Mitsubishi Electric Corporation v Oneplus Technology (Shenzhen) Co, Ltd & Ors

This judgment is organised as follows:

This is my judgment following the first technical trial in these proceedings in which the Claimants seek to persuade the two sets of Defendants who remain to take licences of a pool of patents (‘the MCP Pool’) administered by the Second Claimant which are alleged to be essential to one or more telecoms standards. Originally the first technical trial was due to take place in December 2020 involving a different patent but that trial was settled and, as I understand it, the patent in question will play no further part in this

EP142 is alleged to be essential to version 10.0.0 and all subsequent versions of TS36.322 of the 4^th generation 3GPP Long-Term Evolution (“LTE’) standard.

This case is about the signalling of structure in a digital signal. The digital signals in question are those created and received in 3G and 4G mobile telephone networks in level 2 of the OSI 7 layer model. The case turns on the meaning and scope of the term ‘length indicator’ in claim 1 of EP142.

Although the preferred embodiments in EP142 are written in the context of UMTS/3G, it is Sisvel’s case, as mentioned above, that EP142 is essential to the LTE/4G standard. The Defendants deny essentiality, thereby denying infringement, but also run a series of squeeze arguments to the effect that if EP142 is essential to LTE, then EP142 is invalid on a variety of grounds.

Sisvel makes an unconditional application to amend by way of two requests – Request 1 and Request 2 – which Sisvel advances in the alternative. Sisvel says it makes the application to amend to forestall any argument that the claimed monopoly covers a regime that indicates only if a single SDU is a perfect fit, a point which I return to later.

EP142 is entitled ‘Radio Link Control Unacknowledged Mode Header Optimization’ and has a priority date of 23 August 2005. As is often the case, the key to this action lies in assessing the scope of the claim at the correct level of generality and in the correct context. As is usual in patent cases, this judgment needs to set out a good deal of information in order to define the correct context for the claimed RLC UM header optimization, in the course of which the reader will have to pick up the numerous acronyms which encrust this art.

Mr Adrian Speck QC led Mr Michael Conway for Sisvel and Mr Andrew

Sisvel’s expert witness was Dr Alastair Brydon and Mr Claude Royer was the expert called on behalf of the Defendants. Mr Royer gave his evidence from Canada, so certain of the trial days were set later to accommodate his time zone. In closing, such criticisms as were levelled at the experts were slight, perhaps because each side had extracted what they perceived they needed in cross-examination. Both experts gave their evidence in a straightforward and direct manner. There were times when each of them had a tendency to stick to the party line, slightly more so in the case of Mr Royer.

The experts were agreed that the Patent would be of interest to a person or team working on the implementation of and/or development of telecommunications standards, with a focus on Layer 2 (the data link layer) including the RLC sub-layer. They also agreed that the person or team would be an engineer with typically a higher degree and/or several years’ experience in industry.

There was, however, a mini-dispute as to the size and breadth of expertise of the Skilled Person or team, in particular as to their knowledge and/or experience with other wireless communications standards than UMTS. This dispute was concerned, as I understand it, primarily with the WiMAX prior art. It is not necessary for me to resolve that dispute since it does not affect anything I have to decide.

At the PTR, I ordered the parties to produce a document setting out the agreed CGK together with a list of any CGK points in dispute. This they did and I am grateful for the work done on both sides to produce a very useful document.

What follows in paragraphs 14-65 below is the CGK as agreed between the parties, with some slight editing of my own. It includes references to some developments after the Priority Date (which were obviously not CGK) in order to illustrate the position as at August 2005. I add some further points of CGK at the end of this section.

The architecture of wireless communications networks, including UMTS, is typically described as a "stack" of protocol "layers". Each layer has a specific set of functions and each relies on other layers to perform its function.

The Open Systems Interconnection ("OSI") model was developed by the International Standards Organisation. It describes the following seven layers that are typically present in a communications network in some form.

Generally, the higher layers (5, 6 and 7) are primarily concerned with interacting with the user and implementing the applications that run over the network. The lower layers (1, 2, 3 and 4) are primarily concerned with the formatting, encoding and transmission of data over the network.

Further details of the functionality of layers 1, 2 and 3 are described in the context of UMTS below.

In packet switching, there is no dedicated communication path between the sending and receiving device. Instead, data is broken down into packets and transmitted between network links that are shared by multiple competing communication sessions. Each packet has information that specifies the packet’s destination and how the packet will be reassembled into a whole message when received at its destination. Packet switching is more efficient for communicating some types of data traffic because it allows many users to equally and flexibly share the same set of bandwidth resources. However, a drawback of packet switching is that when the network is busy packets can be delayed. This delay is termed "latency". Too much latency results in poor service quality for certain applications, particularly where real-time interaction is required, for example in voice or video calling.

In third generation ("3G") networks, the information to be transmitted is divided into a voice channel for voice calls and a data channel for other data to be transmitted over the network. The voice channel operates through circuit switching (i.e. a dedicated connection, like a traditional landline call), while the data channel utilises packet switching (in which no dedicated connection is established). The data channel can also be used to carry voice in the case of "voice over IP" or "VoIP".

Universal Mobile Telecommunication System ("UMTS") is a "third generation" ("3G") cellular communication system based on Wideband Code Division Multiple Access ("WCDMA") radio technology. It was first specified in 3GPP Release 99, which was completed at the end of 1999, and commercial UMTS networks were launched from 2001 onwards.

The Release 99 UMTS system provided voice telephony and basic multimedia services, but significant improvements to the packet switched data service capabilities were enabled by High Speed Downlink Packet Access ("HSDPA") in 3GPP Release 5, completed in September 2002, and High Speed Uplink Packet Access ("HSUPA") in 3GPP Release 6, completed in September 2005. Among other developments, the increased data rates provided by these enhancements enabled the introduction of Voice over Internet Protocol ("VoIP") services.

UMTS network architecture comprises a Core Network, the UMTS Terrestrial Radio Access Network ("UTRAN"), and User Equipment ("UEs") such as mobile telephones, as shown:

The Core Network manages the switching and routing of calls and data services to their destinations; the UTRAN is responsible for the radio-related functions of the network and provides connectivity between the Core Network and UEs.

Within the UTRAN, the NodeBs are the radio base stations, each of which may support one or more radio cells. The Radio Network Controllers ("RNCs") are the main control nodes. They manage the resources of the NodeBs connected to them and implement

A simplified diagram of the protocol architecture in UMTS is below.

The Control Plane ("C-Plane") represents signalling information from higher layers. The User-Plane ("U-plane") represents user data.

The solid lines represent communication paths operating between the protocol layers. These can be considered as logical communication "pipes" offered by each layer to the layer below which is used to communicate information between peer entities within the network. A variety of these "pipes" is available at each level in the protocol stack, providing a range of communication characteristics suited to different service requirements. They are variously referred to as "radio bearers", "logical channels", "transport channels" and "physical channels" at different points in the stack, as shown.

Within UMTS, the following layers are involved in the interface between the UEs and the UTRAN:

Control plane signalling and user information to be transmitted flows into Layer 2 from the higher layers before being processed ultimately for transmission over the radio interface at the PHY. Received signals are initially decoded and processed at the PHY before being passed upwards through the stack to the relevant higher layers.

On the network side, the MAC and RLC functions are implemented at the RNC. As the RNC is geographically remote from the base station, the MAC scheduling performed at the RNC cannot instantaneously adapt to changing radio channel conditions between

At the transmitter, the RLC receives data packets known as RLC Service Data Units ("SDUs") and passes on data packets to the MAC, referred to as RLC PDUs. At the receiver, the opposite process takes place, as shown below:

RLC is responsible for packaging variable sized data packets from higher layers (RLC SDUs) into RLC PDUs for onward transmission to the MAC.

The MAC is responsible for scheduling received RLC PDUs over transport channels to the PHY, at regular TTIs. To do so the MAC selects a transport block size (which determines the RLC PDU size) and an associated transport format. There is a limited set of different transport block sizes available to the MAC, depending on how the transport channel was semi-statically configured, and the combinations of concurrent transmissions scheduled in a time interval.

For each TTI the MAC layer indicates to each RLC transmitting entity how many PDUs to prepare and what size they may be, taking into account the amount of data waiting in the corresponding data buffer, its priority relative to other data, and the overall requirements on the radio interface in that TTI. The set of RLC PDU sizes is limited; by virtue of signalling restrictions, only N sizes are available for the MAC for a given bearer, which are semi statically defined during bearer configuration. Limiting scheduler complexity would also bring constraints in the size of the set of PDU sizes to consider at each TTI. For example, in UMTS Release 5 ("HSDPA") the maximum allowed set of PDU sizes is eight.

The RLC entity then applies segmentation, concatenation and padding to pack the SDUs it has received into RLC PDUs for onward transmission to the MAC.

The way in which segmentation, concatenation and/or padding is used depends on the operating mode of the RLC entity in question (see below).

The RLC protocol defines three types of RLC entities characterised by their modes of operation: Transparent Mode ("TM"), Unacknowledged Mode ("UM"), and Acknowledged Mode ("AM").

TM is the simplest mode. No header information is added to RLC PDUs. TM mode can buffer data and perform simple segmentation and concatenation of SDUs, but only to a limited extent due to the lack of header information. TM incurs minimal overhead and delay, but the receiver cannot know if data packets have been received correctly. TM is suitable for circuit-switched voice telephony and can be used for transmission of short messages such as those sent over the Paging Control CHannel ("PCCH").

UM adds header information to each RLC PDU. This includes a sequence number and further information to enable correct reassembly of SDUs. This facilitates more flexible packaging of SDUs by segmentation, concatenation and padding (see below). In UM there is no acknowledgment of receipt of PDUs by the receiver. UM is suitable for services where there is some limited tolerance for data errors but a need for relatively low delay, such as conversational streaming media (including VoIP).

AM provides segmentation, concatenation and padding in broadly the same way as UM. However, it also provides an error correction mechanism, whereby an AM receiver can request retransmission of PDUs that have been received with errors or not received at all. This provides reliable data transfer, but retransmissions result in delay. AM is suitable for services such as email, file transfer, and web browsing.

In UM and AM modes, the RLC PDU header provides the receiver with information to extract segmented and/or concatenated SDUs from the received PDU.

The UM PDU is defined in TS 25.322 section 9.2.1.3. Figure 5 shows its structure.

The length of a UM PDU is always a multiple of 8 bits (commonly referred to as an "octet" or "byte"). Figure 5 shows the series of octets in a PDU from the first octet, "Oct1", at the top to the "Last Octet" at the bottom.

In UMTS the first octet of the header is present in all PDUs. The first octet of the PDU contains a Sequence Number ("SN") and an Extension Bit ("E-bit"). This is followed by zero, one or more Length Indicators ("LI") each with a further E-bit, followed by Data, which may contain one or more SDUs or SDU segments. Padding ("PAD") may also be added. It is also possible that a PDU may contain only padding after the header. Headers contain 'overhead' data meaning that they provide information that facilitates the transmission of payload (user) data, but in doing so they use resources that could otherwise be allocated to payload data.

The SN is a 7-bit number, which increments for each PDU to identify its position in the overall sequence of PDUs. This helps the receiver to reassemble the SDUs and is used in ciphering operation.

In UM mode the E-bit can be configured to have one of two interpretations.

In the normal interpretation, introduced in Release 99, the E-bit is set to 0 to indicate that the next field is data, piggybacked STATUS PDU or padding (i.e. the header is not extended) or to 1 to indicate that the next field is an LI and E-bit (i.e. the header is extended).

An alternative interpretation was introduced in TS 25.322 v6.4.0 (June 2005). If the alternative interpretation is used, the E-bit in the first octet is still set to 1 to indicate that the next field is an LI and E-bit, but if the E-bit in the first octet is set to 0, it indicates that the next field is a complete SDU, which is not segmented, concatenated or padded.

The two definitions are summarised in TS 25.322 v6.4.0 as follows:

Which definition is used depends on the configuration set by higher layers.

The use of the alternative E-bit in a case where the PDU is exactly filled with a single complete SDU is shown below.

Section 9.2.2.8 of TS 25.322 v6.4.0 describes the use of Length Indicators as defined in the UMTS standard. The relevant section is set out in Annex 1, to which I have added certain annotations indicated in italics, which are explained below.

The Data field may contain one or more RLC SDUs and/or segments of SDUs, which may be concatenated to avoid unnecessary padding. The lengths of RLC SDUs are always a multiple of 8 bits.

Any unused space at the end of a PDU is referred to as padding. Padding may have any value and the receiver and sender disregard it. Padding is defined in TS 25.322 section 9.2.2.10.

VoIP refers to various technologies used to carry voice telephony (and other media) over Internet Protocol (IP) networks, such as the Internet.

First generation cellular networks were designed and deployed to offer a reliable mobile voice service. The circuit-switched voice techniques used in the cellular network at the time guaranteed quality of service ("QoS") by virtue of radio resource reservation for the duration of the call. This circuit-switch paradigm endured in the design of thirdgeneration systems, which optimised voice capacity through coded channel resources, again allocated for the duration of the calls.

By August 2005, communication equipment vendors and standard development organisations were driving the industry towards packetised voice techniques that offered more flexibility and lower equipment costs than circuit-switched technology.

VoIP services are carried by packet-switched connections. Voice signals are encoded as a series of data packets (e.g. every 20ms), each of which is given its own IP header information and routed independently through a packet-switched network. In contrast to circuit-switched voice, the VoIP packets allow to flexibly intermix the voice information with other types of data traffic, such as internet traffic. By foregoing the paradigm of reserved specialised radio resources for the whole duration of calls, VoIP can help optimise the sharing of the radio spectrum with data services.

In the early 2000s, there was increasing interest in extending VoIP to mobile networks, in order to achieve similar benefits. However, the voice capacity (i.e. the number of active calls at a given time) of VoIP was often lower than the voice capacity of circuit switched ("CS") voice over the UMTS systems available in 2005. The limited capacity and significant latency (delay) of packet-switched data services in mobile networks were constraints. The carriers were hesitant to abandon circuit-switched technology, as mobile cellular data services were still commercially insignificant relative to voice services.

In UMTS, the improved capacity, data rate and latency of HSDPA (defined in 2002) and HSUPA (completed in September 2005) were significant steps towards making VoIP a realistic service in UMTS. Network operators were interested in launching VoIP services based on a service architecture known as IP Multimedia System (IMS). However, it would take many years to see wide commercial adoption of Wireless VoIP and the 3GPP IMS voice framework (aka "VoIMS") in particular and it was not widely deployed as at August 2005.

Encoding voice signals into IP packets produces speech frame packets of generally equal size at equal time intervals.

The IP protocol adds relatively large headers to the speech packets. The PDCP layer (see above) uses an algorithm called Robust Header Compression ("ROHC") to compress the packet headers. This results in data packets of variable size, and in turn a stream of RLC SDUs that can have different sizes. The possible sizes are limited to a known set of sizes each having a different frequency of occurrence. The Skilled Person would be aware of this, without necessarily being aware of the exact relative frequency of occurrence of different size SDUs.

VoIP services require low delay but they are tolerant to occasional errors in transmission. The most appropriate choice for VoIP transmission in UMTS is RLC UM.

The concept for developing a successor to UMTS (to be marketed as Long Term Evolution ("LTE")) emerged at the end of 2004, with plans to develop an improved radio protocol based on Orthogonal Frequency Division Multiplexing ("OFDM").

Preliminary study items had begun under Release 7 in 2005, but specification work had not yet begun. The Skilled Person would have been aware that early concepts for LTE had been discussed and would have been following its development, but would know that specification work had not begun and that the technical details were not yet defined.

Although the following points were not set out as agreed CGK, I find that they also comprised part of the Skilled Team’s CGK:

In line with its title (set out above), the Patent defines the Field of the Invention in [0002] as relating to optimizing radio link control unacknowledged-mode protocol data unit headers, for example, to better support packet switched voice transmissions or transmission of other real time packet switched services over Wideband CDMA air interface.

In [0003] and Figure 1a, the Patent sets out the results of an experiment which measured the distribution of RLC SDU sizes for a 12.2 kbit/s adaptive multi rate (AMR) voice codec. It explains that the voice codec itself produces equal sized packets but a robust header compression (ROHC) produces variable size SDUs. Re-ordering the most significant sizes (i.e. all those above 1%) in Figure 1a gives the following exampledistribution which happens to reflect the eight PDU sizes which were permitted in UMTS Release 5:

[0003] suggests that to optimize the RLC overhead, the following RLC PDU sizes could be selected: 11, 15, 36, 40 and 98, with 36 and 11 being used for the most frequent RLC SDU sizes (the Skilled Person would understand this would entail the use of the alternative E-Bit), but it notes that there are quite significant amounts of RLC SDUs of 13 and 38 octets which are 2 octets smaller than the RLC PDU sizes of 15 and 40 octets. The Skilled Person would understand that this paragraph is talking about trying to match the RLC PDU sizes so they accommodate the variety of SDU sizes and relevant header information as efficiently as possible.

[0004] continues this focus on the ‘2 octets smaller’ situation:

Paragraphs [0005]-[0009] sit under the heading Summary of the Invention. It is only necessary to refer to paragraph [0005] which concerns the method. After correcting an error, paragraph 0005 reads as follows:

Both at the beginning (in [0011]) and the end (in [0053]) of the Detailed Description, the patentee is at pains to emphasise that, whilst the preferred embodiments are described in a WCDMA-type system (UMTS), the invention may be implemented in other systems ‘such as in 3.9G systems’. Thus, the patentee clearly signalled that the invention may be implemented in the forthcoming 4G system, depending on how it was configured.

Having made that point, the Detailed Description continues with a couple of paragraphs which set the scene, describing RLC, UM, AM and TM modes and their roles in transmissions between sender and receiver. I will quote [0014]-[0018] because they are important in understanding how the term ‘length indicator’ is used in EP142 (emphasis added):

[0019] is a long paragraph which is difficult to follow because it contains long sentences with many sub-clauses. Furthermore, as the experts eventually agreed, in two places

I will return to discuss [0024] a little later, due to the proposed amendments to the specification which seek to take the embodiments described in that paragraph outside the scope of the Patent.

[0026] and [0027] return to the ‘two octets smaller’ discussion and there is no dispute that these paragraphs describe embodiments of the invention.

In discussing the ‘two octets smaller’ situation, [0026] highlights two advantages resulting from the fact that the receiver has all the information required so that it does not need to wait for the next PDU and can send the data payload immediately up to the next layer. The first advantage is that the payload of this PDU is not affected if the next PDU is lost. The second is that delay is reduced because the payload can be delivered immediately without having to wait for the next PDU to arrive. [0027] draws attention to a further advantage: if the last SDU in the PDU is the last in a sequence, the sending of an additional PDU is avoided. These three advantages are equally applicable whether the PDU contains a single SDU or multiple complete SDUs.

The tables which follow [0027] are based on the tables in the standard but to aid the Skilled Person, the amendments to the standard are shown in underline.

It becomes clear that the rest of the detailed description describes three embodiments of the invention which differ only in the particular special length indicator value that is used to indicate both the start and end of an RLC SDU:

It is not necessary to set out all the figures. In general, each of figures 1-12 contains a comparison. On the left, in Column A, PDUs and SDUs of particular sizes are shown in a system without the benefit of the invention. Column B, on the right, shows what happens in a system with the invention. Figure 1 is captioned ‘Sequence of RLC UM PDUs showing the usage of LI, RLC PDU size of 39 octets and RLC SDU sizes of 38 or 37 octets’ and, with the arrows enhanced by Dr Brydon, looks like this:

In the top row, both Column A and B illustrate the use of the alternative E-bit, where E=0 indicates that the PDU contains a single complete SDU which is not segmented, concatenated or padded. In the second row, Column B shows the special length indicator 1111100 of the Patent indicating that the first data octet of the PDU is the first data octet of an SDU and the last octet of the PDU is the last octet of the same SDU (this is an example of the single SDU embodiment). The same is shown in the bottom row, Column B. These contrast with column A where for PDU SN+1, the receiver has to wait to receive PDU SN+2 to receive the indication (LI=0000000) that the SDU in PDU SN+1 ended in that PDU. As Dr Brydon explained, the use of LI=0000000, in conjunction with the absence of any padding indicator in the next octet, implicitly indicates that the first data octet of PDU SN+2 is the first data octet of an SDU. Although not shown in figure 1, the end of the SDU in PDU SN+2 has to be indicated once again in PDU SN+3.

Figure 2, which I need not set out, shows what happens when RLC PDUs of 39 octets carry SDUs of 36, 35 or 34 octets. In those scenarios, Columns A and B are the same, showing that the invention has no application.

Figure 3 shows essentially the same as Figure 1 where the special length indicator of the invention is LI=0000000. Figure 4 again shows that the special length indicator is not used, in the same scenarios as in Figure 2. Figure 5 shows what happens if PDU SN carries the last but complete SDU of a data burst. With the benefit of the invention, Column B features only one PDU (of 39 octets) carrying the complete SDU of 37 octets with the header carrying the SN, E=1, LI=0000000 and E=0, whereas Column A has to include a complete second PDU which contains LI=0000000 to indicate that the last octet of PDU SN was the last octet of an SDU, and then LI=1111111 to indicate that the rest of PDU SN+1 is padding.

A comparison of Figures 5 and 6 shows that when the patent is discussing, for example, the second embodiment, it is not describing a single system. Rather in that embodiment it is illustrating how the value 0000000 can be repurposed in different ways. In Figure 5 that value indicates that the PDU is exactly filled with a single SDU which starts and ends in the PDU whereas in Figure 6 that value provides a different indication. As it says in [0042] ‘Here, the special length indicator=0000000 indicates that one SDU starts and another ends in this PDU…’ There has been no suggestion that a single value in a system would be able to convey two different indications.

In more detail, Figure 6 shows RLC PDUs of 74 octets carrying two RLC SDUs. In Column B, all the information required to unpack PDU SN is included, whereas in Column A, the end of the last SDU has to be indicated in PDU SN+1. Figure 7 shows what happens when the first SDU (of 40 octets) is segmented between PDUs (of size 39 octets), followed by SDUs of 34 and 37 octets. PDUs SN and SN+1 are the same in both Column A and B. The benefit of the invention comes in PDU SN+2, where the special length indicator of the invention (in this second embodiment LI=0000000) indicates explicitly that the last SDU of 37 octets is completely carried by PDU SN+2 without padding (and implicitly indicates that the last octet of PDU SN+1 was the last octet of the last SDU in that PDU). By contrast, in column A, although PDU SN+2 carries a complete SDU of 37 octets, the receiver will have to wait for PDU SN+3 before receiving an indication that the last octet of that SDU was the last octet of PDU SN+2.

For the third embodiment, Figure 8 is equivalent to 1 and 3, Figure 9 is equivalent to 2 and 4, Figure 10 is equivalent to Figure 5, 11 to 6, and 12 to 7.

Overall, the figures demonstrate the benefits of the invention in particular situations but also that it has no application in others.

The claims divide into sets of method and apparatus claims, plus a ‘computer program’ claim. It suffices to deal only with the method claims.

Claim 1 as granted reads as follows:

Indications as to the possible breadth of claim 1 as granted are given by the additional limitations in subsidiary claims:

The patentee applies unconditionally to amend EP142. In the specification, it is proposed to insert a new [0018A] ‘Paragraphs [0019] to [0021] and [0024] below describe embodiments not in accordance with the claimed invention.’ Consistently with that, the words ‘an embodiment of the invention’ are deleted from the start of [0024].

For the claims there are two proposed requests. I will set out Request 1 as broken down into integers by the Claimants, in which Integer R1(e) is the proposed addition:

In Request 2, the amendments proposed to claim 1 are more extensive:

In both requests, original method claims 6, 7, 9, 10, 11, 12 are proposed to be deleted because, it is said, they cover ‘the single SDU embodiment’ i.e. where the indicator indicates that the first data octet of the PDU is the first octet of an SDU and the last octet of the PDU is the last octet of the same SDU. Corresponding changes are made to the apparatus claims. As far as I am aware, this is the only real explanation given for the proposed amendments: the patentee wishes to remove from the scope of the invention the so-called ‘single SDU embodiment’.

The Claimants submit that the combination of integers R1(b), (c) and (e) specify that the length indicator of the claim must be for indicating the case where the first data octet of the PDU is the first octet of an SDU and the last octet of the PDU is the last octet of an SDU for any number of SDUs in the PDU. Whether this submission is correct is a point I address later.

Although it is well established that the alleged infringement cannot be allowed to influence the interpretation of the claims in the Patent, in the circumstances of this case it is instructive to have the alleged infringement in mind in order to detect whether the alleged infringement is illegitimately influencing the arguments on claim scope.

The approach taken to RLC PDUs in LTE has a number of similarities but also some differences to the approach taken in UMTS. The most significant difference of all is that in LTE there is no constraint of a set of fixed PDU sizes for a given session. Instead, LTE uses fully flexible PDUs such that every PDU can be shrink fitted to the SDU payload.

With that point in mind, the other differences and similarities are best illustrated by reference to one of the basic structures of an RLC PDU in LTE in which a 5-bit sequence number is used. A 10-bit SN can also be configured, but the difference is not material.

The fixed part of the UMD PDU header is the first octet. It starts with the 2-bit Framing Indicator, FI field, followed by a single E-bit, and a 5-bit Sequence Number, SN. In the optional part of the header there then follows pairs of a single E-bit with a 11-bit Length Indicator LI. Because each LI is longer than 8 bits, each LI occupies more than one octet and if there are an odd number of LIs, 4 bits of padding must be included at

The LTE specification defines the FI field as follows:

It is common ground and/or not disputed but in any event I find that:

The Claimants’ infringement case relies on the FI=00 value when inserted into the FI field. The Claimants allege that:

I remind myself of the following principles:

I also bear in mind the following two citations. First, the reminder by Floyd LJ in Icescape at [97]:

Second, HHJ Hacon’s observation about the third question in Regen v Estar Medical Ltd [2019] EWHC 63 at [224]:

Although consideration of the inventive concept or the problem underlying the invention is directly relevant when considering the first Actavis question, in the circumstances of this case, I consider it is sensible for me to review the parties’ contentions on these points before I embark on the process of assessment of claim scope.

In this case a great deal of effort was expended on each side on trying to formulate and support a characterisation of the invention with an eye on both normal interpretation and equivalents. In this context, the arguments were directed to the claims as proposed to be amended. For the purposes of analysis I will assume that claim 1 as amended is limited to the situation where the first octet of the data part of the PDU is the first octet of an SDU and the last octet of the PDU is the last octet of an SDU which may be the same as the first SDU or a different SDU. I will consider later whether the proposed amendments actually succeed in achieving that limitation. Reverting then to the arguments, they evolved, with the principal steps being as follows.

For their part the Defendants characterised the invention as solving the ‘N+1 problem’ where N is the number of SDUs in the PDU, but the total size of the combined SDUs is N+1 octets smaller than the PDU. This concept emerged in Dr Royer’s first expert report. It seems to me that this is just another way of describing the ‘two octets smaller’ situation specifically addressed by the Patent. In [0004] the ‘two octets smaller’ problem to which the Patent was addressing itself involved a single SDU so that N=1, but, say, the SDU was 37 octets long and the PDU was 39 octets long. As I have mentioned, [0026] describes the benefits of the invention in the ‘two octets smaller’ situation for both a single SDU and multiple complete SDUs in the PDU.

The Claimants responded in four ways:

The final nuance came from the Defendants’ expert during his cross-examination on Day 4 when he characterised the invention as ‘the length indicator technique’. Although the Claimants poured scorn on this in their Closing Submissions, this was just another shorthand and, in my view, did not affect the arguments one way or the other.

It is convenient to deal with the X1-1 point first. Although I note that Dr Brydon did not give any evidence about X1-1, the illustration was put to Dr Royer. As I understand matters, the purpose of X1-1 was to counter the Defendants’ argument that a ‘length indicator’ (as that term is used in claim 1) could not be positioned in the header before the first E-bit i.e. in the fixed part of the header. The Defendants responded to X1-1 by contending it was artificial. It is not necessary for me to reach any concluded views about X1-1 because, as appears below, I have been able to form a view on interpretation which bypasses this particular dispute.

Leaving aside the X1-1 issue, all the other points seem to me to be different ways of characterising the same thing: that the benefits of the invention can be achieved in certain circumstances, depending on the available sizes of PDU.

I will say a little more about the Claimants’ first and second points.

The phrase ‘perfect packing’ is more than a little dangerous as a shorthand for the invention. As an expression, it covers more than just the invention:

The Claimants’ argument has validity in the sense that if you can adjust the sizes of the set of PDUs, there may be more opportunities to secure the ‘2 becomes 1’ saving and other benefits of the invention. This can be illustrated in two ways:

The specification makes it clear that the problem the Patent is addressing is the situation where the one or more complete SDUs are two octets smaller than the PDU. In a prior art UMTS type system, the RLC PDU UM header would contain two octets: the first being the SN and E-bit and the second being the ordinary length indicator 1111100 to indicate that the first data octet in this RLC PDU is the first octet of an RLC SDU. There would then be no room for the further ordinary length indicator to indicate that the last octet of the or the last SDU was the last octet of the PDU. So the current PDU would have to be followed by another PDU which would signal, using 0000000, that the last octet of the previous PDU was the last octet of an SDU.

The invention changes the meaning of the special length indicators to solve this problem. Taking the first embodiment as an example, the meaning of the special length indicator 1111100 is changed to indicate that the first data octet in this RLC PDU is the first octet of an RLC SDU and the last octet in this RLC PDU is the last octet of an RLC SDU (same or different SDU). The consequence is that all the information required by the receiver to unpack and process the data in the PDU is contained in that PDU. This has three advantages:

The Skilled Person would notice immediately that this is a solution to a particular problem. It is only useful and can only be used when the ‘two octets smaller’ situation happens to arise. The figures in the Patent make this point very clearly because Figures 2, 4 and 9 show situations where the patented solution is not used. As another example, take the situation where the alternative E-bit is configured and set to indicate that the RLC PDU contains a complete SDU which is not segmented, concatenated or padded. No length indicators are required and the SDU is one octet smaller than its PDU. In that particular situation, the alternative E-bit produces the same set of advantages as the invention.

The Claimants approach the Actavis questions on the basis of the following characterisation of the inventive concept of EP 142:

The Claimants further point out that, although it was possible in the existing UMTS standard to convey information identifying the "perfect packing" condition using length indicators, to do so required at least two length indicators: one to identify that the beginning of the first SDU was at the beginning of the PDU, and a second to identify the end of the last SDU in the PDU. The invention thus provides a more compact, efficient way of telling the receiver that for a PDU which contains complete SDUs, those SDUs are complete and can be extracted for processing by a higher layer.

It will be noted that the Claimants’ formulation of the inventive concept uses the term

The Defendants characterise the inventive concept very much in terms of the problem addressed by the Patent: how to indicate when the first octet of an SDU is the first data octet of a PDU and the last octet of the PDU is the last octet of an SDU when there is insufficient room for two ordinary length indicators to provide that information. They go on to submit that the inventive concept is bound up with using what they term the ‘length indicator technique’: this is the idea of using length indicators to indicate specific information, by which I think the Defendants mean ‘structure information’ and not the meaning of an ordinary length indicator.

I do not set out the principles from Kirin-Amgen. Although I have them clearly in mind, to set them out would only unnecessarily lengthen this judgment.

The issues of normal interpretation all reside in this phrase in claim 1: ‘an indicator including a length indicator for indicating’. Before tackling the key issue which concerns ‘length indicator’, it is convenient to make a few preliminary observations. The first concerns the last two words ‘for indicating’. In the usual way, this means that the length indicator must be suitable for conveying the information specified in the claim. Second, neither side made any particular observations about the slightly clumsy phrase ‘an indicator including a length indicator’ aside from their main submissions as to the meaning of ‘length indicator’. However, it is clear (and I find) that the method of claim 1 requires a single length indicator. Third, it can be seen from the specification that the meanings of the special length indicators for AM PDUs does not change. This is consistent with the title of EP142, even though claim 1 is not explicitly limited to UM.

Each side put forward their contentions on construction in terms of the functions that a length indicator performs. The rival contentions were as follows.

For their part, the Claimants contended that the Skilled Person would have understood length indicators to be those parts of the header that provide control information todescribe the payload (i.e. data) portion of the PDU (my emphasis), and the arrangement of the SDUs within it. They allow the receiving entity to separate the RLC-SDUs contained in the data part of the PDU. In some cases they do so by identifying the boundary of a specific SDU in the PDU and in other cases they provide some additional information about the structure of the payload.

In essence, the Defendants submit that the term ‘length indicator’ should be given the meaning it had in the CGK. As to precisely what that meaning was, as I understood the Defendants’ case, the primary function of a length indicator is to be used in the extension part of the header to indicate the size of each SDU (and so the boundaries between SDUs) in a PDU. To perform this function, length indicators included sufficient bits to be able to indicate the maximum length indicator value to be indicated in binary form. The Defendants go on to submit that it was also commonly known that certain predefined values in the length indicator field could be used for a secondary or ‘dual’ purpose of indicating other specific information when needed.

As can be seen, both contentions were quite elaborate – and to a certain extent each one reflected the arguments which the party in question wished to run on other aspects of the case.

Bearing in mind the purpose of a PDU is to carry SDU payload (whether the SDUs contained within it are complete or segmented), when an RLC PDU arrives at the relevant level of the receiver, the receiver needs to know how the various SDUs or SDU segments are packed in the PDU or PDUs, so that it can separate out each SDU and send each one for further processing in the next layer.

For the purposes of analysis I will use the umbrella term ‘structure information’ to encompass all the different types of information the receiver uses to unpack SDUs from PDUs. In the context of the Patent, structure information was conveyed in the following ways:

As the Patent says, length indicators in UMTS were either 7 or 15 bits long. The Skilled Person would have known that length indicators in GSM/GPRS were 6 bits in length. With greater volumes of data being transmitted in each generation, the Skilled Person would also anticipate that in 3.9G/4G even the shortest length indicator might well be greater than 7 bits in length. Whilst noting that the length indicator of claim 1 does not have to be limited to any particular number of bits, it is nonetheless convenient to discuss the essential attributes of a length indicator by reference to the 7-bit entity.

A field of 7 bits allows for 128 values (0-127). In the patented system applied in UMTS, the values of 1-123 were available for use as ordinary length indicators i.e. to indicate the position in the PDU of the final octet of a particular SDU. The values of 0 (in binary 0000000), 124 (1111100), 126 (1111110) and 127 (1111111) were used as special length indicators, to which one can also add 125 (1111101), a reserved value but evidently available for some future special use. As the Patent illustrates, the meanings given to the special length indicators could be changed and/or re-arranged. In theory, the number of special length indicators could be increased. In UMTS, the number of special length indicators was constrained by the use of the last two bits of the 7-bit indicator but in theory one could use the last three bits to define a larger class of special length indicators but that would necessarily cut down the available class of ordinary length indicators, which could be dealt with by using the 15 bit versions.

The Defendants characterised the length indicators as having a ‘dual purpose’, but in fact the uses made of the 7 bit length indicators were mutually exclusive. In a system, any particular value was either an ordinary length indicator or a special length indicator but (for obvious reasons) it could not be both. Both types of length indicator conveyed information about the relationship between the PDU and one or more SDUs contained within it. So, in the class of ‘length indicator’ a value either indicated the position in the PDU of the last octet of a particular SDU or, as a special length indicator, it conveyed a somewhat more elaborate type of structural information, namely the relationship between a PDU and the SDUs which were either contained in that PDU or the preceding PDU.

In my judgment, it follows that in EP142 one of the essential attributes of a ‘length indicator’ was that it was capable of realistically performing the role of an ordinary length indicator, even if particular values could be repurposed to act as special length indicators. I add the qualification ‘realistically’ because the Skilled Person is a practical person. He or she would not see a 1 or 2-bit field as realistically capable of acting as an ordinary length indicator in any practical system.

Next I want to consider whether the term ‘length indicator’ has a limiting effect. Let me consider for the purposes of argument alternative terms such as ‘indicator’ or ‘structure indicator’.

The terms ‘indicator’ or even ‘structure indicator’ in the context of an RLC UM PDU header in UMTS would be sufficiently general to cover ordinary length indicators, the special length indicators and the alternative E-bit. This helps to show that the term ‘length indicator’ was used in claim 1 as granted to exclude the alternative E-bit, there being no other words in the claim which could perform that role. This is why the Defendants encouraged me to consider the claim as granted.

Accordingly, I find that to fall within the term ‘length indicator’ in the Patent, the field in question must be capable of specifying the position in the RLC PDU where an SDU ends, even if certain values are capable of being used to provide the sort of indications of the existing special length indicators. If the field in question is not so capable, then in my judgment, it is not a ‘length indicator’.

It will be noted that this interpretation:

The absence of the restrictions I have just discussed releases the claim from the confines of the particular context in which the preferred embodiments are described, i.e. UMTS, and takes account of the explicit references to 3.9G.

In essence, the Claimants need the term ‘length indicator’ (which includes both (1) ordinary length indicators and (2) the special length indicators of the patent) to exclude (3) the single alternative E-bit but include (4) the 2-bit value FI=00. All four provide structure information about the SDU(s) in the PDU or, to use the Claimants’ phrase ‘control information about the payload’. The distinguishing feature of (3) and (4) is that they only provide structure information and are not capable of directly acting as

Accordingly, my finding as to the proper interpretation of ‘length indicator’ resolves the allegation of infringement on a normal interpretation of claim 1 of EP142.

Notwithstanding that it is not within the normal meaning of the relevant claim(s) of the patent, does the variant achieve substantially the same result in substantially the same way as the invention, i.e. the inventive concept revealed by the patent?

Naturally, the Claimants’ approach to this question depends heavily on their characterisation of the inventive concept which I have set out above. On that basis, the Claimants contend that the FI=00 value performs exactly the same role in the PDU as the new special length indicator of the Patent and conveys exactly the same control information about the payload as in the Patent. They go on to contend that the FI=00 is a single defined value to convey the fact that ‘perfect packing’ is present in the PDU, which also capitalises on the same efficient approach proposed in the Patent, using a single ‘token’ to do the job that previously required two. They also point out that the various values of the FI field as whole perform essentially the function of the special length indicators in UMTS.

The Defendants put forward a series of points as to why this question should be answered ‘No’. Although they were set out in considerable detail, they can be summarised conveniently as follows:

In response, the Claimants say that none of the differences which the Defendants identify matter to the invention.

The variant is when the FI field is set to 00. This value conveys the information that the first data byte/octet of the PDU is the first byte/octet of a SDU and the last byte/octet of the PDU is the last byte/octet of an SDU. Thus this value applies whether the PDU contains a single complete SDU or multiple complete SDUs. The receiver therefore receives exactly the information which the amended claim purports to specify. It is therefore clear that the FI=00 value does exactly the same job as the single special length indicator that is the subject of claim 1 as proposed to be amended.

Does the variant do this in substantially the same way as the invention? On this point I need to examine the Claimants’ characterisations of the inventive concept and their argument that the essence of a ‘length indicator’ is that it conveys ‘control information about the payload’ which allows the receiver to separate out the SDUs contained in the data part of the PDU. As to these:

The invention in the proposed amended claims achieves the result by using a length indicator. The FI=00 achieves the same result in a different way because it is not a length indicator.

The Skilled Person knows that the variant (FI=00) achieves the same result as is specified in the amended claims. Would it be obvious to the skilled person, reading the patent (as amended) at the priority date of September 2005, that the variant achieved the same result in substantially the same way as the invention? In my view it would not be obvious, because the Skilled Person would immediately understand that the variant is not a length indicator. The Skilled Person would also understand that the use of the FI=00 value is a different way of achieving the result.

Would such a reader of the patent have concluded that the patentee nonetheless intended that strict compliance with the literal meaning of the relevant claim(s) of the patent was an essential requirement of the invention?

In the circumstances of this case, the Skilled Person reading the Patent (as proposed to be amended) at the priority date would conclude that the patentee was using the term ‘a length indicator’ to exclude the alternative E-bit from the scope of the claim and therefore that what I found to be the normal interpretation of ‘a length indicator’ was an essential requirement of the invention. The reason is clear. Otherwise the claim would be invalid.

I am conscious that in the particular circumstances of this case that the answers to all three of the Actavis questions have depended on the normal interpretation of ‘length indicator’ and I have considered whether this itself indicates errors in my approach to the Actavis questions. I have concluded it does not. Even if I assume that Actavis questions 1 and 2 yielded positive answers, in the circumstances of this case, question 3 would still require the answer ‘Yes’.

Since I have found that the use of the FI=00 value in LTE does not infringe EP142 either on the basis of a normal interpretation or on the basis of equivalents, I can deal more briefly with the remaining points. On the basis of those findings, the Defendants do not allege the Patent is invalid.

What remains is the Claimants’ unconditional application to amend the Patent.

As I mentioned above, the purpose of the proposed amendments is to exclude the ‘single SDU embodiment’ from the scope of the claims. However, this application takes place against a backdrop where it is abundantly clear that in the Patent as granted, the single SDU embodiment was certainly within the scope of the claims and was certainly taught in the specification as being within the invention. There was very little focus on the motivation for the amendments. However, my understanding is that the reason for the amendments was to avoid the Samsung prior art. This proposed the alternative E-bit, which I have found to be CGK by the priority date.

The Claimants seek to exclude the ‘single SDU embodiment’ by proposing amendments which, they say, ‘specify that the length indicator of the claim must be for indicating the case where the first data octet of the PDU is the first octet of an SDU and the last octet of the PDU is the last octet of an SDU, for any number of SDUs in the PDU.’ The Defendants say that the proposed amendments fail to achieve this aim, since they say that either of the proposed amended claims 1 would be worked by an indicator which could only indicate the single SDU embodiment. The Claimants’ retort is that the Defendants have misunderstood the nature of the method which is claimed: they say the Defendants are muddling up the precise circumstances that may cause the indication to be inserted with the overall rule being applied; they also say that the fact that in an individual instance the SDU that started the PDU can be the same as the one that ends it does not mean the indicator is a single SDU only indicator.

In conceptual terms, what the Claimants are trying to do is clear enough. However, in my view, the proposed amendments do not achieve the aim. It will be recalled that the aim is to exclude the single SDU embodiment but include the indicator which is for indicating ‘perfect packing’ (to use the Claimants’ expression) whether that arises from a single SDU or multiple SDUs. However, in a working system, a single length indicator of a particular value (say 125) can only indicate one condition or the other. If 125 indicates the condition arising from a single SDU, and, say, 126 indicates the condition arising from multiple SDUs, the claim still embraces a length indicator which only indicates the single SDU condition. This arises from the language of the claims as proposed to be amended and I do not see how the amendments to the specification alter the conclusion. After all, the amendments to the specification define the concept which the amended claims are trying to claim but, as I have said, the claims fail in this endeavour. Far more wording would be required in the claims to achieve this.

The Defendants took a variety of other objections to the amendments and the UK IPO took the view that there was a lack of support under s.14(5)(c), but I do not think it is necessary to consider these points further. Many are founded in one way or another on the point in the previous paragraph.

Since the proposed amendments do not achieve the aim, the consequence is as follows. If I had found that the term ‘length indicator’ was broad enough on a normal construction to cover the FI=00 value in LTE, then EP142 would have been invalid over Samsung and the alternative E-bit. I am aware of the maxim of construction that indicates that a patentee does not normally phrase his claim so that it covers something in the CGK, but that is no more than a guideline. It is a guideline which is the more easily displaced when a patentee is attempting to force his claim to read onto an alleged infringement.

On the claims as granted, the reason why the alternative E-bit fell outside the claims was because it was not a ‘length indicator’. In this action, however, the Claimants wanted to establish that the use of the FI=00 value in LTE fell within claim 1 and for that purpose, the term ‘length indicator’ had to accommodate that type of 2-bit indicator. This was, of course, uncomfortably close to the single bit indicator of the alternative E-bit which applied in the scenario where a PDU contained a single SDU without segmentation, concatenation or padding. Accordingly, the Claimants needed to be able to distinguish the alternative E-bit in a different way.

The route the Claimants chose was to propose amendments which purported to exclude precisely the scenario of the alternative E-bit i.e. where an indicator could only indicate where a PDU contained a single SDU without segmentation, concatenation or padding, whilst retaining embodiments where a single indicator could indicate both that situation (i.e. start and end information of the same SDU) and the situation where it indicated the start of one SDU and the end of a different SDU) – this is the ‘same or different’ wording in [0009], [0028], [0032] and [0034].

If I had found that the scope of ‘length indicator’ was broad enough to cover the FI=00 value in LTE only by applying the Actavis questions, I incline to the view that EP142 would again be invalid due to the existence of the alternative E-bit but since this is hypothetical, it is not a necessary finding for me to make and the topic is controversial, I will say no more about it.

In any event, the claim for infringement of EP142 fails whether on the claims as proposed to be amended or on the claims as granted for the reasons set out above. Those reasons apply irrespective of the proposed amendments.

The Defendants made it clear that Qualcomm and WiMAX were pleaded as squeezes against the Claimants’ infringement case, such that the prior art only became relevant if the scope of protection was not limited, as I have found, to indicators capable of acting as what I might call true ‘length indicators’. Since I have found that the scope of protection is so limited, it is not necessary for me to lengthen this judgment with an analysis of the Qualcomm and WiMAX prior art and the various arguments which arise.

Despite the elaboration and effort which the Claimants have devoted to the claim for infringement of EP142, in my view this case boils down to a simple issue: the term ‘length indicator’ was used in claim 1 as granted to exclude the alternative E-bit, amongst other reasons. Although the alleged infringement, the FI=00 value, is not identical to the alternative E-bit, it has considerable similarities. In particular neither is capable of acting as an ordinary length indicator. Accordingly, if ‘length indicator’ is construed so as to exclude the alternative E-bit, it must also exclude the FI=00 from the scope of the claim, whether on a normal interpretation or on the basis of equivalents.

For all the reasons set out in this Judgment, the claim for infringement of EP142 fails.

an AMD PDU can contain a maximum number of 15 "Length Indicators" indicating the end of 15 corresponding SDUs; and

the rest of the AMD PDU space shall be used as padding or as piggybacked STATUS PDU. Length: 7 bits

Length: 15bits