated. I find that for the purposes of assessing a FRAND rate for 3G UMTS in these proceedings, UnwiredPlanet have 1 handset and 2 infrastructure patent families which are essential. The corresponding numbers for 2G/GSM are 2 handset patent families and 1 infrastructure patent family. There is an issue about the way UnwiredPlanet deal with 2G/GSM and 3G/UMTS in relation to the Original MNPA because UnwiredPlanet included as essential patents which did not meet the MNPA cut offs. That may make a difference with regard to the FRAND status of the 2014 offers and I will address that in context if necessary.

The following tables are the same as the previous ones but incorporating the parties’ rival denominators for the industry as a whole and the resulting fractions which are produced. These tables show the values for S, UnwiredPlanet’s share of Relevant SEPs:

Huawei’s case:

UnwiredPlanet’s case:

Now the major differences between the parties can be seen. UnwiredPlanet contend their patents represent 1.69% of the Relevant SEPs for handsets in LTE, in other words S = 1.69% whereas Huawei say the portfolio only contains 0.33% of those SEPs and so S= 0.33%. In other words, on UnwiredPlanet’s case, the value of their patents for handsets in LTE is five times the value contended for by Huawei. A dimension which I have not mentioned yet is how to deal with multimode devices, that comes in the next section.

Similar differences between the parties’ contentions arise when looking at the ratio R between UnwiredPlanet’s SEPs and Ericsson. They arise for the same reasons because each side uses the same methods to count for the industry as a whole and for individual companies.

What accounts for the difference here is the degree to which the rival techniques reduce the number of relevant patents. The starting points are similar but the end points are different. Huawei suggest that the total number of patent families declared essential to 4G/LTE, making certain assumptions, is 6027. UnwiredPlanet used a corresponding figure of 5917, produced in a different way. However the outcome of the HPA, for the number of Relevant SEP families for 4G/LTE handsets is 1812 while UnwiredPlanet’s equivalent is 355.

It will be recalled that by numbers UnwiredPlanet’s patents represented about [Z%] of Ericsson’s portfolio when they were assigned. Huawei contend this [Z%] size ratio acts as an anchor point for testing the credibility of each side’s case on relative E:UP portfolio strength R. Conceptually Huawei is correct although one needs to take care with inherent uncertainties in many of these numbers and with the fact that a small sample size from a large population may not be representative.

Huawei derived values for R at different points in time – pre-MSA; post-MSA preLenovo; and post-MSA post Lenovo. The pre-MSA data is in the following table.

Huawei’s case:

The [A] figures for UnwiredPlanet are the numbers in the totals column for Huawei’s case on the numerators, as before. The [B] figures for Ericsson are the numbers of relevant SEP families in Ericsson’s portfolio before the MSA. All the figures in the table are produced by the HPA. […]

The figures for the two later points in time (based on corrections made in chief and in a Powell Gilbert letter) are:

Huawei also produce a similar set of values for R using numbers of declared patents (making certain assumptions). Huawei submit these numbers are not irrelevant but are not the ones to place much weight on. I include them in the judgment to illustrate the differences which can arise when declared numbers are used. The table for pre-MSA is:

Huawei’s case:

UnwiredPlanet’s rival strength ratios R are presented in a slightly different way whereby the numbers [A] and [B] are presented as the % ratios of either company’s Relevant SEPs to the total number of Relevant SEPs, but the result comes to the same

UnwiredPlanet’s case:

[…] This is the sort of variability which UnwiredPlanet submit indicates the caution which must be exercised when comparison is made to the [Z%] figure. UnwiredPlanet also contend that they do not rely on these individual strength ratios but submit that the right strength ratio to use, if one takes this approach, is a blended ratio taking into account multimode, which is dealt with in the next section. Huawei contend that the idea that UnwiredPlanet acquired […] of Ericsson’s relevant SEPs for LTE is fanciful and this is evidence which shows the flaws in UnwiredPlanet’s approach to patent counting.

It is common ground that one needs some weighting method in order to deal with multimode devices and both sides used the same basic approach, which for multimode LTE devices is to weight the numbers LTE/UMTS/GSM in proportions 70:20:10. For a UMTS multimode (i.e. UMTS/GSM) device both sides used 67:33 UMTS/GSM.

Multimode is a concept which really applies to handsets rather than infrastructure although Huawei did provide figures for weighting the infrastructure numbers. I am satisfied that the FRAND approach would be to weight handset rates but not infrastructure rates because while multimode handsets are very common (subject to a point on China) multimode infrastructure is not.

Applying the multimode weighting factors produces the following results for S, that is UnwiredPlanet’s share of Relevant SEPs overall:

Huawei’s case:

UnwiredPlanet’s case:

The multimode tables above also include references to UnwiredPlanet’s 80:20 approach. That is addressed in the next paragraph. In the course of writing this judgment a small point arose on the RAN infrastructure figures shown in italics in the table above on UnwiredPlanet’s case. Huawei’s Databook produced in closing showed the 0.88% figure in the LTE/UMTS/GSM row and “not given” in the row for UMTS/GSM. Although infrastructure weighting is not important, even bearing that in mind this did not make sense and after considering the written materials it seemed that there had been a muddle about numbers and about LTE/UMTS/GSM and UMTS/GSM. I worked out that 1.88% would be the number for LTE/UMTS/GSM while 0.88% was what the number for UMTS/GSM would be. I wrote to the parties. 0.88% is the right number for UMTS/GSM. 1.88% is what the LTE/UMTS/GSM number would be although Huawei rightly pointed out that UnwiredPlanet had not derived it before. I will include 1.88% because it is simple maths, there was a muddle in UnwiredPlanet’s FRAND Statement of Case and because it cannot prejudice Huawei.

The 80:20 approach derives a ratio which consists of 80% of UnwiredPlanet’s share of

Applying the multimode weighting produces the following results for R, that is UnwiredPlanet’s strength ratio to Ericsson (pre-MSA) on the same bases as before:

Huawei’s case (C13/3/2, in part in databook p8)

UnwiredPlanet’s case:

UnwiredPlanet say that the […] figure for the strength ratio R between their portfolio and Ericsson’s is the right one to use if one is going to draw a comparison with the [Z%] numerical size ratio and the two are not so far apart as to be out of line. Huawei contend the opposite, […] is too far from […] to be realistic.

Having started to set out each side’s case I will mention a problem inherent in that and in grappling with the cases. Both via the valuation experts and in their submissions both sides presented the court with a blizzard of figures. The summaries in this judgment represent a small fraction of the numbers presented. There was a somewhat larger blizzard from UnwiredPlanet than from Huawei but the difference was not significant enough to make a difference. A frequent problem is in keeping track of the bases on which numbers are presented so as to try and make sure one is comparing like with like. In practice, for example, it is impossible to ensure that on every occasion two rival figures are both based on the state of the Ericsson portfolio (pre- or post-MSA etc.), and multimode weighted or not, as well as many other more subtle factors.

From now on I intend to put most weight on figures derived from the post-MSA postLenovo portfolio. The effect of the differences between the three versions of the Ericsson portfolio is smaller than the effect of other uncertainties inherent in the various exercises. While I have aimed to be consistent, striving for perfect consistency with the numbers in this case is not productive.

In the discussion below the terms HWLTER and UPLTER refer to the parties’ rival values for the relative strength ratio R between UnwiredPlanet and Ericsson for 4G/LTE multimode. HWLTER is […] and UPLTER is […].

I will mention rounding briefly. Obviously 0.36 and 0.37 both differ from 0.40 and can all be rounded appropriately to 0.4. It is not mathematically correct to say that 0.36 can be rounded to 0.40. Nor is it mathematically accurate to say that 0.36 is less than 0.4. However the assessment is not a purely mathematical one and the inherent uncertainties in the evidence are much greater than the difference between 0.36 and 0.40. So I will try (a) to state numbers in the evidence in the form they appear, but also (b) to use two significant figures in giving my reasons but (c) not to get too hung up on it all. This will involve heresies like “0.36% can be rounded to 0.40%”.

Related to this is the point that many of the numbers written in the comparable licences are obviously round numbers (such as a royalty in dollars of 1$ for 2G, 2$ for 3G and 4$ for 4G). It is not wrong to unpack these into rates expressed as percentages like

The parties’ rival submissions on royalty rates

To arrive at an equivalent benchmark rate on Huawei’s case one needs to know that the UK uplift applied by Huawei is 48.51% based on […]. Stripping out this uplift from Huawei’s October 2016 proposals and rounding to two significant figures comes to: i) for 4G/LTE: infrastructure 0.041%; mobile devices 0.040%; ii) for 3G/UMTS: infrastructure 0.031%; mobile devices 0.031%;

Mobile devices and handsets are the same thing. Infrastructure refers to RAN infrastructure. In argument the parties focussed on the rates for 4G/LTE multimode handsets and I will do the same. To recap, UnwiredPlanet’s case is that the FRAND rate for its global SEP portfolio for 4G/LTE is 0.13% whereas for 2G/GSM and 3G/UMTS the rate is 0.065%.

Of course a rate of 0.13% is just over triple 0.040% but all the same these numbers demonstrate that the parties are not now so far apart as some of the rhetoric at trial might have led one to believe. At the start the rival benchmark rates differed by an order of magnitude (0.2% for 4G/LTE from UnwiredPlanet and 0.022% for Huawei (based on the 0.034% UK offer and stripping out a 48.51% uplift). Nevertheless the difference is still substantial when one bears in mind that as a royalty it is to be applied to very large revenues.

Huawei’s opening case was encapsulated by the following chart (best seen in colour) which was figure 4 of Mr Lasinski’s 3^rd report and was presented in Huawei’s written opening submissions:

The dashed grey and red solid lines are UnwiredPlanet’s proposals (the October 2016 rates are the same as in July). The directly comparable UnwiredPlanet-Samsung 2016 rate is shown as a green block, three comparable Ericsson licences are shown as blue blocks and the top-down aggregate royalty burden rate is yellow. It is marked “Patent Analysis” or “Huawei Patent Analysis”.

The three Ericsson licences Huawei contend are most probative are [J] licence, the [K] licence, and the [L] licence. The last one is treated as a 2G/3G licence.

The rates for 4G/LTE in the chart are derived in the following way. From the [J licence] one starts with Mr Lasinski’s preferred rate of […] as representative of [a] royalty rate for 4G, […]. Then a relative strength ratio R of […] is applied to produce an effective UnwiredPlanet rate of […]. Note that the […] was taken as R based on the portfolio post-MSA, before Lenovo and using multimode weightings but that figure had been corrected in chief to […], see a letter from Powell Gilbert dated 18^th November 2016 which also explained the change made no difference, as indeed it does not. Mathematically the number ends up as […] as the effective rate for UnwiredPlanet, which is still rounded to […]. Using the HWLTER of […] makes no difference either. Of course using the UPLTER of […] produces a rate of […].

UnwiredPlanet do not deny that the [J] licence is one relevant comparable but contend it should be seen as one of many. In addition, UnwiredPlanet point out that the Ericsson rate E used by Mr Lasinski is based on […]

The 2016 UnwiredPlanet-Samsung licence involves […] and, like the [J] licence, it has other complications too. One complication it does not have is scaling by strength ratio since it is an UnwiredPlanet licence not an Ericsson licence. Mr Lasinski’s evidence is that making many allowances in UnwiredPlanet’s favour to increase the effective rate […]. This, say Huawei, indicates that […] for UnwiredPlanet is generous.

UnwiredPlanet do not agree with this. First they say that the UnwiredPlanet-Samsung licence is not a useful comparable at all because it must be seen in a wider context of a developing relationship between PanOptis and Samsung. Second, they say that in truth the rates in this licence are pitifully small, much smaller than those derived by Mr Lasinski, which is said to be a reflection of the first point.

The [K] licence has […]. [the rate is…] for 4G multimode handsets. Therefore using a HWLTER of […] (as for […]) produces an effective UnwiredPlanet rate of […].

The key terms of the [K] licence, at least the royalty rates, were decided in an arbitration rather than being negotiated between the parties. UnwiredPlanet say this completely undermines the utility of this licence as a comparable. Huawei do not agree.

Huawei’s top down case on aggregate royalty burden derives a 4G benchmark handset rate of 0.028%. This was based on starting from a total burden T of 8% (based on a figure stated publicly by Ericsson as the maximum aggregate royalty range) and taking the appropriate share of the whole industry S as 0.36% for multimode. (I wondered if 0.30% should have been used for handsets but nothing turns on the difference.)

So for 4G/LTE Huawei contend the two Ericsson comparable licences have […] and based on Huawei’s case for the strength ratio (R = […]) that gives a rate for UnwiredPlanet of […]. Huawei contend that the 2016 UnwiredPlanet-Samsung rate is lower at […] and so is the top down case at 0.028% but all this goes to show is that […] is generous to UnwiredPlanet.

There is an issue on “hard-edged” non-discrimination arising from the 2016 UnwiredPlanet-Samsung licence. If it is accepted then Huawei contend the rates applied to them should be same ([…]for 4G) as the rates in that licence. I will treat this as a distinct issue.

UnwiredPlanet’s opening case cannot be encapsulated with a single chart in quite the same way as Huawei’s largely because Mr Bezant produced so many charts. The charts are best seen in colour. This is one from Bezant 6, Appendix 2 (U/10/p1):

The blue “offer line” is 0.13%. The various bars are rates derived from different sources. Although the key contemplates one source was one way lump sum rates, in fact there are none in this chart. The ARR comes from the MSA and […] refers to a rate from the Lenovo licence with an adjustment. All the other bars come from Ericsson licences. For each Ericsson licence the bar is an example of ExR in which the value E is different. There are more bars than licences because Mr Bezant has derived multiple rates from the same licence in various cases.

UnwiredPlanet argue that no individual comparable, particularly the ones singled out by Huawei, can bear the weight Huawei place on it. UnwiredPlanet also argue that this chart indicates the existence of a wide spread of rates in practice.

The key thing about some of the Ericsson licences that UnwiredPlanet rely on in addition to the licences relied on by Huawei is that in them Ericsson is […]

[…]

[…]

[…]

[…]

The 2011 Ericsson-RIM licence […]

That is sufficient to understand how UnwiredPlanet put their case. There is no need at this stage to address the other Ericsson licences in the chart. The further evidence UnwiredPlanet rely on can be put into four groups: the ARR from the MSA, publicly stated rates, the 2014 UnwiredPlanet-Lenovo licence, and licences from other licensors.

The term of the MSA which UnwiredPlanet rely on as a comparable is the ARR, which is […]. Huawei make the point that this is self-serving given that Ericsson benefit from royalties paid to UnwiredPlanet. As a tool for assessing a benchmark FRAND rate today the ARR has no value.

The public statements about rates are addressed below. In terms of top down aggregate royalties generally, UnwiredPlanet contend that while it may be useful as a cross-check in certain circumstances, it is based on a false premise that manufacturers in fact pay everyone who owns any portion of the relevant pool whereas in practice they do not. Nevertheless they also point out that using their case for their share S of the Relevant SEPs (say 1.25%) and applying it to a total aggregate royalty burden T of 8% or 10% produces rates close to their preferred rate.

Turning to the 2014 UnwiredPlanet-Lenovo licence, on its face it contains a lump sum licence payment of […] and running royalties creditable against that lump sum of […] per product in defined “Major Markets” (the MM rate) and […] per product in other territory (the OT rate). In percentage terms […] compares favourably with the 0.2% demanded by UnwiredPlanet in 2014 and maintained until July 2016. Huawei contend that to rely on the stated rates is to ignore the true economics of this agreement. Huawei also point out that […].

The licences from other licensors which UnwiredPlanet addressed, at least at the start of the trial (see opening p103), are licences from Qualcomm (to Huawei (two) and Samsung), licences from InterDigital (again to Huawei and Samsung) and two licences in which Samsung were licensee (from […] and Nokia). To the extent they are significant they can be addressed in context.

In closing I pointed out to the parties that there seemed to be a broad equivalence about their rival cases at least in one respect. It can be seen in the implied aggregate royalty rate. Huawei contend the benchmark multimode 4G/LTE handset UnwiredPlanet rate should be 0.040% and Huawei contend that UnwiredPlanet’s share S of multimode LTE handset patents overall is 0.30%. Conversely UnwiredPlanet contend the final royalty rate should be 0.13% and contend their share S overall is 1.25%. The ratios of these two pairs of figures are close and the similarity can be expressed in terms of the implied total aggregate royalty burden T. On Huawei’s figures the implied total aggregate royalty burden T would be 13.3% while for UnwiredPlanet it would be 10.4%.

Huawei resisted this characterisation of the arguments. Their top down approach starts from a value of T of 8% and works the other way to a royalty rate of 0.028% to support, as generous, a conclusion that the royalty should be 0.040%.

This has caused me to address the question of whether the total aggregate royalty approach is better used as a top down method or as a cross-check. To apply a top down approach one needs to decide on the total royalty burden T as a starting point. The evidence from which Huawei submit an inference should be drawn is evidence of statements by patent owners about what they say the aggregate royalty burden for a given standard should be.

A variety of statements about the total aggregate royalty and statements about individual companies are in evidence. The most significant for 4G/LTE are the following. Some are undated but they are most probably all from the same era 20082010:

For 4G/LTE Huawei contend that the total royalty burden T should be 8% based on the first three statements (the two from Ericsson in 2008 and the one from Huawei). For 3G Huawei rely on a further statement by Ericsson, Nokia, Siemens and NTT DoCoMo that they had, as the owners of “the clear majority” of SEPs for W-CDMA reached a “mutual understanding” to license “…at rates that are proportional to the number of essential patents owned by each company”, which would “…enable the cumulative royalty rate for W-CDMA to be at a modest single digit level”, meaning 5% or less.

Huawei point out that the April and July 2008 statements by Ericsson were regarded as so important that they were formally scheduled as encumbrances on the […] patent portfolio when it was transferred to UnwiredPlanet and also point out that Mr Robbins accepted that Ericsson’s statements about aggregate royalty were obligations UnwiredPlanet was obliged not to violate.

Huawei submit that the court should attach particular weight to early declarations by major patent owners who were predicting what their ownership would be and what the total stack should be. Huawei refer to the evidence of Prof Neven on this (paragraph

Prof Neven’s explanation is compelling as long as one is confident what the total stack should be in the first place and provided some means for enforcing it against all parties exists. However the main conceptual difficulty I have with the using a total stack in a top down approach as opposed to using it as a cross-check is in the selection of the total royalty burden T to start with.

In my judgment the statements set out above have little value in arriving at a benchmark rate today for a number of reasons. The claims are obviously self-serving. The statements about aggregate royalties in particular are statements about other people’s money on the footing that the person making the statement says at the same time that the cake is quite small but they are entitled to a large piece of it. As an illustration, if one assumes Alcatel’s 2% royalty claim means they claim at least 20% of the Relevant

Furthermore, putting weight in these statements do not take into account what implementers and SEP holders have actually been content to agree in the intervening years. Compared to public statements, comparable licences are concrete data points, albeit their interpretation can be uncertain and the factors derived from them even more so. One could use comparable licences to try and derive a figure for the total royalty burden T but to achieve that requires one to have done all the same work which is needed to apply comparables directly anyway, so back calculating T will not add anything.

Moreover the combination of Huawei’s submissions on rates and Huawei’s submissions on what UnwiredPlanet’s share of the Relevant SEPs is, shows that in truth Huawei’s case does not support an aggregate royalty burden of 8%. It supports a higher total burden than that.

Where Huawei undoubtedly have a point is that the cross-check shows that if Huawei’s case on UnwiredPlanet’s share S of SEPs overall (0.30%) is right, the benchmark rate claimed by UnwiredPlanet of 0.13% cannot be supported. It would imply a total burden T of 43%. That is far too much. Conversely if UnwiredPlanet are right about their share S of SEPs overall (1.25%), a benchmark of 0.040% implies a total burden of 3.2%. That is much less than Huawei themselves are prepared to countenance in these proceedings.

The MNPA and HPA techniques

I will now address each party’s patent counting techniques (the MNPA and HPA), explain the criticisms which are made and then address them. Rather than focus on one technique completely and then the other, the two methods need to be explained and evaluated side by side so that the assessments of each can be understood in context.

The MNPA was devised by UnwiredPlanet as a technique to use in licensing negotiations. It is applied to 4G/LTE and in the original method consisted of the following steps:

The Revised MNPA was produced in 2016 in response to points made in the litigation. It differed from the original MNPA in two major respects. At step (2) the way the standards are identified was changed in such a way as to incorporate more standards. At step (7) a different approach entirely is taken to what UnwiredPlanet call applying essentiality filters. In the Revised MNPA, instead of the three percentages at step (7), a figure derived by Dr Cooper was used based on a detailed analysis he carried out on a sample of Samsung SEPs. The figure used is 16.6%.

The numbers produced by the original MNPA are the following:

On this basis a starting list of 5915 patents is reduced to 418 Relevant SEPs for the LTE standard and for handsets and a residue of 2702 other patents relevant to handsets which were declared as essential.

The numbers produced by the Revised MNPA (in Mr Bezant’s appendices to his third report) are as follows

On this basis a starting list of 6619 patents is reduced to 355 Relevant SEPs for handsets and a residue of 2983 other patents relevant to handsets which were declared as essential. Note that the number of Relevant SEPs (355) is not exactly 16.6% of 2128. That number would be 353. The difference is explained in a footnote to Mr Bezant’s third report. I am satisfied 355 is the appropriate number to use.

Both the Original and Revised MNPA produce numbers for the industry as a whole. The way UnwiredPlanet derive figures for individual companies (apart from UnwiredPlanet itself) is by identifying the patents at step 6 by company and then applying the relevant fractions to those totals. This gives figures for individual companies.

The 80/20 approach seeks to attribute some value to the other handset patents in the residue. It does so in a mathematically simple way by attributing 80% of the royalty to a company’s Relevant SEPs in these tables and 20% of the royalty to a company’s figure for the residue.

UnwiredPlanet use the same MNPA approach to derive a total number of Relevant SEPs for infrastructure (by which they mean the air interface and eNode Bs rather than core network). The original produces a total of 3280 which UnwiredPlanet confusingly call the “Core LTE” (see step 5 of the Original MNPA table above). From this 2071 were identified as having handset claims (see step 6) which leaves 1209 families as infrastructure only (3280 = 1209 + 2071). From the 1209 UnwiredPlanet estimate most will be core network (i.e. not air interface or eNode Bs) and only 15% will be relevant infrastructure. 15% of 1209 is 181. To this 181 has to be added the share of the handset families which also includes relevant infrastructure. That is 1337 giving a total of 1518. That figure is treated in the same way as the handset figure at step 7 to produce 306 as the number of Relevant SEPs for infrastructure. As I understand it when UnwiredPlanet revised their approach to counting patents they did not revisit the numbers for infrastructure but simply reduced the infrastructure offer in the same proportion as the handset offer.

The way UnwiredPlanet deal with 2G and 3G is simpler than the MNPA technique. They start with a figure for the total pool of Relevant SEP families for 2G or 3G based on a published report. For 2G UnwiredPlanet use the report “Analysis of Patents Declared as Essential to GSM as of June 6, 2007” by Goodman and Myers of Fairfield Resources International published on 31^st December 2008. For 3G UnwiredPlanet use a similar paper published by the same group on 6^th January 2009 entitled “Review of Patents Declared Essential to WCDMA Through December, 2008”. In these papers the authors report the outcome of detailed reviews by a team of experienced engineers of the patents declared essential to wireless standards with a view to determining how many are actually essential.

The figure from the Fairfield report for the total number of truly essential 2G patent families is 158 while the Fairfield report for 3G reports the equivalent number as 529. UnwiredPlanet then subtract from these totals a number for the patent families which solely relate to infrastructure. That produces a total for handsets which is 102 for 2G and 324 for 3G.

For infrastructure UnwiredPlanet used the figures from the reports, identified patents relating to infrastructure both alone and with handsets in the same way as for the approach to infrastructure with the MNPA and came up with figures for the total Relevant SEPs for 2G and 3G. Those numbers are 85 for 2G and 274 for 3G.

The HPA was carried out by a team including Dr Kakaes and consultants at Thomson

Once these five steps were completed one could derive numbers representing UnwiredPlanet’s “deemed” essential patents identified this way. They are the basis for Huawei’s case on how many Relevant SEPs are held by UnwiredPlanet. One could also derive numbers for the industry as a whole and for other companies such as Ericsson and Huawei. They are the basis for the figures set out above. There are various different ways of deriving these figures but there is no need to get into that detail.

The totals produced by the HPA are the following:

These are the numbers presented in Huawei’s FRAND Statement of Case. They differ slightly from the numbers used in the figures set out in this judgment above for the denominators because adjustments were made during the proceedings but the changes are small and do not alter the substance.

Huawei level a sustained attack on UnwiredPlanet’s MNPA both internally (i.e.

Huawei’s external attacks on the MNPA characterise it as “patently unreliable and selfserving”. They submit that in cross-examination Mr Saru accepted that it was never designed for the purposes for which it has been pressed into service in this trial. They submit the results it produces are counterintuitive and contrary to both UnwiredPlanet’s own fact evidence and the available third party studies. They contend that the relevant experts for UnwiredPlanet, Dr Cooper and Mr Bezant, were both keen to emphasise that they had no hand in its creation and that “neither sought with any conviction to defend its results”.

In summary Huawei contended that the evidence points clearly to UnwiredPlanet having around 6% of Ericsson’s portfolio (the strength ratio). They also submitted that the MNPA was the only method which came close to giving UnwiredPlanet a 1.5% plus share of the industry’s Relevant SEPs and that all the other methods gave figures of less than 0.5%; and that the figures differ by “a long way”.

UnwiredPlanet mounted a significant attack on the HPA and its status in these proceedings. In its FRAND Statement of Case (para 132) Huawei had presented the HPA as something which was undertaken given the flaws in UnwiredPlanet’s methodology. However during the trial it emerged that this was not true, as Huawei now accept. The HPA was in fact carried out for the arbitration between Ericsson and Huawei which led to the 2016 Ericsson-Huawei licence and in which Dr Kakaes and Mr Lasinski were both witnesses. UnwiredPlanet also submitted that the HPA depends on an extremely cursory 30 minute analysis and contains an inbuilt presumption of essentiality. UnwiredPlanet ties this in to the arbitration point because, they submit, what also emerged was that in the arbitration the HPA was no more than a filter to identify patents that Dr Kakaes should look at properly. They argued that for Huawei to put the HPA forward as the actual assessment of analysis was regrettably misleading.

As an outcome, UnwiredPlanet maintain that a strength ratio of 10.50% for LTE multimode between UnwiredPlanet and Ericsson is not inconsistent with the evidence

Some of the criticisms made relate to the utility of these methods for unpacking. I will not address them because I am not satisfied that the differences between the counting techniques make enough of a difference to unpacking to be worth it.

I have no doubt that the exercise of devising the original MNPA involved a degree of self-interest on the part of UnwiredPlanet. The idea that it was devised in an entirely objective fashion is fanciful and if Mr Saru’s evidence was intended to persuade me that is was, then it did not succeed. That said I also reject the idea that the whole thing was a cynical exercise designed purely to attempt to justify UnwiredPlanet’s preordained licensing policy. UnwiredPlanet knew they needed to come up with some method of assessing the value of their patent portfolio by reference to the industry as a whole. The original MNPA was devised with that in mind but as an exercise, its utility depends on its objective characteristics which are addressed below. If it is objectively reasonable then the fact it was devised with a degree of self-service does not justify rejecting it as relevant evidence.

Huawei are correct that the MNPA was not devised to compare UnwiredPlanet’s portfolio with other companies’ individual portfolios but the fact that UnwiredPlanet now seeks to use it in this way too does not matter. What matters in that respect are detailed issues.

Huawei are right to criticise UnwiredPlanet for suggesting (or aiming to leave one with the impression) that Dr Cooper was responsible for the MNPA or parts of it as a method. Aside from his work on the sample from the HPA, he was not. Neither was Mr Bezant. On the other hand, apart from specifics dealt with later such as the 10%/20% optional/mandatory point I do not recognise the suggestion that somehow Dr Cooper or Mr Bezant thought the MNPA was so flawed that they were “keen” to emphasise they had no hand in its creation or that, overall, they did not seek to defend its results “with any conviction”.

The more significant external criticism made by Huawei is that it produces results which are counterintuitive and contrary to other, reliable, evidence. The highpoint of this is the comparison of what Huawei call “implied essentiality rates”. These rates represent the application of the MNPA to a particular company’s patents. For reasons explained below I will not use the label “implied essentiality rate”. I will call these rates the “MNPA Relevant SEP ratio” for a given company. Huawei produce a table for all the patent families in the original 6619 pool used by the Revised MNPA. […]. The figures are derived for the whole industry but it is only necessary to mention the MNPA Relevant SEP ratios for Samsung (6.72%), Qualcomm (7.41%), Huawei (3.24%) and Nokia (8.07%). Using the same approach, the MNPA Relevant SEP ratio for UnwiredPlanet is 12.00%. That is different from the ratios for UnwiredPlanet’s portfolio deployed by UnwiredPlanet in argument because the 12% comes from applying the MNPA to both the numerator and the denominator (the numbers are 3/25).

Huawei submit that this shows that the MNPA assesses UnwiredPlanet’s portfolio as being far stronger than the portfolios such as Ericsson (of which Huawei contend the UnwiredPlanet portfolio was intended to be a representative cross-section), Nokia (which Huawei point out Mr Saru explained was “careful in declaring”) and Qualcomm (which Mr Bezant said in his report had a “strong portfolio”). Huawei argue that these differences are systematic, very substantial and irreconcilable with the fact evidence or indeed any reasonable experience of the industry.

In argument for comparison with the MNPA Relevant SEP ratios for third parties

Mr Bezant’s view was that this could be explained as an artefact of UnwiredPlanet’s small portfolio size. There is something in this. One would expect that as portfolios get smaller the relative effect on the ratios of small changes in absolute numbers will increase. That is borne out by Huawei’s analysis as a whole. All the MNPA Relevant SEP ratios over 10.00% are in the smaller portfolios (below 100 in size). The same point can be made another way – if the numerator for UnwiredPlanet had ended up at 2 instead of 3 the MNPA Relevant SEP ratio would have been 8.00% […] rather than 12.00%. So I accept that one cannot place much weight on the fact that a company with a small portfolio like UnwiredPlanet has an MNPA Relevant SEP ratio which is larger than the company from whom their patents were selected.

Nevertheless this still does not mean Huawei do not have a real point. They do. These numbers expose a fallacy in the way UnwiredPlanet present the results of the MNPA both in argument and in evidence. Huawei’s name for this ratio (“essentiality rate”) was reasonable because UnwiredPlanet use similar language to describe the same thing. UnwiredPlanet have presented the number produced by the MNPA which is used as a denominator as if it represents the number of “truly essential LTE patents” or words to that effect. UnwiredPlanet’s FRAND Statement of Case calls this number the “True LTE handset pool”. These descriptions are wrong and misleading. The MNPA includes rates for the essentiality rate (28% in the original method and 16.6% in the revised) but it also has other features. The justification for the cut offs in the method apart from the essentiality rate, such as the pre-2009 cut-off, is not essentiality. It is an attempt to differentiate between the value to a licensee of two different categories of patents even though both may be truly ESSENTIAL within the meaning of the ETSI IPR policy. That is a key conceptual difference between the MNPA and the HPA. The reason different companies have different MNPA Relevant SEP ratios is not because their implied essentiality rates differ, it is because of the kinds of patents they have relative to things like the particular LTE standards to which they are declared, the priority date, and the presence of handset claims, differ. So the low rate for Huawei relative to Samsung and Ericsson is explicable by the combined effect of steps (4) and (5) whereby families with no EP/US member and the pre-2009 cut off has more impact on Huawei than Samsung or Ericsson, which in turn is consistent with the phenomenon that Chinese companies have increased their patent filings outside China only in recent years.

In my judgment the external criticisms do not undermine the utility of the revised

The first point is about step (2). The MNPA does not look at all patents declared to LTE in general, rather the MNPA takes a defined list of particular standards and deals with patents declared to those. UnwiredPlanet started with a list of 49 standards from a licensing pool called VIA known as the VIA 49. The VIA pool includes industry giants AT&T and NTTDoCoMo. Mr Saru was cross-examined about the decision to use it. While I agree the choice had an element of being self-serving, I was not persuaded it was an unreasonable choice to make. The list had been made by a third party. Later in the proceedings following criticism UnwiredPlanet used a much longer list of standards but this did not make a major difference to the end result in the light of the other filters which were used. By the closing there was less to this issue than at earlier stages in the litigation. It is not in dispute that there is no generally accepted view of what constituted a correct list of LTE standards. Part of UnwiredPlanet’s rationale for doing something along these lines was to focus on the parts of LTE which they thought licensees would be interested in. That was not unreasonable. I reject the point on step (2). The other criticisms are lesser points in any event nor do they make enough of a difference to matter.

The point on step (4) is that the patents chosen were restricted only to families containing a US or EP member. This was justified by Mr Saru on the basis that serious industry players would seek patents in Europe and the USA as major markets if they thought the patents were essential. Huawei disputed this, submitting that Dr Kakaes was obviously right not to agree in cross-examination that other markets such as China are not valuable and important. Huawei pointed out that the MNPA excluded around 709 patents for having no US or EP member.

Dr Kakaes was correct in cross-examination, all the same the evidence was clear that the licensing rates in the US and Europe are higher than elsewhere. Both the MNPA and HPA have a step like step (4) because it is a sensible thing to do. The difference is that the HPA includes a family if it has a Chinese member even if there is no US or EP. One can understand why that might be done given that Huawei is Chinese and also given the evidence that many Chinese companies will only file in China for many applications. Based on Mr Cheng’s evidence I would expect Huawei today to file SEPs internationally once the first application was made in China, given their importance. In the end I am not satisfied that this difference between the MNPA and the HPA makes any material difference to the issues I have to decide. A serious player in the telecommunications market, including a major Chinese company, would likely file SEPs in the US and/or Europe. A method which included Chinese patents when the family had no US or EP member at this stage would present a more complete picture of the landscape but the differences overall are modest. In my judgment no significant systematic error is introduced by not doing so. The nature of Huawei’s portfolio means that it will have an effect on that portfolio but I am not satisfied this matters. If the differences between unpacking methods mattered, this would be important, but they do not.

The debate about step (5) of the MNPA is important. At this step UnwiredPlanet select only patents with a pre-2009 priority date to take forward. UnwiredPlanet’s rationale is that there is an inevitable time-lag between the priority date of a patent, the invention making its way into a frozen release of a standard and then that standard being implemented. UnwiredPlanet say the fundamentals of LTE as a system were determined in LTE Release 8 and that was fixed at the end of 2008. So only patents with pre-2009 priority can be part of it. A later dated patent could not be valid and essential to this “core” system. Huawei point out that this step excludes well over 1,500 LTE families from the pool and argue it is completely unjustified.

UnwiredPlanet do have a point in that LTE Release 8 was the first and fundamental release of LTE however Huawei contend that this approach gives no value for later releases and is flawed.

The relevant releases after LTE Release 8 are Releases 9, 10 and 11. The term “LTEA” for LTE-Advanced sometimes appears. It can be taken as the same as releases after and including Release 10 of LTE. Release 10 was released in 2011 and enables downloads and uploads ten times faster than Release 8. In the UK the network operator EE began implementing LTE-A in 2013. The implementation was across the whole network. On the evidence I find that the really important aspect of the releases after release 8 is a feature called carrier aggregation. It is clearly significant, particularly relating to infrastructure and to network operators.

Mr Saru explained that the cut-off was justified in a licensing context because technology in later releases was not as critical to LTE as implemented in the products on the market at the time (by which he meant 2013/14 but the point is general, that there is a lag). In his oral evidence Mr Saru distinguished between what technology has been released in a standard and what drives the market. Huawei pointed out that Mr Saru accepted that this filter had been chosen by UnwiredPlanet knowing that it would have a relatively minimal effect on the UnwiredPlanet portfolio. UnwiredPlanet sought to mitigate this on the basis that Mr Saru’s view was that it was simply a reflection of the fact that UnwiredPlanet had deliberately selected good patents which would be strong from a licensing perspective. However I do not accept that that would justify the step even if it is really what UnwiredPlanet thought.

Mr Yang gave evidence on this for Huawei. He said that in technical terms Releases 9 to 11 involve significant developments many of which have been deployed, while in commercial terms Releases 9 to 11 are highly valuable to Huawei. UnwiredPlanet submitted Mr Yang was a fact witness but gave opinion evidence. So he did but Mr Yang was a good witness, generally qualified to discuss the topics he covered, and was simply seeking to explain his companies’ point of view.

Mr Yang supported his evidence with material extracted from three papers said to show the widespread deployment of LTE-A networks over time. (There is a point that figures

As it states, this is a graph of incremental deployments of LTE-A Release 10 networks as compared to incremental deployments of Release 8 and 9 networks. In the same paper is another graph as follows:

These two graphs are best understood in colour. To be precise the second graph distinguishes between the two LTE duplex modes: FDD – which is used e.g. in the UK

Taking Mr Yang’s evidence as a whole I find the position is as follows. In general different features are adopted by the industry at different rates. Carrier aggregation has been adopted much faster in Asia than in Europe. At a technical level features commercially deployable in later releases may well have their technical roots in Release 8 (e.g. Voice over LTE).

Dr Kakaes’ view about the pre-2009 filter was that it unreasonably excluded things which by 2013/14 were being frozen into the standards and implemented. Nevertheless he also accepted that there are features in standards which are not commercially implemented, for a range of reasons, and implementers commonly will decline to license patents relating to features they do not implement. UnwiredPlanet submitted that Dr Kakaes accepted that if it was possible to take account of these commercial realities then it was better to do so, and he accepted that the HPA did not try to do this at all. The latter submission is correct. The former submission does not precisely reflect what Dr Kakaes said in the cross-examination relied on but taking his evidence as a whole, a fair reflection of Dr Kakaes’ position was that it was reasonable to take account of the reality that there are features in standards which are not implemented.

Dr Cooper supported the existence of a time lag in terms of implementation of features and explained that features in Release 8, in contrast to later releases, were required to be used if one is implementing LTE, whereas it would be up to implementers to what extent they would implement features in a later release. However he also made clear that he “did not necessarily accept that simply post Release 8 developments should be dismissed since they can and often do add ‘value’”. That supports Huawei. There was a point about early R&D investments being higher risk and deserving higher reward but it was nebulous and I do not accept it.

There is more to these arguments than this summary but I have dealt with the major points. In my judgment LTE Release 8 does represent the fundamental technology on which LTE is based and FRAND licence negotiators would take that into account in assessing the value of patents. Later Releases of LTE are still based on the fundamentals of what is in that first working Release. Taking a cut-off of patents with a pre-2009 priority date is a FRAND approach to licensing Release 8.

On the other hand once later releases exist and are licensed, a method which gives no value at all for the technology in later releases is flawed and does not reflect FRAND. The impact of this problem changes as time goes on. It is an inherent difficulty arising from the fact that standards develop over time.

For LTE, assessed as at 2014, I find that the absence of value for post-2009 patents is not significant (either in Europe or anywhere else). However assessed today (2016/2017) the absence is significant given the way LTE-A has been implemented over time. For LTE some value has to be given in assessing the FRAND value of a portfolio for patents essential to later releases (and which therefore may have been excluded by a pre-2009 cut off). On the other hand, a method which gives equal value to any patent essential to anything in Releases 9 to 11 will inevitably overstate that value. Release 8 is still the fundamental technology in LTE and while carrier aggregation is important in the later releases, other aspects are not.

There are a limited number of ways in which one can deal with this. Unless one is going to make a list of Releases 9 to 11 features and identify each patent relevant to that feature, which would be impractical, the only alternatives are broad brush. One can include all patents knowing that this overstates the value of post-2009 patents, which is the HPA method, or exclude them all knowing this understates the same value, which is the MNPA. The 80:20 approach by UnwiredPlanet is an attempt to mitigate this problem, among others, because it gives some value for patents put to one side by the pre-2009 cut-off. In that sense the intention behind the 80:20 approach is sensible but I am concerned that it is so crude as to be arbitrary.

The points on step (6), the handset and infrastructure filter, were mostly concerned with unpacking and Huawei’s portfolio. There is no need to engage with that. UnwiredPlanet’s detailed approach to handsets and infrastructure involves an assumption that 15% of the families with no handset claim are RAN rather than core network. Having listened to Dr Kakaes’ oral evidence on this I find that UnwiredPlanet’s approach was reasonable. The MNPA figure for handsets will be a lower bound but I doubt it is all that far from the true figure.

The next step is step (7), the essentiality filter. Here the Revised and the Original MNPAs differ. In the Original MNPA three fractions are used. The first is the 28% essentiality ratio. This was used to deal with over-declaration. There is no question that over-declaration is a major problem. The question is - how big? 28% was derived from studies reported in papers by Fairfield/Goodman & Myers on 2G and 3G. Dr Kakaes criticised this because there were also similar studies from the same group (Fairfield) and other groups for 4G which gave higher essentiality ratios. Mr Saru thought 28% was likely to be a ceiling and a lower percentage might be more accurate. Two 4G papers were put to Mr Saru (from Cyber Creative and FRI) in which essentiality rates of 50% or more are given.

In an annex to Huawei’s opening skeleton is a summary of third party essentiality studies on 4G. There are four: iRunway at 8.2%, Fairfield at 50% and two Cyber Creative studies at 53.8% and 56.0%. Huawei submitted the iRunway study was not representative.

However as UnwiredPlanet point out, Huawei’s own HPA produces a lower overall essentiality ratio than the 50%+ rates from Cyber Creative and Fairfield. Dr Kakaes reported overall essentiality ratios for 4G of 35.8% and 34.1% from the HPA on slightly different bases (the differences do not matter). In his third report Dr Kakaes set out a table for sixteen individual companies’ 4G essentiality ratios derived from the HPA. They range from 18.6% for Google’s patents (338 declared, 63 deemed essential by the HPA) to 82.3% for Sharp’s patents (79 declared, 65 deemed essential by the HPA). Most of the companies in the table (13) have ratios within 22%-50%. The portfolios range from 64 to 771 declared and 14 to 228 deemed. The ratio for Huawei is 43.5% and the ratio for Samsung is 23.5%.

In cross-examination Dr Kakaes did not accept it was reasonable to use a 2G/3G study for the purposes of considering a 4G rate although he accepted that there was no technical reason why essentiality ratios for 2G, 3G and 4G should differ and indeed had made that point in his third report. Unusually for Dr Kakaes, who was a good witness, on this particular point he had lost some objectivity.

Dr Cooper addressed the papers which reported essentiality ratios. He thought they all had problems and weaknesses and he thought the 4G Fairfield paper produced a ratio which was too high. In his opinion there was no reason to think the essentiality ratio for 4G was likely to be different from 2G and 3G. He thought 28% for 4G was generous and this was a view formed before he had conducted a review of Huawei’s and Samsung’s portfolios. He thought the true essentiality ratio was likely to be between 10% and 20%.

In my judgment adopting 28% as an overall essentiality ratio for 4G cannot be criticised. It was reasonable both when UnwiredPlanet adopted it in 2013/14 and later on.

A further point which relates to this but is convenient to address now is Dr Kakaes’ opinion that using a different method to assess the numerator and the denominator in the strength ratios is not appropriate. It will be recalled that UnwiredPlanet do this whereas Huawei do not. Although superficially it might appear to be a sound criticism, in my judgment it is not a valid point in these proceedings. Of course in general one usually seeks to compare like with like. Therefore it is meaningful to present a ratio for UnwiredPlanet against another company or the pool as a whole based entirely on figures provided by the same technique – as in the HPA. This is particularly so when the technique does involve some consideration of each patent rather than figures applied across the board as in the MNPA. However it is also meaningful when one wants to make a comparison between an identified collection of patents and the pool overall to do what UnwiredPlanet did and analyse the identified collection individually while applying a broader brush technique like the MNPA to the wider pool, since after all it is entirely impractical to analyse the whole pool with that same rigour. Moreover this is all the more legitimate when the identified portfolio is small, since an average is less likely to be accurate when applied to a small pool than a large one.

The other two percentages used by UnwiredPlanet (10% optional and 20% mandatory non-deployed) were criticised as arbitrary. The problem is not with the concept that there are optional features to which SEPs are essential and mandatory features in a standard which are not deployed. I have no doubt both exist (as Dr Cooper explained), albeit it is also true that some strictly optional parts of LTE are really important (e.g. MIMO as Dr Kakaes explained). The problem is justifying deductions of this magnitude. Combined together these two fractions reduce the denominator by over a quarter and therefore correspondingly could increase UnwiredPlanet’s royalty rate by a third (1/100 = 1%, 1/75 = 1.33%). And in my judgment the problem is made much worse given the pre-2009 cut off and a limit on the number of standards considered at step (2). Dr Kakaes was right that there was no reliable empirical basis for either fraction. Mr Saru’s attempt to justify them in his oral evidence referred to his experience in a vague way and to informal calls to old friends. That was hopeless.

In my judgment the 10% and 20% fractions cannot be supported alongside the pre-2009 cut-off at step (5) and the limit on standards at step (2). Combining the fractions with steps (2) and (5) is not FRAND.

The Revised MNPA was devised with the criticisms of the original MNPA in mind. At step (7) the Revised MNPA uses a single fraction of 16.6% derived by Dr Cooper. It arose as follows. Dr Cooper was asked to review the findings of a sample of the patents which the HPA deemed to be essential to an LTE handset that had a pre-2009 priority date. Dr Cooper randomly selected a sample of patents of a size that would allow him to draw conclusions with at least 90% confidence about the pool from which the sample was drawn. This resulted in Dr Cooper reviewing 38 Samsung and 30 Huawei patents and he spent 5-6 hours per patent family. He concluded that the essentiality rate of the Samsung patents (excluding optional features) was at most 16.6% and then revised that further to 15.9%. For the Huawei patents he concluded that the essentiality rate (excluding optional features) was at most 9.4%. UnwiredPlanet used that 16.6% figure at step (7) of the revised MNPA.

UnwiredPlanet point out that in his second statement Dr Kakaes was not surprised that having spent 5-6 hours per patent family, Dr Cooper had found a number of patents not essential which the HPA had deemed to be essential. They point out that Dr Kakaes went on to agree with Dr Cooper about a substantial number of the patents in his study. The major criticism made by Dr Kakaes was about the sampling process. I will deal with that after the other points.

The detailed points were these. First, there were patents excluded based on Dr Cooper’s definition of LTE. However I am satisfied that at best this would make little difference to the end result. At best the point changes the result for two patents. The impact of that can be seen from the fact that changing the result for one patent moves the answer from

A question was whether it was right to use a rate for Samsung as in effect an industry average. Huawei submitted there was no empirical evidence that a rate for Samsung was representative of the industry as a whole. The choice was UnwiredPlanet’s rather than Dr Cooper’s. He explained that he would not expect the rates for different companies to be identical but he could not see an a priori reason why there should be big variations between companies. In my judgment the evidence, as best it is, is that the rates for different companies can differ considerably (see the table above from the HPA) but there is no systematic reason why one company’s rate should be different from another. In my judgment using a rate for Samsung as representative of the industry is not illegitimate given that Samsung is a major player. I doubt Samsung has an essentiality rate which is significantly below average. There are significant uncertainties in all these exercises and this is another but it does not render the technique meaningless. Choosing to use the number derived by Dr Cooper for Samsung rather than Huawei was the conservative choice.

Huawei also submitted that it was inappropriate to use a figure derived from the HPA in the MNPA. This was for three reasons. First because the filters in the MNPA produce a different starting pool of patents from that in the HPA. Second because the different approach to LTE means many Samsung families found essential in the HPA would not have made it through the filters in the MNPA but were then chalked up as inessential. Third because the way of identifying a family as a handset family differs. In substance these are either another way of putting the detailed points I have already considered or they relate to the major sampling issue which comes next.

For Dr Kakaes the key problem with Dr Cooper’s approach was that while a random sample had been taken from the pool which was sampled, the pool which was sampled was skewed. This was in two respects: first the pool from which the sample was taken consisted of the patents deemed essential (and held by Huawei or Samsung); and second the pool was actually only a subset of that because it was also limited to patents which met certain MNPA filters such as the pre-2009 cut-off. I have already dealt with the second point above but that does not address the first point. As to this, Dr Kakaes acknowledged that there will inevitably be errors in the evaluation process in the HPA but he said those errors would point in both directions, i.e. essential patents could be deemed not essential as well as not essential patents deemed essential. Therefore sampling only from the pool deemed to be essential was skewed. To do something like this sort of sampling appropriately, a random sample should have been selected from the pool as a whole, before evaluation. For Dr Kakaes this undermined the exercise entirely. Dr Cooper had sampled from what Dr Kakaes called a very, very, very biased universe.

The strength of Dr Kakaes’ view about this point came across in his oral evidence.

Without resolving this argument about the HPA I cannot complete my consideration of the Revised MNPA. In order to decide issues as much as possible in their proper context I will therefore suspend consideration of the Revised MNPA, this being the last major issue, and turn to the HPA. Once I have dealt with the HPA I will consider the implications of those decisions on this aspect of the Revised MNPA and reach my conclusions on both the Original and Revised MNPA.

The HPA was run as an exercise as part of the Ericsson – Huawei arbitration. That is not in dispute. The way the HPA was presented by Huawei in these proceedings in the FRAND Statement of Case was wrong and should not have happened. Huawei cannot simply blame Ericsson for demanding secrecy about the arbitration. Huawei have an excellent UK legal team and the matter could have been raised with the court (or even conceivably with another judge although I cannot see that that would have been necessary). What has happened is that the truth about how the HPA was devised and the reasons for it were not presented properly from the outset. Although more came out at trial I am not satisfied the full picture has been presented to the court.

Huawei maintained in closing that:

“The exercise was overseen by Dr Kakaes and a team of engineers from Thomson Reuters. The evaluators were not informed of the identity of the ultimate client (i.e. Huawei) or of the opposing party in the dispute for which the analysis was originally prepared (i.e. Ericsson), so as to preserve neutrality.”

I accept that the engineers who undertook the evaluation exercise at step (5) of the HPA did not know that the client was Huawei when they carried out their work. That is a virtue and in that sense the exercise was neutral. However it was clear from Dr Kakaes’ answers in his oral evidence that Huawei’s US lawyers, Sidley Austin LLP, were involved at various stages in the decision making and possibly drafting of documents and I am not satisfied that the HPA can be regarded as something set up independently of Huawei’s interests. It is not possible to say. However just like the MNPA, what matters most is the objective reasonableness of the steps, not the motives of the devisers.

UnwiredPlanet take a number of points about the HPA but in my judgment none of them matter except one, which is the submission that the evaluation step (5) was in fact no more than a coarse filter to identify patents that Dr Kakaes should look at properly later and has a tendency built into it in favour of increasing the number of patents in the pool deemed to be essential. The other points taken by UnwiredPlanet (about the initial dataset and technology categories) are similar to the points I have rejected which Huawei took against the MNPA. If the coarse filter point succeeds UnwiredPlanet do not need to place further emphasis on the other issues and if it fails, they are not significant enough to undermine the HPA outright. Just as they do for the MNPA, the extra points serve to emphasise the inherent uncertainties in the exercise.

The evaluation exercise which was carried out was a huge undertaking. Even then the average time per family was ½ hr. In a much smaller exercise on a small subset of patents which Dr Cooper conducted he spent 5-6 hours per family. He was not wasting time. UnwiredPlanet detected that in cross-examination Dr Kakaes tried to resile from the onerous nature of the task. I do not believe that is what he was doing. He was simply emphasising that in parts some of the elements of the task may not be that difficult. In his reports Dr Kakaes had emphasised that the analysis was not a rigorous and thorough assessment of essentiality of all declared SEP families in the relevant group, since carrying that out was not plausible without employing vast resources. The exercise was based on what he called a “relatively quick assessment”. In my judgment, given the number of families to deal with, and the inherent complexities of the patents, standards and the task itself, it would not be possible to make a definitive assessment of essentiality for the number of patents in issue in the time available. I do not believe Dr Kakaes suggested otherwise.

In his written fact evidence Dr Kakaes addressed the Essentiality Review Protocol. In paragraph 31 of his first witness statement Dr Kakaes described the evaluation as follows:

“31.

Accordingly, the second stage of the study was to analyse the 11,384 Group 1 patent families to seek to determine whether or not a patent that was declared essential to ETSI is, in fact, “essential”. Conclusively confirming actual essentiality is a complicated and involved legal and technical task. In this document, I use the term “is essential” (and similar terms) to mean that, after evaluation, we have determined that there is a reasonable basis for treating a patent as essential. In each such instance, we reviewed the patent specification and claims and did not identify an apparent reason to exclude the patent from being essential. Thus, a more precise interpretation of this phrase is that such a patent has passed a screen to exclude non-essential patents.”

UnwiredPlanet say this shows that patents were deemed essential as long as there was a reasonable basis to treat it as such and only excluded if an apparent reason to exclude it had not been identified. The method was in Dr Kakaes’ words a screen to exclude non-essential patents. On its face this description accords with UnwiredPlanet’s submissions.

In paragraph 41 of the same statement Dr Kakaes explained that if the standard being considered required all the elements of one of the claims being considered, then the patent family was deemed essential. Expressed that way there is no tendency either way but in a footnote to this paragraph Dr Kakaes then said: “To be more precise, the reviewers determined that the declared standard specification(s) did not provide a clear reason to rule out the patent as being essential.” UnwiredPlanet say this reflects the same tendency they contend can be seen in paragraph 31.

UnwiredPlanet also pointed to the protocol document exhibited by Dr Kakaes which provided that the evaluators in which the word “substantially” appeared in a context which expanded the scope of what would pass as essential. The text is:

“Compare the selected claims with the declared standard specifications and determine whether the standard specifications substantially require all the elements of the claim.”

(my emphasis)

These points were all put to Dr Kakaes in cross-examination. He did not accept UnwiredPlanet’s characterisation of the effect of these passages. One suggestion he made in cross-examination and repeated in re-examination was that the explanation related to a detail of ETSI IPR Policy which is involved in considering whether it is not possible on technical grounds to do otherwise. I accept Dr Kakaes’ evidence that this detail of the IPR Policy, which would involve proving a negative, did not form part of the assessment of essentiality but I do not accept that it is what passages in the written evidence were talking about.

I accept Dr Kakaes’ testimony that he checked numerous entries and found errors going both ways, including patents the evaluators should have placed in the deemed essential collection but had not done so, perhaps because they read the claims too narrowly or missed additional standards. I also accept that he spent hundreds of hours checking results and answering queries from the evaluators. This supports Huawei’s submission that the aim of the HPA was to apply a consistent approach to all the patents considered. I am sure a consistent approach was applied. The debate is to properly characterise what the approach was.

UnwiredPlanet also put to Dr Kakaes something he said about Dr Cooper’s detailed analysis of the sample deemed essential by the HPA. Dr Kakaes had said he was not surprised that Dr Cooper’s more detailed studies had found that a number of patents deemed essential in the HPA were not in fact essential. That lack of surprise supports UnwiredPlanet’s point but when asked about it Dr Kakaes said he just meant that he was not surprised Dr Cooper had reached different views. I do not accept that explanation. The point was not simply that Dr Cooper had reached different views, the point was that for patents deemed essential, Dr Cooper had found quite a number of them not to be.

Some of Dr Kakaes’ answers on this topic were rather difficult to follow. I think a partial explanation for this was that in all his evidence about the HPA as a methodology, particularly before the point at which the arbitration point emerged fully but also afterwards, in the back of his mind Dr Kakaes was worried about the arbitration and about what he thought he was and was not allowed to say. This did not help but even taking that into account I am left with one characterisation of the HPA in Dr Kakaes’ written evidence and a different one in his cross-examination and re-examination.

Weighing up the evidence I prefer to place weight on Dr Kakaes’ written evidence. It was clearly written taking care to present a balanced explanation of the exercise and its limitations. It is also inherently credible that an exercise of this scale, which could only ever be a “relatively quick assessment”, would err on the side of placing a patent family in the deemed essential collection unless there was a sufficient basis not to. There is nothing wrong with that provided it is understood that that is what is happening. It is a sensible way of proceeding. Dr Kakaes felt a personal ownership of the HPA and I think in the cross-examination he regarded the questions on this topic as implying that the HPA was flawed. Therefore he sought to defend it.

I find that it is accurate to describe the evaluation step in the HPA as a step which errs on the side of including a patent in the deemed essential pool.

I turn to consider the significance of UnwiredPlanet’s case about the HPA’s role in the arbitration. The submission is that in truth the HPA was designed to be just a coarse filter to identify patents that Dr Kakaes should then look at properly, or in other words a screen to exclude clearly non-essential patents. If that is right then it enhances UnwiredPlanet’s case on the nature of the evaluation step.

In cross-examination it was put to Dr Kakaes that Huawei needed to make the process of assessing essentiality manageable by curtailing how long a single family was to be considered. Dr Kakaes did not agree and wanted to explain why but to give a complete answer required him to explain how the results of the HPA were used. At this point it emerged that he felt unable to give a full answer because of non-disclosure obligations he felt he owed Ericsson as a result of the arbitration. Once in private and reassured that he could speak freely Dr Kakaes explained that the HPA, which he called the “study” was “just to figure out what the landscape is”. The study had two steps, the census (i.e. steps (1) to (4) as described in this judgment) and the essentiality study (step (5)). The information was passed on to Mr Lasinski but Dr Kakaes said his (Mr Lasinski’s) usage of it was minimal. What Dr Kakaes also did was analyse a subset of patents that were deemed essential in the HPA. They were patents held by the parties to the arbitration - Ericsson and Huawei. He said “I looked in detail, and in the subset of Ericsson essential patents, and identified their importance and so on.” and added “a

UnwiredPlanet say this proves their point. Huawei do not agree. In closing counsel for Huawei placed emphasis on the word “importance” in these passages and submitted that the further detailed study which Dr Kakaes was talking about was not a study of essentiality, it was a study of “importance”. That is a term he had used elsewhere in his report as relating to the value of an invention, i.e. the importance to the standard of the technology covered by the patent. In other words it is accepted (plainly rightly) that Dr Kakaes here was explaining that there was further detailed study and analysis of patents placed into the deemed essential pool by the HPA, but Huawei argues that the nature of that further study was about importance and so does not support the idea that the HPA was a coarse filter on essentiality on the footing that patents could always be weeded out later on more careful consideration of that aspect.

I readily accept that although it was not apparent at the time he was speaking, Dr Kakaes was using the word “importance” there in the same sense as elsewhere in his report. But I am not persuaded that this takes one as far as Huawei seek to go. First of all, if it matters, Dr Kakaes did not say in these passages that “importance” was the only thing considered in the further study. Secondly, “importance” is concerned with evaluating the importance of the patent’s technology to the standard and therefore cannot help but traverse the same ground as essentiality. The idea of undertaking a further study of importance without noticing whether a patent is essential is unreal.

The HPA was devised for and used in the arbitration and regrettably the court has not been presented with a full picture of the HPA. I find that what we call the HPA was devised not simply as a scheme to produce an end result in itself, but as a form of filter or screen to produce a pool for further study. That is consistent with all of what I know now. On that basis there is nothing surprising about the idea that the evaluation would err on the side of essentiality since there was going to be a further detailed study which involved considering the patented technology and the standard. Given that, there was no harm in including more patents in the deemed pool than would turn out to be essential on detailed study. What one would seek to minimise was missing patents from the deemed pool which might be essential. UnwiredPlanet’s characterisation of the nature of the evaluation step in the HPA is correct.

The task the HPA performs is an inherently difficult one. The answers can only ever be approximate. In the HPA the essentiality evaluation step is and was intended to be a coarse filter to screen out non-essential patents and to err on the side of including a patent in the deemed essential pool. This does not mean the method is flawed or unreliable. I am satisfied that the HPA has applied a consistent yardstick and produces meaningful results. It is a reasonable attempt to deal with over-declaration and derive information about how many essential patents there really are. When comparing large numbers on a like with like basis, the tendency built into the evaluation step matters much less. However as an absolute value, the numbers from the HPA over-estimate the true number of essential patents. In other words, if a number derived from the HPA is used as the denominator in a fraction in which the numerator is a number derived by considering the patents in more detail, the result will understate the significance of UnwiredPlanet’s patents. Furthermore for smaller pools the coarse nature of the filter is likely to matter more and produce a greater uncertainty in the numbers.

Huawei derive the HWLTER of […] using the HPA on its own and so they can fairly submit it is the result of applying the HPA consistently. However that number is based on a numerator which gives the same number of Relevant SEPs in UnwiredPlanet’s portfolio as Dr Cooper’s more careful analysis. I find that Dr Cooper’s analysis is likely to be closer to the true figure. Compared with this, raw figures from the HPA tend to be overestimates. The impact of that will apply to the denominator. I find the true strength ratio R should be somewhat higher than [the HWLTER].

I can now return to the Revised MNPA. To recap the point is that Dr Cooper performed a more detailed study of a sample of patents belonging to Huawei and Samsung which were in the deemed essential pool of the HPA. Huawei submitted I should place no weight on Dr Cooper’s assessment because by sampling only from the deemed essential pool, the exercise was badly skewed. UnwiredPlanet’s answer was that the nature of the evaluation step in the HPA meant it was reasonable to focus on patents which passed the filter and assume that those patents which were discarded as not passing the filter would not have been found essential by Dr Cooper. I have accepted that this step of the HPA does increase the pool of essential patents and errs on the side of putting a patent in the deemed pool. Accordingly, one would not expect there to be as many patents in the discard pool which would in fact turn out to be essential after a detailed 5-6 hour analysis, as there would be patents in the deemed essential pool which turn out not to be essential. There will be errors going both ways, as the evidence established, but the inherent tendency built into the evaluation exercise means that it is reasonable to expect many fewer patents in the discard pool as having been wrongly rejected, than there are patents in the deemed essential pool which turn out not to be essential. So while selecting only from the deemed essential pool will inevitably skew the result a bit, I am not satisfied that the skewing will be anything other than small. It is a point to keep in mind when placing weight on the result but it is not strong enough to justify rejecting the approach.

Obviously more effort would lead to more statistical rigour, but the effort of evaluating the number of patents Dr Cooper’s exercise did with 5-6 hours per patent family is already considerable. Even within the limits of the enormous sums spent in costs by

In my judgment Dr Cooper’s study was a reasonable effort to assess the essentiality rates of Samsung and Huawei.

Having now been through all the points in detail I will stand back and consider the MNPA as a whole. Broadly the HPA and MNPA are aimed at the same difficult task. The MNPA has flaws but, apart from one aspect of the Original MNPA, overall in my judgment the Original MNPA was and the Revised MNPA is a reasonable attempt to derive information which allows one to assess the strength of a portfolio of patents declared essential to LTE as against the industry as a whole, from the point of view of what licensees would be interested in. There are two critical caveats.

First, as with the HPA, one needs to take care with the results because the error bars are wide. However the results of the MNPA are not meaningless and do not systematically favour UnwiredPlanet, as long as one does not think the results are the true essentiality rates. The MNPA has a tendency to understate the value of patents in China because of step (2) but for a global benchmark the MNPA has utility.

Second, with the MNPA, something like the 80:20 approach is necessary. UnwiredPlanet’s description of the final number as the “True LTE handset pool” is wrong. To use the Revised MNPA fairly demands the incorporation of some step which gives some value for the patents which fall outside the so called “True LTE handset pool”. That is a serious weakness.

Huawei pointed out correctly that when the 80:20 approach was applied in the Original MNPA it was applied differently, not to calculate a number representing UnwiredPlanet’s patent share but rather to apply to the imputed royalty stack. That is true but this way of putting UnwiredPlanet’s case was advanced at the trial and it is right to consider it.

Whether another ratio apart from 80:20 is a better reflection of the different value of patents in the two pools is not something addressed in the evidence. A majority of the residue patents will not be essential at all but a good number will be essential to options and later developments of significance to LTE (e.g. carrier aggregation, TDD and later MIMO patents). In terms of individual patents, given the different sizes of the two LTE pools using UnwiredPlanet’s figures, 80:20 makes an individual patent in the Relevant SEPs pool about 34 times more valuable than residue. I think that is much too high. That may be because the pool of Relevant SEPs is too small relative to the residue pool or because the 80:20 ratio is too generous to UnwiredPlanet or some combination of the two.

The focus of the debate on the MNPA has been on 4G handsets but the weaknesses exposed in it also apply to the numbers UnwiredPlanet contend for in relation to infrastructure on 4G. The points made do not apply to the same extent to UnwiredPlanet’s case on 2G and 3G.

Findings about the strength of UnwiredPlanet’s portfolio

The strength of UnwiredPlanet’s portfolio for multimode 4G handset licensing is represented by two numbers: S (the share of the total Relevant SEPs) and R (the ratio of UnwiredPlanet to Ericsson). For 4G multimode handsets UnwiredPlanet’s number for S is 1.25% and for R is […] (the UPLTER). These are based on the MNPA and

A further aspect to keep in mind is that these numbers are supposed to reflect various ratios of numbers of patents in different categories to one another and they are linked in complicated ways. A simple illustration that the differences between the parties are not simply in the magnitudes of S and R is that UnwiredPlanet’s R is about 8 times bigger than its S whereas Huawei’s R is about 20 times bigger than its S. I do not mean to say that that relationship means anything in particular, the point is a reflection of underlying differences.

I am satisfied that both methods produce the wrong answer. The problem is whether there is a better way to arrive at the right answer than doing my best to choose values for S and R somewhere between the parties’ extremes.

I thought initially that a virtue of what one might call the Revised MNPA + 80:20 approach, based as it is on percentages such as the 16.6% at step (7), would be that the method itself was more readily adjustable than the HPA. For instance one could for example decide as a matter of judgment that 16.6% ought to be 28% (the figure used in the Original MNPA). However the complexity of the 80:20 adjustment, layered on top of the multimode adjustment and, if one is considering R, also taking into account figures for Ericsson too, makes the Revised MNPA + 80:20 approach impossible to adjust in a credible manner. The only way it can be adjusted would be so broad brush that it would be mere pretence to suggest it was more meaningful than doing my best to just choose values for S and R somewhere between the parties’ extremes.

The problem posed by the HPA is different. At its heart is the evaluation of the team of evaluators which is not adjustable at all. Nevertheless the way in which the key numbers are produced using the HPA as a method is simpler and more transparent than the Revised MNPA + 80:20 approach. I have concluded that the right way to reach a conclusion is to apply adjustments to the figures derived from the HPA. The basis for the adjustments is my qualitative evaluation of the evidence as a whole, primarily Dr Kakaes and Dr Cooper, and including the indications given by the Revised MNPA + 80:20 approach. Since this is the approach I will take to 4G, I will take the same approach to 2G and 3G.

The significant overstatement in the HPA is the number produced for the total pool of Relevant SEPs. The number for 4G handsets is 1812 and is much too high. The corresponding number in the Revised MNPA is 355 but that number is much too low if it is to represent all Relevant SEPs. I think both values are out by about a factor of two. Half of 1812 is 906 while twice 355 is 710. Splitting the difference takes one to 800. Standing back, about 800 is fair and in my judgment an appropriate figure for the pool of 4G/LTE patents. Applying that as the denominator in a fraction to determine the share S which UnwiredPlanet’s patents represent from the pool gives 6/800 = 0.75%.

This all produces the following tables:

Turning to the ratio R between UnwiredPlanet and Ericsson and taking the numerators as a given, the critical numbers are the numbers of relevant Ericsson patents. For this exercise I will not try to distinguish between handsets and infrastructure but just use

All of these numbers are close enough to [Z%] so as not to be out of line with the number of patents transferred to UnwiredPlanet from Ericsson’s portfolio. The small sample sizes involved mean that reasonable deviations from [Z%] are unsurprising.

So for 4G multimode handsets I have concluded that UnwiredPlanet’s share S of the total pool is 0.70% while UnwiredPlanet’s ratio to Ericsson R is 7.69%. In principle these numbers ought to be linked by Ericsson’s share of Relevant SEPs but the uncertainties mean that perfect consistency is not realistic and I will not strive to find it.

The comparables in this case

Having considered how the UnwiredPlanet patents stand as compared to the industry and to Ericsson, the next step is to evaluate the various comparable licences in evidence. The UnwiredPlanet licences may also allow me to arrive at a rate directly. The bulk of the licences are Ericsson licences and the ultimate objective with those is to arrive at a figure for the value E in order to do the sum E x R.

2014 UnwiredPlanet - Lenovo

The 2014 UnwiredPlanet-Lenovo was introduced in the section on UnwiredPlanet’s case on rates above. The major debate about this licence is whether any weight should be placed on the running royalty rates on the face of the licence. Mr Lasinski said they were cosmetic. The rates are expressed in cents per product but making sensible assumptions they compare favourably to a royalty rate of 0.2%. The point is that the licence contains two lump sums adding up to $100 million. On the face of the agreement […] is defined as a prepayment of royalty while the […] balance is attributed to the sale to Lenovo by UnwiredPlanet of certain patents. On its own terms therefore no further running royalties will be due until the […] is exhausted. Huawei contend that Lenovo wanted to attribute the whole $100 million to royalty pre-payment but accepted the […] split because they thought they were safe that […] would not be exhausted during the term (5 years plus an additional possible 2 years). Other factors which bear on this are these: […] Also there is a dispute about the attribution of the […] licence element.

A factor which does not have much significance is the difference between the higher MM rate […] and the lower […] OT rate applied elsewhere […]. Within the limits of the uncertainties in this exercise these two come to a similar percentage rate when the difference in product prices in these two markets is taken into account.

The oral evidence about the Lenovo licence was given by Mr Robbins, who had been personally involved in the negotiations on the UnwiredPlanet side. There are also some UnwiredPlanet documents. The import of the evidence is fairly clear. I find that both Lenovo and UnwiredPlanet thought it was highly unlikely that the prepayment would be exhausted. (Mr Robbins in cross-examination said “certainly unlikely”.) […]. Nevertheless it does not follow from this that Lenovo did not care what the running rate was. That is for three main reasons. First, it was not inconceivable that Lenovo’s sales would be large enough over the term to exhaust the royalty. It was possible that they would be that large if Lenovo enjoyed a very high rate of growth, comparable to that of Samsung. The running royalty will determine the rate at which the lump sum is used up. […] For these reasons I find that the running rate was the product of genuine negotiation. The […] and, of course, the lump sum pre-payment itself, meant that Lenovo’s interests were protected to a high degree in any case but I find that Lenovo still had an interest in negotiating a lower rate. Their interest was modest compared to the negotiation of the lump sums and other terms, but it was tangible.

The allocation of the lump sums between the patent purchase and the licence as it appears on the face of the documents is not reliable. Mr Bezant and Mr Lasinski were agreed about that. As I understand the case presented by each side, neither party seeks to unpack a lump sum notionally attributable to the licence in order to generate a comparable royalty rate nor does either party seek to use a sum attributed to the value of the patents sold to generate useful evidence for the value of UnwiredPlanet’s patents. Therefore it is not necessary to reach a view about what the proper attribution would be. If I had to do so I would find the large majority of the value should be attributed to the patent purchase.

Focussing on the licence itself, it is a licence for SEPs and implementation patents but as drafted there is no information to allow one to make an attribution between these two.

In his Sixth report Mr Bezant plotted what the comparable Lenovo rate might look like if one assumes […]. To do this also involved taking Huawei’s sales profile as a royalty base. This is not a reliable comparable at all. I will not place weight on it.

I conclude that the Lenovo licence is not a useful comparable from the point of view of setting a FRAND rate today given the other evidence now available. However its utility depends on the other evidence available and so, from the point of view of UnwiredPlanet in 2014, who were not privy to the terms of any licences to which they were not a party, it may bear more weight. I will address that in context if necessary.

UnwiredPlanet-Samsung 2016

As part of a settlement of these proceedings and after PanOptis acquired UnwiredPlanet, the UnwiredPlanet-Samsung licence was entered into on 28^th July 2016. Huawei contend it is in principle the best comparable in the case while UnwiredPlanet contend it is a poor comparable.

As Huawei put it, they rely on the licence for both the FR and ND elements of FRAND. At this stage I will focus on the weight and significance to be attached to this licence as evidence of the fair and reasonable value of the UnwiredPlanet portfolio in 2016. Hard edged non-discrimination will be dealt with later.

Under the licence Samsung paid UnwiredPlanet US […] and assigned a portfolio of 20 patent families in return for a worldwide licence under UnwiredPlanet’s SEP and nonSEP portfolio until […] together with a release of any past damages. Before one decides how much weight to place on any royalty rate information derived from the licence, UnwiredPlanet contend that this licence cannot be seen in isolation and needs to be considered in the context of a wider arrangement between PanOptis and Samsung and the distressed financial position UnwiredPlanet was in when acquired by PanOptis. This depends on Mr Ware’s evidence. Huawei’s case is that the facts of what went on are now sufficiently clear to show that the wider factors make no material difference. UnwiredPlanet disagree and contend that the two issues of rate and context interact directly because any royalty rate derived from this licence is truly much lower than the rates which Huawei put at the forefront of their argument on this licence and that this is a reflection of context.

So in order to derive a royalty rate from this one needs […], ascribe a value to the Samsung patents assigned to UnwiredPlanet, take into account the value of the nonSEPs and work out a way of assigning value as between 2G, 3G and 4G. The way Mr Lasinski assigned value between 2G/3G and 4G is not in dispute.

Mr Lasinski derived a range of possible rates and presented them in two tables, one for […] and the other assuming […]. The provisions […] in the licence are complicated but do not need to be explained. Each table then shows the implied royalty rate depending on the value attributed to the assigned patents – from […], and the percentage of royalty attributable to SEPs rather than non-SEPs from 25% to 100%. As the value of the assigned patents rises the royalty goes up because in effect Samsung have given more value for the licence. Also as the percentage rises the rate rises too, because it is a rate for the SEPs rather than the non-SEPs. The 4G rates range from […]. The 2G/3G rates vary accordingly from […] on the same basis.

Mr Lasinski made a point relating to the similarity of rates implied by the [K] licence and the [J] licence. Similarly in figure 4 of his third report which is set out above Mr Lasinski plotted rates derived from the UnwiredPlanet-Samsung 2016 licence on the chart together with rates from those two Ericsson licences and two other data points. UnwiredPlanet criticised this and submitted its effect was to make the rates derived from the […] licence look closer to the other three Ericsson licences than they really are. That is because the rates used were the highest rates Mr Lasinski derived from the UnwiredPlanet-Samsung 2016 licence and while the difference is still a factor of […], if more realistic rates were used for the UnwiredPlanet-Samsung 2016 licence it would be shown to be an outlier. Mr Lasinski did not accept that was why he had chosen to plot those rates in Figure 4 and since the idea that Mr Lasinski was setting out to mislead was not put squarely to him in this context, it would not be fair to him to make a finding on that.

However, objectively speaking, by including only the highest rates from Mr Lasinski’s tables for the UnwiredPlanet-Samsung 2016, Figure 4 is capable of misleading. The assumptions on which the highest rates are based are […]. However, Mr Lasinski accepted […] and accepted he had used a much lower SEP percentage (about 30%) when performing a similar calculation on the Lenovo licence. His explanation that this was because he did not regard Lenovo as a good comparable does not justify this difference. On the assigned value Mr Lasinski took Mr Ware’s acceptance of […] despite having earlier expressed the view […] and despite generally not accepting Mr

There are major uncertainties deriving implied rates from this licence but the figures used in Mr Lasinski’s figure 4 are too high. […]. On these assumptions the 4G rate ranges from […] and the 2G/3G rate ranges from […]. These are all far lower than the other rates in Figure 4 and relied on by Huawei as best comparables. It supports UnwiredPlanet’s case that the UnwiredPlanet-Samsung 2016 licence is an outlier and that UnwiredPlanet are right that Mr Lasinski’s purported generosity to UnwiredPlanet in his calculations is a wolf in sheep’s clothing.

Mr Bezant’s view of this licence was that there were a number of factors which significantly reduced its reliability for the purposes of assessing FRAND offers in 2016. He addressed the implied aggregate royalty burden produced by using Mr Lasinski’s figures for this licence combined with the HPA and also with the MNPA. While I can see Mr Bezant’s point, it does not add anything to the analysis because it is just another way of explaining that the rates are low.

In principle, it is obvious that one would expect a licence granted under the same portfolio, to one of the parties in the proceedings, would be an excellent comparable. Huawei pointed out that in British Phonographic Society v MCPS [2008] EMLR 5, the Copyright Tribunal held that a settlement by a co-defendant can be an “outstanding” comparator. There is no doubt it can be, the question is whether it is. Before turning to the context in which the licence was entered into, having now analysed the licence itself, it can be said that the terms on their face raise a question mark over this licence as evidence of the fair and reasonable value of the UnwiredPlanet portfolio in 2016. Even taking into account the uncertainties, the rates are significantly lower than the rates Huawei contend for in these proceedings and are significantly lower than the rate implied by the […] licence, which is an important comparable on any view.

UnwiredPlanet’s version of the context relevant to understanding this licence is the following. PanOptis is a licensing company. It has an existing relationship with Ericsson. It had considered buying the UnwiredPlanet portfolio in 2014 but did not. In March 2015 PanOptis offered $75 million for the portfolio but UnwiredPlanet wanted $100 million and no deal was done. From about July 2014 PanOptis started having commercial discussions with Samsung. They included the possibility of Samsung taking a licence under other PanOptis telecoms patent portfolios and by the summer of 2015 they included the possibility of a wider strategic partnership. In July 2015 UnwiredPlanet approached PanOptis again, this time about purchasing the licensing companies themselves. In September 2015 PanOptis offered to buy UnwiredPlanet for $35 million. […]

Mr Ware’s evidence was that the reason that PanOptis were interested in this deal was because it would be “solving a problem for Samsung that would significantly assist the development of the wider strategic relationship we were in the process of negotiating and that could ultimately be of enormous commercial value to us”. He also said that without this the acquisition of the UnwiredPlanet portfolio in late 2015 did not fit with his strategic vision.

[…], Ericsson also approached PanOptis encouraging it to purchase UnwiredPlanet. Ericsson was keen for PanOptis to purchase the portfolio as it considered PanOptis to be a safe pair of hands. The fact that Ericsson wanted PanOptis to purchase the portfolio was an additional reason why PanOptis were interested in doing the deal because it would strengthen PanOptis’ existing strategic partnership with Ericsson. Ericsson also indicated that it would be prepared to waive the revenue sharing arrangements, which would allow PanOptis to license the UnwiredPlanet portfolio as it saw fit and which would avoid Ericsson needing to be part of any litigation with prospective licensees.

There were various negotiations and by March 2016 PanOptis dropped its offer price from $50 million to $40 million due to UnwiredPlanet’s worsening financial position.

Mr Ware said that PanOptis was able to purchase UnwiredPlanet for a price which did not represent the value of UnwiredPlanet’s patents. In his view that was because UnwiredPlanet was on the verge of insolvency. It had told shareholders that it would run out of cash reserves in July 2016 and was desperate to get out of the licensing business, to a significant degree as a result of the difficulties UnwiredPlanet had encountered in trying to license the portfolio and the cost of litigation. UnwiredPlanet characterise this as a fire sale. […].

Once PanOptis had purchased UnwiredPlanet it approached Samsung and the licence was concluded in very short order. Under that licence Samsung agreed to pay […] in cash and transfer the patents mentioned already for which I have used a value of […]. Mr Ware emphasised what he called other considerable benefits that PanOptis gained from concluding the licence with Samsung in addition to the cash and transferred patents. These were: the fact that it […], the fact that it […] and strengthening the foundations for a far wider commercial relationship with Samsung in the future.

Huawei do not agree with the way the transaction is characterised by UnwiredPlanet. They say in response:

Huawei submit that the suggestion that PanOptis accepted a […] price from Samsung because of the […] element to the acquisition does not sit with the facts of the case. Whether or not UnwiredPlanet was in financial distress is irrelevant. The licence was

I have set out the parties’ rival cases on this licence at length because it plays an important role in this case. If it is sound evidence of the value of the UnwiredPlanet portfolio then that would reduce the fair, reasonable and generally non-discriminatory royalty rate. My findings on the context in which the 2016 UnwiredPlanet-Samsung licence arose are as follows. By the time it was purchased UnwiredPlanet was in serious financial trouble. The only licence UnwiredPlanet had been able to agree was with Lenovo and UnwiredPlanet was engaged in very expensive multinational patent litigation in an effort to establish its rights. By late 2015 – early 2016 UnwiredPlanet was close to insolvency. I accept Mr Ware’s evidence about what happened. The price PanOptis paid for UnwiredPlanet was lower than the market value of the patent portfolio because of the serious financial difficulties UnwiredPlanet were in at the time. As regards his discussions with Samsung, the picture Mr Ware painted of the reality of high level negotiations with that major multinational organisation was convincing and credible. PanOptis had the ability and the means to buy UnwiredPlanet in any event but I find that the key reason why PanOptis did buy UnwiredPlanet when they did and for the price they paid was in order to build trust with Samsung and because Samsung were prepared to take a licence under the portfolio in a deal in which the cash component […]. The purchase was being “de-risked”, as Mr Ware put it. The long term benefits to PanOptis which would derive from this were regarded by PanOptis as important and are in fact potentially very valuable. The arrangements did not give PanOptis a contractually enforceable right to the benefits derived from building trust with Samsung but that does not mean it was not well worth doing.

These findings about the context of the licence together with the findings about low rates in the licence itself support one another. I conclude that the licence does not represent useful evidence of the market value of the UnwiredPlanet patent portfolio. (c) Ericsson-Huawei 2016

[…]

The issue to resolve concerning the Ericsson-Huawei 2016 licence is the significance of the fact that [certain] terms were the product of an arbitration. I can deal with this shortly because I believe the answer is clear. Terms which were settled by an arbitrator are not evidence of what willing, reasonable business people would agree in a negotiation. In that sense a royalty in the licence is not probative of the market value of the portfolio under licence at all. Decisions of other courts may have persuasive value but that will largely depend on the reasoning that court has given to reach its conclusion. An arbitral award is at least capable of having a similar persuasive value, but reasoning supporting the terms in this licence is not available. I know that the arbitrators [made findings about the effective royalty rate to be paid …] and I also know that this is not far above the […] rate from the […] licence […]. It is […]. Dr Kakaes and Mr Lasinski were witnesses in the arbitration and they have given some very brief evidence about what the arbitrators did or did not rely on. Mr Lasinski said that the arbitrators […]. I am not prepared to place weight on this evidence in relation to the value of Ericsson’s portfolio. There is no good evidence of what the terms of the […] or […] licences are. Moreover without seeing the reasoning of the arbitrators one cannot see how they arrive at the conclusion they did.

Huawei submitted the licence was relevant because it was a rate someone (Huawei) was paying. So they are, but since […], the fact they are doing so is not evidence of the value of the portfolio.

[…] Without sight of the arbitrators’ reasons I cannot accept that submission.

Ericsson-Samsung 2014

[…]

[…]

[…]

Mr Lasinski referred to a witness statement of Mr Kim of Samsung. Mr Kim was going to attend trial but following the settlement he did not. In the relevant paragraph Mr Kim said that Samsung had agreed […]. Relying on this Mr Lasinski did not take account of the […]. UnwiredPlanet criticised Mr Lasinski for this but I do not see the force in that criticism since Mr Lasinski’s report makes his approach transparent so that the court can understand the basis on which he approached it.

UnwiredPlanet contended that little weight should be placed on Mr Kim’s evidence because (i) he did not in the end attend trial, (ii) […], (iii) as with Lenovo, there are good reasons why Samsung would not want a licence to contain rates they thought were unreasonably high since those rates could be used in later negotiations or in court, and (iv) although Ericsson was also a party to this proceeding Ericsson’s fact witnesses had not been permitted to see this part of Mr Kim’s evidence and so were unable to comment on it. The fourth point came as a surprise to me during the trial (I had assumed Ericsson had had a full opportunity to answer this evidence) and it seemed that Mr Lasinski had made the same assumption.

[…] At least without hearing the witness give evidence, I would be reluctant to accept the idea that […]

Standing back, this licence represents a solid piece of evidence of what reasonable people in the industry would do. The two parties have broadly equivalent economic strength. It has been freely negotiated rather than set by an arbitrator. There was litigation between Ericsson and Samsung before it was agreed but I doubt parties of the size and sophistication of these two were troubled by that. This licence is solid evidence

Ericsson-Huawei 2009

[…]

[…]

[…]

[…] Consistent with his approach for […]. Mr Bezant used the MNPA along with the 80:20 approach. […]

UnwiredPlanet submitted that Mr Lasinski’s approach to this licence was another wolf in sheep’s clothing in that the assumptions he had made were not really in UnwiredPlanet’s favour at all. That was because the resulting rates derived from this process were presented as corroborative of the rates derived from the other major comparables relied on, whereas in fact the rates in this licence […] and do not corroborate Huawei’s case. Another symptom of this, submitted UnwiredPlanet, was that while Mr Lasinski had applied a 50% adjustment to raw rates from other pre-2013 licences such as […], because he said the legal landscape changed around 2013 […], he had not applied that adjustment to the rate derived from this 2009 licence when it was presented in Mr Lasinski’s figure 4 (3^rd Report) which Huawei put at the forefront of their opening submissions.

It is not right to say that Mr Lasinski did not apply the 50% adjustment to the rates from this licence at all. They are in his reports. However it is fair to point out that the derived rate which was given prominence was derived using the unadjusted figure. Mr Lasinski’s second report suggested his approach was justified because the 2009 licence was not entered into under the direct threat of an injunction but that would apply to other licences for which he did apply the adjustment.

[…]

[…]

[…]

In order to apply that […] rate to 2014 or 2016 I need to address two further points.

Mr Lasinski said he had observed a general decline in rates since 2013 and he referred to evidence […]. Mr Lasinski attributed this to the evolution of FRAND case law and the fact that there was limited guidance about FRAND before 2013. Mr Lasinski referred to four US decisions cited above (Microsoft v Motorola, In re Innovatio IP

I accept Mr Lasinski’s view that there is some evidence of a decline in some rates over time and I am sure that at least part of the explanation is the emergence by 2013 of decisions in which courts were prepared to set FRAND rates, which in turn strengthened the bargaining position of licensees by reducing the power of the threat of an injunction. However the trend is not simple since […]. In any event I am not convinced that one simply adjusts for all this by taking any rate derived before 2013 and halving it. The evidence does not justify such a simplistic approach.

Ericsson – [M] 2013

[…]

[…]

[…]

[…]

[…]

[…]

[…]

[…]

[…]

Mr Lasinski applied his 50% adjustment to the rates in this licence for the developments in FRAND case law given that it was signed in 2013. I have rejected such an approach as too simplistic.

Ericsson: [N,O,P]

These are three […] licences […]:

There was a suggestion that UnwiredPlanet or Mr Bezant singled out these licences as opposed to licences with larger licensees such as […], Huawei or Samsung but UnwiredPlanet explained that that is not what happened. The case management directions included provision for parties to choose certain licences to inspect. The choices were sequential and UnwiredPlanet’s choice was made after licences from bigger players had already been chosen.

[…]

Ericsson – ZTE 2011

ZTE is a very large Chinese company with worldwide reach. Huawei accepted that ZTE was more similar to them than [M] or RIM.

[…]

[…]

Other factors to take into account are these: first, as a 2011 licence Mr Lasinski’s point on FRAND case law is relevant; second, Mr Lasinski explained this licence followed lawsuits in Europe; third, […].

Mr Bezant plots three figures derived from this licence in the chart above. That is unwarranted.

[…]

Ericsson-RIM

[…]

Mr Bezant accepted this licence was more difficult to interpret than [M] or ZTE. […]. RIM make or made the Blackberry handsets which in their time were popular, successful and expensive devices in the USA and Europe. Huawei made an additional point in closing that this licence was concluded before a general decline in selling prices of handsets experienced from 2013 onwards but in his oral evidence when the point was put Mr Bezant accepted there was pressure on prices because of new products but did not think it was as absolute as was suggested.

[…]

[…]

Ericsson – Apple 2008

[…]

[…] The exercise adds a further level of significant uncertainty onto an already uncertain exercise. Apple’s interest was to negotiate how much they had to pay, not to negotiate what the benchmark rate would be.

Ericsson-Sony 2012

[…]

[…]

Mr Lasinski’s view was that this licence should not be included in the assessment, mainly because it is a related party transaction and also owing to the other uncertainties. UnwiredPlanet pointed out that this stance was inconsistent with the stance taken on the UnwiredPlanet-Samsung 2016. I agree. They are not identical situations but the impact of the wider context in which each transaction exists reduces the weight one can put on it. […]

The Ericsson licences as a whole

The only licence on Mr Bezant’s chart which has not been dealt with so far is […] but it is so old that I will not rely on it.

Having now reviewed all the relevant Ericsson licences, looking at them as a whole is a useful exercise. The ones on which I am prepared to place any weight are:

[…]

I find that an appropriate rate to use as representative of the value of E of Ericsson’s 4G SEP portfolio in licensing a multimode handset is 0.80%. That is arrived at by looking at the range of [rates from] the [Q] licence, starting from the top of that range, and adjusting upwards by a factor which is […]. I have arrived at this number and approach after considering all the material. The main reasons for this are as follows. The [Q] licence is the best place to start. The [range] represents the payments [the licensee] will have been most concerned about and it makes sense to start there. Taking the upper end of that range […] is appropriate because Mr Lasinski’s preferred figure is […] anyway and because of the existence of […]. So I derive […] as the rate on which to place weight from the [Q] licence. A further increase it is appropriate to take into account is in the other Ericsson licences and in particular the fact that all the […] rates are much higher than […] including the […] rate which one would expect to be relatively low. But it would not be appropriate to raise the [rate] up to the level of these other rates because of Mr Lasinski’s point about FRAND case law. The figure for E for multimode 4G of 0.80% is more than […] but less than […]. The final number was also derived as a number expressed to one decimal place (0.8%) to recognise the uncertainties but in keeping with the convention used in this case I have written it as 0.80%.

For 2G and 3G I derive a single rate for E of 0.67%. That is arrived at by scaling 0.80% with the ratio of Mr Lasinski’s 2G/3G rate to his 4G rate in the […] licence. It is appropriate to do this because […].

Turning to infrastructure, […] Taking all this into account I conclude that there is no sufficient evidence of a consistent difference between rates for single mode infrastructure and for multimode handsets which is of sufficient magnitude to justify arriving at a different rate for infrastructure. I find that the value for E for infrastructure should be the same as the multimode 4G and 2G/3G handset rates, i.e. 0.80% for 4G infrastructure and 0.67% for 2G or 3G infrastructure.

The other inference to draw from standing back and considering […] is that based on […] the rate for China should be half the rate for the rest of the world. […]

Other licences

The only other licences worth mentioning at all are by Qualcomm. The rates are much higher […]. I will not place weight on the absolute levels of Qualcomm’s rates in assessing the level of a benchmark rate.

Other indications relating to rates

Mr Bezant’s charts often include the ARR, that is the royalty floor rate in the MSA. I have already rejected its use in this way.

Rates have been set in some of the decisions of foreign courts cited by the parties.

The Microsoft v Motorola, Innovatio, and Ericsson v D-Link judgments are not concerned with 2G, 3G or 4G technology, they were mostly about the IEEE 802.11 standard (Microsoft v Motorola was also about the ITU H.264 video standard). Therefore the rates set in these are not readily comparable. The German FRAND decisions do not set or approve rates.

The IP High Court in Japan in the Samsung v Apple case [R6/1] used the top down approach by deciding that the aggregate royalty burden for 3G should be 5% and deciding that from families declared essential there were 529 patents that are or are likely to be essential, I think based on the Fairfield/Goodman & Myers study.

In China in the Huawei v InterDigital case the Guangdong High People’s Court upheld the FRAND rate of 0.019% for InterDigital’s portfolio. This was arrived at by comparison with two rates derived by unpacking other InterDigital licences. They were an InterDigital-Apple licence with an unpacked effective rate of 0.0187% and an InterDigital-Samsung licence with an unpacked effective rate of […]. These rates cannot be converted into rates relevant to this case without carrying out another portfolio comparison which it not necessary. What this decision does do is support Huawei’s case that rates in China are low.

None of these other sources provide any reason to adjust the benchmark rate I am seeking to derive. It is notable that the Japanese court and the Chinese court have approached the determination of FRAND rates using the same techniques as have been addressed in this case.

What is the benchmark FRAND rate for UnwiredPlanet?

The outcome of considering the comparables is that I have benchmark rates for Ericsson of 0.80% for 4G and 0.67% for 2G and 3G and no other reliable comparables. Given the previous conclusion for the strength ratio R of UnwiredPlanet to Ericsson for 4G multimode handsets as 7.69%, that indicates a benchmark royalty rate for UnwiredPlanet for a 4G multimode handset in 2016 is 0.062%.

Applying the total royalty burden as a cross-check produces the following. A benchmark royalty rate for UnwiredPlanet for a 4G multimode handset of 0.062% coupled with a figure of 0.70% for UnwiredPlanet’s share S of the Relevant SEPs for

The figure for 4G infrastructure will be 0.072% using R = 8.95% for 4G infrastructure (0.072%=0.80% x 8.95%). The cross-check T comes to 7.0% (using S for 4G infrastructure of 1.02%) which is appropriate.

The 4G figures as well as the corresponding figures for 2G and 3G are in this table:

The total royalty burden T implied by each of these rates falls within an appropriate range. The value of T for 3G multimode handsets at 5.6% is not far out of line with the judgment of the internationally respected IP High Court of Japan. The 3.1% value for 3G infrastructure is somewhat lower than 5% but not far away.

Subject to the hard-edged non-discrimination issue I conclude that these benchmark rates are FRAND. They are fair, reasonable and generally non-discriminatory.

The impact of hard edged non-discrimination on the FRAND rate

Huawei submit that pursuant to the non-discrimination limb of FRAND UnwiredPlanet are obliged to offer the same or similar rates to Huawei as they have extended to Samsung in the 2016 UnwiredPlanet-Samsung licence. In closing Huawei called this a “hard-edged point” (Closing para 210) and I have adopted that expression for the sake of clarity. This is the place in the analysis to consider Huawei’s hard-edged point. To recap, whereas general non-discrimination was taken into account in setting a benchmark rate for the portfolio, general non-discrimination did not discriminate between licensees. The benchmark would be equally applicable to a major player like

At times it was not completely clear whether Huawei were running this hard-edged point as an aspect of their case that 0.041% was the right rate for 4G, albeit a generous one, or whether this was a distinct issue in the sense that even if 0.041% would otherwise be the right rate, the Samsung licence means that a lower rate corresponding to that should apply. In closing Huawei clarified that this is a distinct issue (Day 19 p194-195) such that if I accept Huawei’s non-discrimination case I should set the rate as the same as the Samsung rate.

On Huawei’s case in closing the difference between Samsung’s rate and the benchmark was a substantial, […] for Samsung vs a benchmark of 0.041%). Given the findings I have made the difference is even greater. A fair assessment of the 4G rate in the Samsung licence is between […]; whereas the 4G benchmark FRAND rates arrived at above are 0.062% for handsets or 0.072% for infrastructure. Therefore this hard-edged point is important and is capable of making a significant difference to the outcome.

A significant aspect is an issue of principle which, so far as the parties have been able to discern, no other court has had to face. The issue is about how closely aligned the requirements of the non-discrimination limb of FRAND pursuant to the ETSI FRAND undertaking are relative to the corresponding requirements of competition law. It is common ground that competition law (Art 102(c)) only prohibits discriminatory behaviour to the extent that behaviour is capable of distorting competition. The question is whether that condition or something akin to it, which is not mentioned expressly in the ETSI FRAND undertaking, is nevertheless a relevant aspect of ETSI FRAND. But in order to get to this there is some ground to be cleared.

Obviously FRAND has a non-discrimination limb, as I have already explained. Huawei pointed out that both Mr Bezant and Mr Lasinski agreed that that non-discrimination obligation “means that licensors should treat similarly situated licensees similarly”. Huawei submit that in the jargon of non-discrimination, Samsung are “similarly situated” to Huawei and so UnwiredPlanet are obliged to offer the same or similar rates to Huawei as they have extended to Samsung in the 2016 UnwiredPlanet-Samsung licence. UnwiredPlanet do not accept the hard-edged point put by Huawei. Their case is that UnwiredPlanet are not obliged to offer Huawei the same rate as the Samsung rate. That is because Huawei are not “similarly situated” to Samsung; the Samsung licence is not an equivalent or comparable licence to the Huawei licence being considered; and, even if those two points are wrong, the non-discrimination limb of FRAND contains the same or an analogous aspect as the requirement in competition law only prohibits conduct which is capable of distorting competition. UnwiredPlanet point out that Huawei have disavowed any attempt to conduct the economic analysis necessary to establish that in this case. In reply Huawei submit that no such analysis is necessary, citing British Airways v Commission Case C-95/04 [2007] ECR I-2331.

Competition law non-discrimination forms part of abuse of dominance. As it relates to prices it can be summarised as follows. First the underlying principle is that comparable situations must not be treated differently and different situations must not be treated alike unless such treatment is objectively justified (relying on Franz Egenberger C313/04, EU:C:2006:454 at [33]). Second, Article 102(c) TFEU prohibits “applying dissimilar conditions to equivalent transactions with other trading parties, thereby

Stated at this level these principles are not controversial. Both sides approached this issue on the basis that concepts such as similarly situated parties, equivalent/comparable transactions, and objective justification, were the same under the non-discrimination limb of FRAND as they are in competition law. Mind you, none of those concepts are mentioned expressly in the ETSI FRAND undertaking either.

I will say now that I find Samsung and Huawei are “similarly situated” on any view. Huawei do have a far larger focus on infrastructure than Samsung but this does not justify a finding that Huawei and Samsung are not similarly situated. Since this case is concerned with a worldwide benchmark rate it is appropriate to consider the worldwide position of the two undertakings. They are both very large multinational telecoms manufacturers active in the same handset and RAN infrastructure markets. They are both in the top three handset vendors worldwide (the other is Apple) and while Huawei is the top 4G RAN infrastructure maker, Samsung is another major supplier in that market.

UnwiredPlanet’s approach was to argue that the factors in Mr Ware’s evidence which made the Samsung licence not a reliable comparable in the context of setting a FRAND rate were also relevant to non-discrimination. Huawei’s argument was the same albeit from the other direction, in other words its case was that the Samsung licence was a reliable comparable for both purposes for the same reasons. Both sides effectively equated the concept of an equivalent/comparable transaction in the context of nondiscrimination with a comparable licence used in the analysis of royalty rates. Clearly the concepts overlap but in my judgment the two are not identical and the distinction is acute in the context of the Samsung licence. Used in an assessment of an appropriate rate in accordance with the ETSI FRAND obligation, comparable licences act as evidence of the value of the property being licensed. A FRAND rate could still be set even if no licences existed at all and even if there was no licence from UnwiredPlanet to Samsung. Using that kind of evidence inevitably takes into account the general nondiscrimination aspect of the ETSI FRAND obligation mentioned already. But a hardedged non-discrimination obligation, if it means anything, arises as a consequence of entering into particular transactions with particular licensees. The issue is what effect in law arises from the fact that UnwiredPlanet have granted a licence to Samsung under the same portfolio. The effect of this grant is that a major competitor of Huawei, one who is otherwise “similarly situated” to Huawei, has licensed the same portfolio.

Mr Ware’s evidence explains the motives leading to the 2016 UnwiredPlanet-Samsung licence and explains why it was entered into on the terms it contains. I have accepted that this shows why the terms are not reliable evidence of the value of the portfolio but

Mr Ware’s evidence is sufficient to show that the weight to be attached to the pricing in this licence is low, as a result of the other benefits PanOptis perceived would flow from it and the circumstances UnwiredPlanet were in at the time, however those benefits and circumstances do not derive from any objective characteristics of the transaction itself. It is in the end nothing more than a patent licence (with the associated assignment). UnwiredPlanet’s or PanOptis’s motives for selling this licence cheaply on that occasion do not change the fact that they did sell the licence cheaply. The consequence of the licence is that PanOptis has been able to enhance its general relationship with Samsung and therefore to have a relationship with Samsung which it does not have with Huawei, but I reject the suggestion that this means that the transaction has features vis a vis Samsung which make it different in any objective sense relevant in this context from the licence Huawei is entitled to.

UnwiredPlanet repeatedly emphasised that one has to take a realistic common sense view and that non-discrimination cannot mean that businesses have to charge all customers the same price for their goods or services, citing Purple Parking v Heathrow Airport [2011] EWHC 987 (Ch) (Mann J), and Attheraces Ltd v The British Horseracing Board Ltd [2007] ECC 7 (Court of Appeal), as well as two textbooks:

I accept these broad statements, however the circumstances in this case are striking. The licensee (Samsung) is, at best, one of a handful of major licensees. The licence is for the same portfolio and relates to the same acts. It is contemporaneous. I find that for the purposes of considering hard edged non-discrimination, the Unwired PlanetSamsung 2016 licence is an equivalent or comparable transaction to the putative licence under consideration in this case, the one between UnwiredPlanet and Huawei.

So given these findings, in terms of competition law and assuming abuse of dominance, that would leave the issues of distortion of competition and objective justification to be considered. Before going further however it is worth reflecting on the position which has been reached. UnwiredPlanet’s counsel described the discrepancy in pricing in the Attheraces case as a very striking discrepancy. That was a three-fold difference (£361 vs £900). The difference between the benchmark rate and the Samsung rate is much larger than that. On one view it is a […] difference. Moreover the discrepancy between

I turn to consider whether distortion of competition is part of the non-discrimination limb of FRAND. Huawei’s simple case was that it was not mentioned in the words of the undertaking. However as I have already noted, none of the concepts which both sides agree need to be considered to address non-discrimination are mentioned in the undertaking either so that simple point is not as strong as it might seem at first sight. Conversely beyond submitting that it should be included, UnwiredPlanet’s argument did not give reasons why.

In terms of authority, there is no case on the point. In terms of the economics literature

Prof Neven had exhibited the paper and referred to it as influential (albeit in another context). I infer from the paper that there is no wide agreement amongst economists about how the non-discrimination limb of FRAND applies. In the passage quoted the paper proposes three possible interpretations. First, the “standard economic definition”. That is one in which all licensees pay identical rates on identical terms. It is not supported by either party before me. I am not surprised. If that is what the ETSI undertaking was supposed to mean it could readily have been written in that way. While such an approach has the virtue of simplicity it would be impractical in practice. It

The second version of non-discrimination referred to is one which “some economists” have proposed. It is very weak. Since the ETSI FRAND undertaking already obliges licensors to offer licences to everyone, it does not add anything to that. It is also weaker than the benchmark FRAND rate approach, which at least applies to all licensees with licences of the same type. The benchmark approach does not mean that all licensees must pay exactly that rate but it provides a benchmark against which such a rate can be judged. In practice it will stop licensees having to pay much more than the benchmark set by reference to the value of the portfolio. Competition law can intervene to penalise the imposition of excessive prices much higher than the benchmark FRAND rate. Neither side supported this weak second version of non-discrimination.

The third interpretation proposed by Prof Carlton is based on similarly situated firms paying the same rate. It is based on the same concepts as competition law. In effect this is the one Huawei contend for although one cannot take that too far. The paper is not purporting to set out a fully worked out proposal on the correct approach in law to interpret the FRAND undertaking. It is not focussed on the distinction between the arguments of Huawei and UnwiredPlanet.

The final text from the economics literature to refer to is Dr Niels’ textbook Economics for Competition Lawyers (2^nd Ed, Oxford 2016). At paragraph 8.51 he states that the non-discriminatory condition in FRAND “is usually interpreted in the same manner as the general criteria for anti-competitive price discrimination under the abuse of dominance rules”. Although Dr Niels does not cite a reference for that, it may be that he had in mind the kind of thing proposed by Prof Carlton. As with Prof Carlton’s paper, Dr Niels’ paragraph is not focussed on the distinction between the arguments of Huawei and UnwiredPlanet.

In my judgment the ETSI FRAND undertaking should not be interpreted so as to introduce the kind of hard-edged non-discrimination obligation supported by Huawei without also including consideration of the distortion of competition. Competition law does not seek to prohibit different prices being charged to different customers. An important aspect of the way that result is assured in competition law is by the requirement that only terms which are sufficiently dissimilar to distort competition are prohibited. In other words, the various elements of the competition law applicable discriminatory pricing operate as a whole to achieve a fair balance. Splitting off some parts without the others is unbalanced and risks unfairness.

Having got this far it seems to me that it is not necessary to read this hard-edged nondiscrimination obligation into the ETSI FRAND undertaking at all provided one takes a benchmark rate approach to assessing a royalty under the ETSI FRAND undertaking.

Therefore I conclude that the true interpretation of the ETSI FRAND undertaking from the point of view of non-discrimination is that a benchmark FRAND rate should be derived which is applicable to all licensees seeking the same kind of licence. That is what I have called general non-discrimination. If, contrary to this view, the FRAND undertaking also includes a specific non-discrimination obligation whereby a licensee has the right to demand the very same rate as has been granted to another licensee which is lower than the benchmark rate, then that obligation only applies if the difference would distort competition between the two licensees.

I turn to the point in British Airways. Huawei made clear that they had not attempted to conduct the kind of economic analysis which would be required to establish that the dissimilarity between the rates in the Samsung licence and the rates demanded by UnwiredPlanet distort competition. They contend it is not necessary in this case and rely on British Airways as authority for the proposition that such an analysis is not always necessary. The case was about travel agents’ commissions and a reward scheme practised by BA. The CFI had held that the competition between them had “naturally” been affected by the discriminatory payment conditions in BA’s reward scheme. On appeal BA submitted this was not sufficient to justify a finding of breach of what is now Art 102(c) and what was needed was concrete evidence of competitive disadvantage. This is recorded at paragraph 142. The CJEU rejected that submission in paragraphs 143 to 149.

The CJEU’s analysis starts by reaffirming the prohibition on discrimination in what is now Art 102(c) and emphasising the requirement that a distortion of competition is required:

“144 Therefore, in order for the conditions for applying subparagraph (c) of the second paragraph of Article 82 EC [now Art 102] to be met, there must be a finding not only that the behaviour of an undertaking in a dominant market position is discriminatory, but also that it tends to distort that competitive relationship, in other words to hinder the competitive position of some of the business partners of that undertaking in relation to the others (see, to that effect, Suiker Unie, paragraphs 523 and 524).

145 In that respect, there is nothing to prevent discrimination between business partners who are in a relationship of competition from being regarded as being abusive as soon as the behaviour of the undertaking in a dominant position tends, having regard to the whole of the circumstances of the case, to lead to a distortion of competition between those business partners. In such a situation, it cannot be required in addition that proof be adduced of an actual quantifiable deterioration in the competitive position of the business partners taken individually.” 506. Then the CJEU turned to the CFI’s decision on the facts, as follows:

“146 In paragraphs 237 and 238 of the judgment under appeal, the Court of First Instance found that travel agents in the United Kingdom compete intensely with each other, and that that ability to compete depended on two factors, namely 'their ability to provide seats on flights suited to travellers' wishes, at a reasonable cost' and, secondly, their individual financial resources.

147 Moreover, in the part of the judgment under appeal relating to the examination of the fidelity-building effect of the bonus schemes at issue, the Court of First Instance found that the latter could lead to exponential changes in the revenue of travel agents.”

In this context the CJEU then rejected BA’s submission that concrete evidence of competitive disadvantage was needed:

“148 Given that factual situation, the Court of First Instance could, in the context of its examination of the bonus schemes at issue having regard to subparagraph (c) of the second paragraph of Article 82 EC, move directly, without any detailed intermediate stage, to the conclusion that the possibilities for those agents to compete with each other had been affected by the discriminatory conditions for remuneration implemented by BA.

149 The Court of First Instance cannot therefore be accused of an error of law in not verifying, or in verifying only briefly, whether and to what extent those conditions had affected the competitive position of BA's commercial partners. The Court of First Instance was therefore entitled to take the view that the bonus schemes at issue gave rise to a discriminatory effect for the purposes of subparagraph (c) of the second paragraph of Article 82 EC. The second part of the fifth plea is therefore unfounded.”

The CJEU’s judgment therefore is that on the facts of that case the CFI were entitled to infer that the ability of the travel agents to compete with one another had been affected by BA’s discriminatory payment terms without “any detailed intermediate stage”. The court explained (at paragraph 145) that additional proof of actual quantifiable deterioration in the competitive position of the victims taken individually was not needed in a case in which the behaviour complained of tends, having regard to the whole of the circumstances of the case, to lead to a distortion of competition between those victims.

As UnwiredPlanet pointed out, in that case the court had had evidence of the competitive position in the downstream market (i.e. between travel agents). It was able to find in paragraph 146 that they compete intensely with each other and that their ability to do so depended on two factors, one of which was their individual financial resources. It was also able to find in paragraph 147 that the BA payment scheme could lead to exponential changes in the revenues of travel agents. As UnwiredPlanet also pointed out, there has been no analysis in this case of the extent of competition in the phone or infrastructure market, nor has there been any analysis of the factors necessary to compete in that market to justify a finding like the one referred to in paragraph 146,

The principle to be derived from the British Airways judgment is that there must be some evidential basis from which an inference can be drawn that the behaviour complained of tends to distort the relevant competitive relationship, but if such an inference can properly be drawn then it is not necessary to prove the existence of a quantifiable impact on the competitive position of the victims.

What evidence is there in this case? Huawei rely on two strands of evidence: first, Ericsson’s intentions and the genesis of UnwiredPlanet, and second, direct financial evidence.

On the first issue Huawei rely on internal Ericsson emails and other documents. […]

Although in the original configuration of the trial both Ericsson and Samsung would have been there and called witnesses, following the Samsung settlement and the consequential rearrangements neither of those parties appeared or called witnesses. So these documents have not been put to a witness nor, as far as I am aware, were they subject to a Civil Evidence Act Notice. In their closing Huawei referred to a number of disclosure documents. UnwiredPlanet did not submit that the various disclosure documents were inadmissible (given CPR PD32 paragraph 27.2) but cautioned as to the weight to be attached to disclosure documents put in this way. I accept the submission in relation to other documents Huawei referred to, but the two documents referred to above really just illustrate a point which has never been seriously disputed, that Ericsson’s purpose in transferring the patent portfolio was to make more money. In paragraph 16 of the judgment on the competition law strike out application [2015] EWHC 2097 (Pat) Ericsson’s case about its motives was summarised this way:

“16.

Ericsson describes its motive in transferring part of its portfolio to UnwiredPlanet as being to enable it fairly to earn more revenue. Its concern is that while the patents remain within Ericsson’s very large portfolio, its ability to earn a fair revenue in respect of those inventions is hindered. Once the patents are transferred, UnwiredPlanet will be able to obtain fairer and therefore greater remuneration for them than Ericsson was able to obtain while still ensuring that any royalties collected in respect of essential patents are FRAND.”

Since greater remuneration from the patents has to come from the rest of the industry,

In financial terms Huawei’s complaint is presented by reference to two pie charts which are intended to illustrate the disparity between Samsung and Huawei comparing the July 2016 UnwiredPlanet proposal against the Samsung licence (using Mr Lasinski’s figures) and taking into account the difference in revenue which would be licensed between the two undertakings. The figures are annual and in dollars. They are for 2G,

I include the pie charts because Huawei put them forward with some prominence but just like UnwiredPlanet’s chart of rates included in this judgment and like any visual presentation, care needs to be taken with them. UnwiredPlanet’s chart presents far too many rates from the same licence as if they were independent data points. In Huawei’s pie chart here the area of the left circle is about $103 billion whereas the area of the right circle is $48 million and so, scaled with the left circle, the right circle should be tiny. There is no question that the rates applied to Samsung under the Samsung licence are far lower than the benchmark rates derived above and lower still than the rates claimed by UnwiredPlanet in July 2016. I will take it that the value of Huawei’s annual licensed revenues is likely to be about half Samsung’s.

In cross-examination counsel put figures to Dr Niels. The rate for Samsung was put as in effect […] per device for Samsung (assuming […]). The equivalent rate for Huawei was put at about 50 to 75 cents per device. These are reasonable assumptions, which if anything favour UnwiredPlanet. A figure for Huawei’s profit margin was taken from public information as being between about $6 and $19 per device. Counsel put to Dr Niels that in that context the difference in rates ([…] vs 75 cents) would appreciably distort competition. Dr Niels did not agree. He was prepared to assume the figures were correct but explained that one could not draw the conclusion that competition would be distorted because given the average selling price of the products (for Huawei assuming $164 to 185) the rate is very, very low. He acknowledged the rate was a higher percentage of the profit margin but that did not change his view. He said that no-one had done a detailed analysis of distortion of competition and to do that you would have to look at how the market works downstream and how operators set prices. In his opinion one still has to ask what ultimately happens to the small price differential, is it passed to end consumers, mobile operators or is it absorbed by Huawei? Even

I accept Dr Niels’ evidence on this. Although the relative difference in royalty rates is large, to be considered in the context of possible distortion of competition they must be expressed relative to the margins on the relevant products. Expressed that way they are very small percentages. The available evidence does not support an inference that the behaviour complained of tends to distort the competitive relationship between Huawei and Samsung.

In giving his answers Dr Niels at one stage said that he was referring to conducting a full analysis “under the competition law rules” and was “not commenting here on the ND in FRAND”. What the witness meant was that he was alive to the point that there was a debate whether distortion of competition was relevant to the ND in FRAND. Since I have held that it is, Dr Niels’ evidence is germane.

Finally I should mention that Huawei were entitled at one stage to think that UnwiredPlanet were conceding that distortion of competition was not a necessary part of hardedged non-discrimination. Huawei did not suggest that they did not call further evidence on this topic because they thought UnwiredPlanet had conceded the issue. If UnwiredPlanet did change their stance, the change did not prejudice Huawei.

So I reject Huawei’s hard-edged non-discrimination point. There is no need to consider objective justification.

Rates - conclusions

Since the hard-edged non-discrimination point has been rejected, I find that the benchmark FRAND rates for the UnwiredPlanet portfolio are as set out in the table above. It follows that none of the offers made by UnwiredPlanet in 2014, 2015 or 2016 involved rates which were FRAND. They were too high. It also follows that the benchmark rates on which Huawei’s offers have been based were not FRAND either. They were too low.

The Other Disputed Terms

Aside from a rate, the parties do not agree about other terms. The major disagreement is about scope (UK or worldwide) then there are disagreements about a few of the terms of a UK licence. Since Huawei did not engage with UnwiredPlanet on the terms of a worldwide licence, I only have UnwiredPlanet’s position on that.

What licence scope is FRAND – UK or worldwide?

Aside from the rate, the question of scope is the most significant point in the case. The parties are diametrically opposed. Huawei are willing to take a licence under Unwired

The same topic arises under the ETSI FRAND undertaking and in competition law. It makes more sense to consider it once. The answer to the scope question has a bearing on the question of an injunction and equitable refusability but those issues are best addressed separately. For the purposes of the competition law aspect I will assume UnwiredPlanet is in a dominant position.

Huawei submitted that it is a fundamental principle of EU competition law that a dominant undertaking cannot tie or bundle together, with a product or service in respect of which it holds a dominant position, some other product or service which does not fall within the same market. They relied on Case T-201/04 Microsoft Corp [2007] ECR II-3619, in which the Court of First Instance approved a fourfold test which had been used by the Commission in that case (see paras 842, 859, 862) as follows:

“- first, the tying and tied products are two separate products;

-

second, the undertaking concerned is dominant in the market for the tying product;

-

third, the undertaking concerned does not give customers a choice to obtain the tying product without the tied product; and

-

fourth, the practice in question forecloses competition.”

The first three points are fairly self-explanatory. The parties did not agree about the fourth point. Huawei submitted that the fourth test is simply whether the conduct is “capable of restricting competition” (Microsoft at paragraphs 866-867) and that this is consistent with the overall position under Art 102, which requires proof only of potential effects on competition. Huawei referred to paragraphs 868, 977 and 1035 and submitted that in these paragraphs the CFI observed that it is ordinarily presumed in tying cases, in the absence of specific circumstances suggesting otherwise. Huawei submitted that UnwiredPlanet had not suggested any such specific circumstances in the present case and that both the bundling practices in issue, the multi-jurisdictional bundling and also the bundling of SEPs with non-SEPs which took place in UnwiredPlanet’s April 2014 offer, pose an obvious and real threat of distortions in competition.

UnwiredPlanet submitted that Prof Neven had not been asked to analyse bundling in terms of a standalone abuse (which was correct). They submitted that an appreciable effect on competition always had to be shown (citing Post Danmark C-209-10 ECLI:EUL2012:172 at paras 24, 39 and 44, as well as the EU Commission’s Guidance on enforcement priorities in applying Art 102 (24^th Feb 2009 OJ EU C45/7) at paragraphs 20-22).

On the issue of principle, Huawei’s submission about the Microsoft case does not completely capture what happened in it. The point was that the Commission had not done what it normally does and just made an assumption, rather the Commission had examined the actual effects on the market. In paragraph 868, the CFI explains the Commission had decided that in the circumstances of the case “it could not merely assume as it normally does in cases of abusive tying, that the tying of a specific product

For abuse to be established there must be a finding that the practice in question forecloses competition. The legal principle I take from the authorities is simply that such a finding may be based on inference but the inference must be justified in all the circumstances. Just because it is normally assumed, it does not follow that it will always be assumed. The circumstances may be such that such an assumption cannot be made and a close analysis of the actual effects is required.

Prof Neven addressed a topic he called “Demanding royalties for acts which are not within the scope of the patents in suit”. The example he gave, which he described as odd and without clear economic justification, was of a patentee with a patent in two jurisdictions which had been upheld in one but revoked in the other but insisted on being paid in both. Stated that way the Professor makes an important point but one has to be careful with the example. As soon as patent portfolios are being licensed it is not straightforward to say that a demand for payment for a portfolio licence is for payment for a given patent in the portfolio. In the present case it is common ground that the licence will be for all declared SEPs even though the royalty is set by reference to a subset – the Relevant SEPs - and this applies to the UK only portfolio as much as to the worldwide portfolio. In such a case the licensor is not really demanding payment for each declared SEP. If one of the declared but non-Relevant SEPs in a portfolio was revoked, leaving Relevant SEPs behind, it would not change the benchmark royalty rate.

In any event there are mechanisms which can be agreed in a licence to cater for the situation described by Prof Neven. For instance one could have a term which carves out a country completely or reduces the royalty rate if there is a material change to the number of Relevant SEPs (or declared SEPs). In other words the problem is not inherent in multijurisdictional portfolio licensing.

As Prof Neven accepted, portfolio licensing is common industry practice and has efficiency benefits. It saves transaction costs for both licensors and licensees and obviates the need to determine a royalty on a patent by patent basis.

As far as I am aware every patent licence in the trial bundles in this case is a worldwide portfolio contract. The vast majority are worldwide licences. There are a few in which a given territory is carved out in such a way that no licence is granted for sales in that territory whereas the rest of the world is licensed. The best example is China in the […] licence. Even with a carve out of some kind the contracts are still worldwide agreements although it would be inaccurate to call them worldwide licences.

Of course just because everyone is doing something does not mean that the conduct is necessarily unproblematic. Nevertheless, the inference I draw from the totality of the evidence is that multijurisdictional portfolio licences themselves are unlikely to have inherently anti-competitive effects and that a demand for a worldwide licence is not inherently likely to distort competition. It may be that a worldwide rate is demanded which is excessive but that is a matter related to the rate. It may also be that a given portfolio does not justify a worldwide licence but that is a point on the facts. Assuming

There was a suggestion that a worldwide licence might create a disincentive to challenge the validity of patents in other jurisdictions. A similar disincentive applies to any portfolio licence. It is a factor to take into account but not enough on its own to make a portfolio licence (worldwide or national) inherently anti-competitive.

So far the points have not been specific to the circumstances of UnwiredPlanet and Huawei. Turning to the specific circumstances, there are three aspects. The first is that UnwiredPlanet’s portfolio does not have patents in every state of the world. The second is that UnwiredPlanet is engaged in litigation in this country. Its stance is that if it is entitled to insist on a worldwide licence then it will do so and if Huawei refuses, an injunction should follow. The third is that UnwiredPlanet is also engaged in litigation with Huawei in Germany and China on patents within the portfolio but in the licence UnwiredPlanet is demanding that Huawei should pay royalties for Germany and China.

The first aspect of the circumstances sounds more significant than it is. Just like UnwiredPlanet, neither Ericsson nor Huawei have patents in every state. In terms of geographical coverage UnwiredPlanet’s declared SEP portfolio covers most of Europe,

In addition to the well-rehearsed issue on China, there are two further aspects which Huawei draw attention to. First, UnwiredPlanet’s coverage of 3G/UMTS and 2G/GSM is much weaker than for 4G/LTE, as Mr Saru accepted. However, South America and South East Asia include jurisdictions in which Huawei makes a very considerable volume of sales, especially of 3G/UMTS and 2G/GSM equipment, in which UnwiredPlanet have no relevant coverage. Also for countries where Huawei sells only single mode handsets rather than multimode, there are no grounds on UnwiredPlanet’s case for it to be receiving any royalties going forward at all. Second, Huawei contend they manufacture handsets in Venezuela, in which UnwiredPlanet have no patents.

UnwiredPlanet’s response was to point out that one needs to consider manufacturing as well as sales. A handset sold in a country in which there is no patent may still have been made in a patented country in which a licence was required. The fee for the licence, which includes licence to manufacture, may well be a royalty calculated by reference to the sale price, with a royalty bearing event being sale. That this would apply whether the sale happens to be made in a patented or non-patented country is not necessarily a problem given that manufacture needed to be licensed.

In practice, subject to Venezuela, the country of manufacture referred to is China. Huawei did not challenge the principle but submitted a focus on China brings back the debate about the ongoing Chinese invalidity proceedings. As to Venezuela, UnwiredPlanet pointed out that Mr Zhang’s evidence was that the manufacturing facility there puts together components made in China.

Where UnwiredPlanet’s coverage differs from that of Ericsson or Huawei is in numbers of patents. UnwiredPlanet have many fewer patents in general and Mr Zhang and Mr Cheng both referred to the small size of UnwiredPlanet’s portfolio. The portfolio is much smaller than that of the biggest players like Huawei, Samsung and Ericsson, but in my judgment it is not so small as to be a portfolio which can or would be treated in a different way from some of the larger portfolios. The portfolio is large enough that it would be impractical to fight over every patent and neither party is doing this. The explanation of coverage above has been given by reference to declared SEPs since they represent the property which would be licensed, even though it is the much smaller number of Relevant SEPs which determines the royalty rate. The small size of UnwiredPlanet’s portfolio and the low number of Relevant SEPs is reflected in the FRAND royalty rates.

Before turning to the impact of the litigation, this is a convenient point to ask what sort of licence for UnwiredPlanet’s portfolio would be FRAND in terms of its geographical scope when applied to a multinational licensee like Huawei? I will start by asking what a willing licensor and a willing licensee with more or less global sales would do. There is only one answer. UnwiredPlanet’s portfolio today is (and in 2014 it was) sufficiently large and has sufficiently wide geographical scope that a licensor and licensee acting reasonably and on a willing basis would agree on a worldwide licence. They would regard country by country licensing as madness. A worldwide licence would be far more efficient. It might well have different rates for different regions and for different standards but that is another matter. The employment of different rates would not lead the parties to abandon a worldwide licence and go for country by country licensing. Assuming the licensee was a Chinese multinational like Huawei, they might well agree on different rates for China as for the Rest of the World but again they would not go for country by country licensing. If the multinational had a significant manufacturing base in another country in which the portfolio was weak, again that could be taken into account.

A point arose in the arguments on the terms of the UK only licence which Huawei called a manoeuvre. UnwiredPlanet insisted that the UK only licence should have a term in it precluding entry into the UK of unlicensed Huawei handsets. This seems to have caught Huawei by surprise but it really should not have done. It was a manifestation of the point that the UK only licence is only a licence under the UK patents. It illustrates one reason why country by country licensing is inefficient for goods like mobile telecommunications devices which will move across borders but I do not regard this as a major point in the present context. It would have to be addressed but that could be done (see below the section on the UK only licence where it is addressed in context). The real inefficiency of country by country licensing is the effort required to negotiate and agree so many different licences and then to keep track of so many different royalty calculations and payments. No rational business would do this if it could be avoided.

What a willing licensor and willing licensee would do is not the only question because necessarily if the result would amount to unlawful bundling then it is not FRAND. I turn to consider the four-part test in Microsoft.

The first part of the four-part test in Microsoft is that the products are separate. A patent in one state is clearly separate from a member of the same patent family in another state. They may have an entirely different scope. However the way in which FRAND royalty rates are determined in practice treats these two distinct patents as a single thing. It is common ground that the industry assesses patent families rather than individual patents within the family. The FRAND obligation actually applies to patent families, for good reason. It would be wrong to ignore this fact about how the licences are negotiated and agreed in practice. Assessing portfolios on a family basis inevitably involves tying in a patent in one state with a patent in another.

The second part (dominance) is assumed at this stage.

The third part is the choice. Here UnwiredPlanet wish to give the licensee no choice but accept the worldwide licence. This insistence is clearly a form of tying in. However again there is more to be said about this. Take a SEP with family members in two countries. If a multinational implementer sells in those two countries and needs a licence in one of them, the implementer probably needs a licence in both, hence the efficiency of this sort of licensing. Of course portfolio licensing is more complicated still because patents in some states may not have counterparts in others, but for this purpose they are all SEPs.

The fourth part is competitive foreclosure. Take the same example of a SEP with family members in two countries A and B and a multinational implementer selling in both. If the implementer does need a licence in both then the licensor is the only source of that licence. Tying in a licence under a SEP in state A with a licence under the SEP in the same family in state B does not preclude a third party from competing with the licensor in state B because the licensor is the only source of licences for this SEP in state B. This is an aspect of the dominant position of the licensor. It also helps explain the efficiency of worldwide portfolio licensing.

Given the prevalence of worldwide licences and the prevalence of assessment based on patent families, I am not prepared to assume that the tying of a SEP licence in one country to a SEP licence in another country has by its nature a competitive foreclosure effect. A close analysis of the actual effects would be required and that has not been done.

Huawei submit that the Motorola decision of the European Commission is in their favour. In that case Motorola sued Apple in Germany for infringement of a SEP for GPRS called “Cudak”. Apple made various “Orange Book” offers (the case was decided before Huawei v ZTE – see the analysis of that case below). The Commission decided that Apple’s second Orange Book offer indicated a willingness to take a licence on FRAND terms and so after that Motorola’s actions in pursuing its claims for a patent injunction were abuses of dominance (see paragraphs 125, 127, 301-303, 419, 423, 433, 434-436 and 440). Huawei point out that one of the issues was that the second Orange Book offer was to take a portfolio licence for Germany (para 125(a)) and that one of Motorola’s complaints was that they (Motorola) wanted a worldwide licence (127(f)).

I agree with Huawei that here the Commission are rejecting Motorola’s complaint that the offer was limited to Germany. The Commission’s decision is firmly in Huawei’s favour. However the decision is premised on the view that a licence limited to Germany is FRAND. Making the sensible assumption that the GPRS “Cudak” SEP had counterparts outside Germany, that decision can only be on one of two bases. Either national portfolio licences are FRAND and worldwide licences are not FRAND, in which case the majority of real licences are not FRAND because they are worldwide; or both national and worldwide licences are FRAND at the same time, in which case FRAND cannot be enforced save by imposing one kind of FRAND licence even if another more realistic kind is also FRAND. I have already addressed these issues above and the answer I have arrived at is that there is only one set of FRAND terms in a given situation. The Commission’s decision does not look at the matter in that way and, although they find in clear terms at the end of the quoted paragraph that Apple is offering to enter into a licensing agreement on FRAND terms, the Commission do not analyse the national/worldwide scope issue in the manner I believe is required. I will not follow this aspect of the Commission’s reasoning.

Huawei took a series of eight further points in support of the submission that UnwiredPlanet’s insistence on a multijurisdictional bundled licence in the present case was not FRAND. I will address the first four points and the sixth point. I have already addressed the fifth and eighth points, about the true geographical scope of Unwired

First Huawei submitted that any consideration must start with the proposition that at least in English law, there is no such thing as a portfolio right. Citing Lucasfilm Ltd v Ainsworth [2012] 1 AC 208 Huawei submitted that the English Courts have no jurisdiction to entertain an action in respect of the validity of a foreign patent, and are reluctant to entertain an action for infringement of a foreign patent even where validity is not in issue. Therefore, insofar as UnwiredPlanet wishes to complain that Huawei is infringing SEPs in other jurisdictions, Huawei contend that such complaints are in principle for the Courts of those other jurisdictions.

I accept that there is no such thing in law as a portfolio right. At least from the perspective of English law, UnwiredPlanet should sue on SEPs in the countries in which they exist. However this does not preclude a finding that worldwide licences are FRAND. The first submission is relevant but not determinative.

Huawei’s second and third points are best addressed together. In the second point Huawei submitted that it follows that there is a fundamental difference of principle between the bundling of all rights enjoyed within a given jurisdiction and the bundling of rights across different jurisdictions. In the latter case, there is an obvious risk that the threat of a territorial injunction may function, unfairly and unreasonably, in effect to reverse or at least negate the impact of foreign proceedings in respect of foreign patent rights. In the third point, Huawei submit that the facts of the present case are a practical manifestation of the concerns expressed by me in Vringo, referring to both [2013] EWHC 1591 (Pat) and [2015] EWHC 214 (Pat). Huawei reminded me that I said that proper scope of the issues to be adjudicated by the Court is “…not simply a matter of case management and consideration of the court’s resources” (para 58), but rather arises from the very “character of …underlying rights” on which a patent infringement action is based. Huawei referred to the following passages from the earlier judgment about the position of ZTE in that case:

“44 … ZTE has said it is willing to take a FRAND licence on any patent found valid and infringed. In my judgment, a defendant accused of patent infringement by a patentee who claims to have a standards essential patent is and must be entitled to say, “I wish to know if this patent is valid or infringed or not before I take a licence”. Such a stance cannot fairly be described as unwillingness.

45

So here the defendant is entitled, in my judgment, to adopt a contingent position. In a contingent case like this, there is no basis on which the court could compel the defendants to accept a licence arrived at by approaching the matter as if the licensee was willing to take a licence without having a judicial determination of validity and/or infringement.

46

Looking at it the other way, if, once the patent trials are heard, for example, say Vringo's alleged SEPs were found invalid, it would be absurd for Vringo to say it still wanted to have a trial to determine a FRAND royalty rate applicable in the United Kingdom. The rate would be zero. Equally, say Vringo won all the patent trials hands down and then conducted a FRAND trial, it would equally be absurd for ZTE at that trial to say, “Oh but these are weak patents likely to be invalid or not infringed and the royalty should correspondingly be less”.”

[Huawei’s emphasis]

Huawei then referred to the further decision in which I observed as follows (at 107109):

“…I can see that the aggregate of individual FRAND rates for patents taken alone and on a territorial basis may well be far more than global portfolio rates and so a rational defendant may well prefer to take a global portfolio licence rather than a series of individual ones. Moreover I accept, as Vringo urges on me, that global portfolio licences are the kinds of licences industry normally enters into.

However this is very different from saying that somehow the fact that a global licence on a portfolio of patents is FRAND necessarily means that a defendant in one jurisdiction faced with one patent is forced to take a global portfolio licence in order to stave off a national injunction on that one patent.

I could see a very different circumstance if Vringo had made a FRAND offer for the patent EP 1,212,919 (UK) itself and that offer had not been accepted. Then an injunction might well follow. In that sort of case, unlike the one based on the global portfolio licence, the threat of the injunction, which is after all a territorial remedy, would not be being used to create some sort of international coercion or coercion about other patent rights.”

[Huawei’s emphasis]

As Huawei also pointed out, I repeated the same views about “international coercion or coercion about other patent rights” in my judgment from the March 2015 CMC in these proceedings (para 23-27). Huawei submit this is exactly this type of “coercion” which UnwiredPlanet is seeking to practise in the present case. It is the same concern which is expressed by the European Commission in Motorola at paragraph 434.

What I called the Vringo problem has already been addressed above but I have set out Huawei’s submissions here at length because the point is important. There are a number of things I wish to say about this.

The earlier Vringo judgment is not dealing with the same problem which confronts the court now. Today the CJEU’s judgment in Huawei v ZTE, makes clear that an implementer can challenge a patent even after a licence has been concluded. What is more I can only say that having heard this trial, what seemed clear to me then does not seem to me now to be as clear cut. I remain of the view however that UK court’s power to grant an injunction against an implementer will only arise once at least two conditions are satisfied - there must be a FRAND licence available for the licensee to accept (or subject to case management some proper means of settling one) and there must be a SEP held by the court to be valid and essential. Only an implementer who refuses to take a FRAND licence will be at risk of an injunction.

The second Vringo judgment raises the Vringo problem which has been addressed already. The essential difficulty is that if it is possible that both kinds of licence – UK only and worldwide – are FRAND in the circumstances of this case then FRAND cannot be enforced at all. UnwiredPlanet submitted that a counterpart to the international coercion identified in Vringo was a similar kind of international coercion which was taking place in this case because the act of granting an injunction is no more an act of ‘coercion’ than the act of not granting an injunction. The FRAND commitment is a restriction on a patentee’s exercise of its patent rights and if, properly construed, it is a restriction that only requires a worldwide offer, and Huawei refuses or frustrates that, then not granting an injunction has the effect of compelling the patentee to either give up payment in other jurisdictions or accept a much lower rate because it cannot practically sue separately around the world.

I do not entirely accept UnwiredPlanet’s submission, since in the end the only rights a patentee has are national patent rights and the only places in which they can be enforced

Any licence involves elements of risk. A licence for a period of years with an unchanging rate involves a risk to one side or the other as the market changes over time. Similar kinds of potential risk exist in a licence for a whole national portfolio as opposed to a single patent and the same applies again for a worldwide portfolio. There is necessarily a difference in scope and therefore risk between a national injunction on a single patent, possibly directed to a single release of a standard, on the one hand and a licence under a portfolio whether it is national or international on the other hand. There may be differences in risk and scope for other reasons too. However all these distinctions are differences in degree, not differences in kind.

Huawei referred to evidence of Mr Zhang on this but his evidence was directed at a global blended royalty rate, which is a different point. Also since his evidence was given I am told Huawei’s challenge to one UnwiredPlanet patent in China has been rejected although Huawei have appealed.

The fourth point taken by Huawei is this approach is contrary to the Brussels I Regulation (Art 22(4)) as would amount to the grant of a UK injunction because a defendant had not taken a licence under a different national instance of an EP patent in a different Member State of the EU, contrary to the CJEU in Case C-4/03 GAT v LUK [2006] ECR I-6509, Case C-539/03 Roche v Primus [2006] ECR I-6535 and Case C616/10 Solvay v Honeywell (12 July 2012, Grand Chamber). There are existing infringement and validity actions in, for example, Germany and Huawei submit that the English court should not make a final determination that they are required to take and pay for a licence in Germany and issue injunctive relief if Huawei do not do so, when that issue is currently before the German courts. To do so, Huawei say, would amount to circumventing the jurisdictional rules laid down by Art 22(4).

Art 22(4) provides that in proceedings concerning the registration or validity of patents, the courts of the Member State in which the registration has been applied for shall have exclusive jurisdiction. In GAT v LUK, the CJEU held that Art 22(4)’s predecessor, Art 16(4) Brussels Convention, applied whatever the form of proceedings in which the issue of a patent’s validity is raised, including, for example, if it were challenged as a defence or counterclaim to a claim for infringement. The CJEU’s reasoning had three parts: (a) first, it held that to decide otherwise would undermine the binding nature of Art 16(4) by allowing a party to circumvent the rules and have questions of validity determined in the wrong jurisdiction (b) second, this would undermine the predictability and certainty of the rules of jurisdiction laid down by the Convention and (c) third, it considered that to allow courts other than those of a State in which a particular patent is issued to rule indirectly on the validity of a patent would multiply the risk of conflicting decisions across jurisdictions. This reasoning was followed in the two other cases that Huawei rely on: Roche v Primus and Solvay v Honeywell.

In my judgment, the Brussels I Regulation and the CJEU case law cited has nothing to do with what the terms of a FRAND licence should be. If a worldwide licence is FRAND then requiring Huawei to take and pay for one would not amount to determining questions of validity in relation to which courts of other Member States have exclusive jurisdiction under Art 22(4). Taking Huawei’s example of the on-going German proceedings, the German courts would remain free to determine the relevant patents’ validity. A FRAND licence should not prevent a licensee from challenging validity or essentiality of licensed patents and should have provisions dealing with sales in non-patent countries. So if the German courts decide all the relevant patents are invalid (or not essential), that would simply result in whatever consequences the worldwide licence provided for. Since the licence is a FRAND licence those consequences are FRAND too. The binding nature and clarity of Art 22(4) are not thereby undermined and, most importantly, there is no risk of the decisions in England and Germany conflicting. I reject Huawei’s fourth point.

The sixth point is about the conduct of the litigation. The key issue is a complaint that UK only licensing was not offered (on an open basis) by UnwiredPlanet for a year after the litigation began. However this rings hollow since Huawei have never (on an open basis) offered to take a worldwide licence at all.

Does the litigation in this jurisdiction make any difference to that conclusion? In my judgment it does not. A different question is whether, following Huawei v ZTE, UnwiredPlanet’s approach to the litigation and injunctive relief is premature and an abuse of dominance. At this stage I am only concerned with the mere fact that UnwiredPlanet and Huawei are engaged in patent litigation and insisting on a worldwide licence. Given that a worldwide licence is FRAND, the fact that litigation is in existence does not seem to me to make an insistence on it abusive or not FRAND.

The final point is to assess the impact of the outcomes of the litigation in Germany and China. In Germany the proceedings are bifurcated. UnwiredPlanet have sued Huawei for infringement in the local courts while Huawei have brought invalidity proceedings in the Federal Patent Court. In China Huawei have brought invalidity proceedings. So far there have been wins, losses and appeals on both sides and no final outcome, in the sense of final without appeals.

In my judgment this other litigation does not make any difference to the analysis. As I have already said in the context of Huawei’s Brussels I and GAT v LUK argument a worldwide FRAND licence should include an appropriate mechanism to deal with countries which become effectively non-patent countries.

I conclude that a worldwide licence would not be contrary to competition law. Willing and reasonable parties would agree on a worldwide licence. It is the FRAND licence for a portfolio like UnwiredPlanet’s and an implementer like Huawei. Therefore, UnwiredPlanet are entitled to insist on it. It follows that an insistence by Huawei on a licence with a UK only scope is not FRAND.

Should the court settle the FRAND terms of a worldwide licence?

All of UnwiredPlanet’s worldwide offers since 2014 have been put as licences with single global blended royalty rates. That needs to be addressed along with any other relevant terms.

Huawei did not engage with UnwiredPlanet’s proposed terms of a worldwide licence because they contended it was not open to UnwiredPlanet to insist on it both in principle (i.e. the FRAND and competition law bundling points I have rejected) and from a procedural point of view. Huawei’s case on the procedure is that a UK only licence is the inevitable and mandatory outcome of this case as a consequence of an earlier case management ruling and certain subsequent procedural steps.

In summary Huawei’s procedural point is this. During these proceedings UnwiredPlanet have offered licences with three different scopes – individual patents, UK portfolio and worldwide. In the summer of 2016 Huawei informed UnwiredPlanet that they would accept a UK portfolio licence and would accept whatever rate the court decides is FRAND for such a licence. Subject to the manoeuvre, Huawei would also accept whatever terms the court thinks fit for a UK only licence. Since a UK portfolio licence is one of UnwiredPlanet’s offers and since Huawei have accepted any rate the court thinks fit for that licence, the only licence which the court can settle is a UK portfolio licence.

I am not impressed with the procedural submission, for the following reasons.

From the beginning Huawei have contended that UnwiredPlanet’s offers such as the 2014 offer were not FRAND and contrary to competition law and sought certain remedies accordingly, such as refusal of an injunction. In one sense such an issue is narrow in that one could simply decide whether an offer is or is not FRAND etc., and if it is not, go no further. A difficulty could be that the offers concerned were never put together with fully worked out contract terms but that might not matter in practice since no doubt the big issue of the rate could be addressed and, on Huawei’s case, certain terms such as bundling SEPs and non-SEPs together were such that they could never be FRAND.

UnwiredPlanet applied to amend to seek certain declaratory relief. This came before me at the case management conference on 19^th-20^th March 2015 (judgment on 24^th April 2015 [2015] EWHC 1029 (Pat)). UnwiredPlanet sought leave to amend its Statement of Case to include as relief declarations about the FRAND status of their offers to date. I allowed one declaration but refused the other. The declaration which was refused was a device to simply ask the question – what would be FRAND? It was not anchored to anything. In refusing this wide declaration I referred to the court not having the sort of open ended jurisdiction available in the Copyright Tribunal. I allowed UnwiredPlanet to claim a declaration in narrower form which referred expressly to specific offers already made. In allowing this amended claim I held that the court could declare that a given set of terms were or were not FRAND and also that the court could, within the framework of a concrete proposal, find that a set of terms which differed in some respect from the concrete proposal was also FRAND even if the terms as proposed were not. I also rejected the objection to allowing this amendment that it would require the court to value UnwiredPlanet’s global patent portfolio. That was because that issue was going to be in the case anyway.

Huawei rely on this judgment, but it is against them. A key step in Huawei’s reasoning is the idea that it is only because Huawei have agreed to accept whatever rate the court thinks fit that the court has jurisdiction to set a rate rather than just adjudicate in a black and white fashion on rates suggested by the parties. I doubt such a thing was ever appropriate but in any event the case management judgment made clear that in dealing with the declaratory relief permitted by the amendment, the court would not be constrained in that way. It is not Huawei’s acceptance that it will take whatever rate the court thinks fit which gives the court jurisdiction to decide what the FRAND rate should be. The amendment permitted in April 2015 does that.

Huawei have always known that UnwiredPlanet ranked their offers in the sense that their first preference was for a worldwide licence. Huawei do not suggest that UnwiredPlanet ever indicated that if Huawei said they would take a UK portfolio licence then the issue of scope was resolved. The case is not put on that basis. Huawei referred to the odd aspect of Mr Bezant’s evidence but it does not bear out the argument. The oddity of Mr Bezant’s approach is that it applies just as much as setting a UK only rate as anything else.

It is true that fully worked out terms were only put forward openly after the start of the trial, as a result of me asking the parties to produce and engage with them. That cannot make a material difference. If Huawei thought they were prejudiced by the absence of such terms earlier in the action they could have asked. But I am left with a set of worldwide terms proposed by UnwiredPlanet and no engagement with Huawei. In other circumstances it might have been appropriate to hand down a judgment dealing with the big issues, such as rates, and then leave it to the parties to agree the smaller points or isolate the differences down to a few points. That is often done in the Copyright Tribunal. However this issue was discussed extensively during the trial. This trial is meant to be the point at which the court decides whether to grant or refuse a final injunction against Huawei following findings of infringement of valid claims over a year ago. The only way that can be done fairly is if at the point the injunction would bite a fully worked out licence is available in relation to which the court has decided FRAND, competition law, and equitably refusability. These proceedings have gone on long enough. There is no justification in allowing yet another round of evidence and argument and so I will do the best I can with the material.

FRAND rates in a worldwide licence

The evidence before me is clear that willing parties would agree that a worldwide licence in this case would have a different rate for sales in China. They would not have a carve out which left China unlicensed ([…]) because UnwiredPlanet’s portfolio of declared SEPs clearly includes China. The rate is determined by the number of Relevant SEPs but a licence is a licence on declared SEPs.

The appropriate rate for China is not complicated to arrive at. The comparable licences show that rates are often lower in China than for the rest of the world. The relative factor varies. I find that a FRAND licence would use a factor of 50%.

However that is not the end of the story. UnwiredPlanet’s portfolio is smaller in China than in elsewhere and UnwiredPlanet have fewer Relevant SEPs there than the numbers used to set the benchmark rates above. However this can readily be catered for. A fair and reasonable approach consistent with everything which has gone before would be to