Nokia Corp v Interdigital Technology Corp

This is an action in which the claimant (‘Nokia’) seeks to establish that the inventions claimed in a number of the defendant’s (‘InterDigital’s’) patents are not essential to the third generation mobile telecommunications standard in Europe. Twenty-nine patents which had been notified to the standards-setting organisation ETSI as essential to the standard according to ETSI’s rules were originally specified in the claim, but immediately before the exchange of evidence seven only were still in dispute. InterDigital did not ultimately advance any case in relation to three of these, so that by the time the matter came for trial only four patents remained in contention. Two of these are a parent and a divisional patent whose specifications are very similar, and the greater part of the evidence and the submissions according concentrated on three only of the patents.

The relief sought is a

‘Declaration that the importation, manufacture, sale, supply, offer for sale or supply keeping or use of (i) mobile telephones and (ii) system infrastructure equipment, or either of them, compliant with the FDD mode of operation as set out in 3GPP TS 21.101 Release 5 or any revisions to this or any later Releases as at the date hereof, without the licence of the Defendant, does not require infringement of [the listed patents] or any of them, such that the Patents and each of them are not Essential IPR for the FDD mode of operation of 3GPP TS 21.101 Release 5 or do not remain or have not become Essential IPR for any revisions to this or any later Releases as at the date hereof.’

The phrase ‘Essential IPR’ is taken from the Intellectual Property Rights Policy of ETSI, the European Telecommunications Standards Institute. ETSI is an industry-established body responsible for setting the technical standards for mobile telecommunications equipment and infrastructure in Europe. The standard with which I am concerned is established by the 3GPP or Third Generation Partnership Project, and is generally popularly known as 3G. This is a field in which acronyms are a constant source of doubt and confusion, and I take the following from the preface of one of the very helpful books with which the parties provided me, “WCDMA for UMTS” 3^rd Ed by Holma and Toskala:

‘Second generation telecommunication systems, such as GSM, enabled voice traffic to go wireless: the number of mobile phones exceeds the number of landline phones and the mobile phone penetration exceeds 80% in countries with the most advanced wireless markets. The data handling capabilities of second generation systems are limited, however, and third generation systems are needed to provide the high bit rate services that enable high quality images and video to be transmitted and received, and to provide access to the web with higher data rates. These third generation mobile communication systems are referred to in this book as UMTS (Universal Mobile Telecommunication System). WCDMA (Wideband Code Division Multiple Access) is the main third generation air interface in the world and deployment has been started in Europe and Asia, including Japan and Korea, in the same frequency band, around 2GHz.’

The development of such a system involves a vast amount of technical work by the participants, and this work produces patents. Patents are inimical to technical standards if no licence is available when compliance with the standard requires use of the patented invention, and ETSI has developed an Intellectual Property Right Policy which is intended to ensure that participants in the standards-setting process make licences available under patents whose use is essential to compliance with the standard. The notification to ETSI of patents believed by the proprietor to be essential to a standard is described in an interlocutory judgment of the Court of Appeal in this action ([2006] EWCA Civ 1618) and I shall not repeat it here. From the Court of Appeal’s judgment I think it is established that there is jurisdiction to entertain an action like the present where negative declarations as to the essentiality of a patented invention to a standard are sought is established if the Court has personal jurisdiction over the defendant and if sufficient facts are alleged that it is possible that the Court might grant declaratory relief. Whether declaratory relief will be granted is a matter of a discretion to be exercised on all the relevant available material in every case. As Lord Woolf MR put it in Messier Dowty v Sabena SA [2000] 1 WLR 2040,

‘39. As against these authorities there are the numerous cases where without objection negative declarations have been granted. There are also the judgments of Lord Denning M.R. in this court and Lord Wilberforce in the House of Lords in Camilla Cotton Oil Co. v. Granadex S. A. [1975] 1 L1oyd's Rep. 470; [1976] 2 L1oyd's Rep. 10. Lord Denning M.R. said [1975] I L1oyd's Rep. 470, 474-475:

"It has been said that a declaration as to non-liability ought very rarely to be made, see Dyson v. Attorney-General [1911] I K.B. 410 and Guaranty Trust Co. of New York v. Hannay & Co. [1915] 2 K.B. 536. And In re Clay [1919] 1 Ch. 66 is sometimes cited for the proposition that it cannot be made. But it is nothing of the kind. In modern times, I think that a declaration as to non-liability can be made whenever it will serve a useful purpose. I would not limit it in any way."

40.

In the House of Lords Lord Wilberforce stated the position in his own words but the effect was very much the same. He said [1976] 2 L1oyd's Rep. 10, 14:

"The declaration claimed is of a negative character and as Lord Sterndale himself [as Pickford L.J. in Guaranty Trust Co. of New York v. Hannay & Co. [1915] 2 K.B. 536, 564] had said: '... a declaration that a person is not liable in an existing or possible action is one that will hardly ever be made.' 'Hardly ever' is not the same as 'never' but the words warn us that we must apply some careful scrutiny. So I inquire whether to grant such a negative declaration would be useful. The liability which the English court is asked to negative is any possible liability of the respondents on the basis of agency ..."

41.

Lord Wilberforce and Lord Denning M.R. differed in the circumstances of that case as to whether the declaration would serve a useful purpose. However, if it would, that it would then be appropriate to grant a declaration was agreed. The approach is pragmatic. It is not a matter of jurisdiction. It is a matter of discretion. The deployment of negative declarations should be scrutinised and their use rejected where it would serve no useful purpose. However, where a negative declaration would help to ensure that the aims of justice are achieved the courts should not be reluctant to grant such declarations. They can and do assist in achieving justice. For example, where a patient is not in a position to consent to medical treatment declarations have an important role to play… As Sir Thomas Bingham M.R. said in In re S. (Hospital Patient: Court's Jurisdiction) [1996] Fam. 1, 19:

"Any statutory rule, unless framed in terms so wide as to give the court an almost unlimited discretion, would be bound to impose an element of inflexibility which would in my view be wholly undesirable."

He considered that the different situation he was there considering was "pre-eminently an area in which the common law should respond to social needs." So in my judgment the development of the use of declaratory relief in relation to commercial disputes should not be constrained by artificial limits wrongly related to jurisdiction. It should instead be kept within proper bounds by the exercise of the courts' discretion.’

It is obvious from this quotation that the principal factor affecting the exercise of the court’s discretion, apart from such matters as the sufficiency of the description of the device or system to which the invention is said to be inessential, is the utility of the negative declaration sought. Would the declaration if granted be the legal equivalent of shouting in an empty room, or is there some point in it? It was on this issue that I heard the evidence of Mr Richardson, called as an expert by InterDigital. I did not hear evidence from any negotiators employed by the parties themselves, but Nokia deployed written evidence given at an earlier stage of the action by Mr Boles of InterDigital.

Nokia also reserved the right to challenge the validity of any patent which claimed an invention I found to be essential to the standard. The subject matter of these patents, and of the standard, is complex and I can see a real advantage in restricting the attack on validity to those patents found to be essential. Accordingly I heard evidence from the experts, Professor Purat, Professor Manikas and Professor Marshall, going to the issues of construction and essentiality only. I was also given a tutorial over two days by Mr Wiffen, who was selected by the parties to educate me on the relevant technical aspects of the European 3G system generally. This introduction was very helpful, and as the subject matter of patent actions becomes increasingly complex such a non-controversial introductory course for the judge seems to me to be highly desirable. I also think that there may be something to be said for dealing with the expert evidence at trial in a manner slightly different from that to which we are accustomed. I was struck in this case by the strain placed on the cross-examiner and on the witness, and I suspect that had I been more enterprising in the management of the case, I would have tried to ensure that the experts were aware of the main points that would be raised in cross-examination, notwithstanding that they had read each other’s reports and should therefore have been aware of the main differences between them. This would have been particularly useful in the present case because Professor Manikas liked to explain points as many lecturers do, by adding to diagrams and expressions on a whiteboard, and this is often difficult to follow when considering the effect of the evidence as a whole. Professor Purat did not use a whiteboard, but I did not think that he found cross-examination in his second language, with no indication of the line being followed and with questions containing a great deal of nuance, particularly easy, although he acquitted himself very well.

UMTS is a cellular system, in which base stations (properly called ‘Node Bs’) communicate with mobiles (properly user equipments, or ‘UEs’) within physical range of the base station. The base stations communicate over a complex digital infrastructure which may involve many kinds of physical link. The patents in this action are concerned with the air interface, that is, the manner in which data is transmitted between the mobile and the base station using WCDMA. This system is very complex, and the patents are concerned with only a small part of it. Their subjects are (1) the control of the power of the signal transmitted by the mobile (’610 and ’807) (2) the use of multiple global pilot channels by a base station (’749) and (3) the use of multiple transmitting antennae at the base station for improving the quality of the signal received by the mobile (’777).

Underpinning the inventions are a number of basic facts about the air interface which need to be borne in mind if one is not to be overborne by the complexities of the underlying physics and of the standard itself. The first of these is the basic communication method, code-division multiple access, or CDMA.

It is rather obvious that in a communication system used simultaneously by more than two users there must be a way of keeping the various communications separate. If they interfere with each other, they will be lost. In an old-fashioned land-line telephone system, the conversations are kept separate between the subscribers and the exchange by using a separate pair of copper wires for each telephone. When the telephone handsets communicate with the exchange not by individual copper circuits but by radio, the individual channels of communication must still be kept separate, and, because only a limited number of communication channels are available, they must be made available to the users as and when required. There are a number of ways of doing this, three of which are particularly relevant. In order of increasing complexity, the three techniques are frequency-division multiple access (FDMA), time-division multiple access (TDMA) and code division multiple access (CDMA).

Frequency division is the easiest concept. Like the different radio stations on the radio receiver dial, each handset transmits and receives at particular defined frequencies which may be dynamically allocated to it. The task of allocating frequencies to handsets is performed by the base station, or by software associated with the base station. In time division multiple access, all communications are encoded and divided into blocks, which are transmitted at the same frequency but in identified time slots, the blocks being reassembled into speech or whatever at the other end. Software at the base station manages the allocation of blocks to particular handsets and carries out all the other management tasks.

CDMA exploits a mathematical quality of so-called orthogonal pseudo-random code sequences. Professor Purat in paragraphs 19 and 20 of his report described them like this:

‘In spread spectrum communications systems each user accesses the shared communication channel at the same time and with the same frequency. In order to distinguish between the different users each mobile station is allocated its own so-called spreading code. The code is used to change the information that the user wants to transmit in such a way that it can easily be distinguished from other users' information. This spreading operation is simply a multiplication of the information data signal with a spreading code signal. The information data signal consists of the values -1 and + 1 that represent the binary information digits 0 and 1, which are called bits. The spreading code signal also consists of the values -1 and + 1, which are called chips to distinguish them from the bits. Therefore, the spreading code signal is also called chipping code signal. The values of the chipping code signal changes much faster than the values of the information data signal which causes an increase (spread) of the frequency bandwidth of the information data signal, hence the name spread spectrum. This multiple access technique is also called code division multiple access (CDMA).

All spread information data signals from the different users interfere with each other on the communication channel. However, the [spreading] codes are constructed in such a way that the receiver in the base station can discriminate between the different information data signals and can recover the information signals of one user from the mixture of spread information signals.’

There are a number of points in this case which it is helpful to remember. The first is that in both the downlink (communications from the base station to the UE) and in the uplink, the spreading code may be regarded as a per-user thing, so far as the users communicating with a particular base station are concerned. Certain of the signals put out by the base station are broadcast to all UEs currently in communication with that base station: what differentiates the different base stations is a different scrambling code, which does not further spread the spectrum of the signal:

This figure, which is taken from page 89 of Richardson, WCDMA Design Handbook, one of the books supplied to me, shows the spreading of the signal and the scrambling of the spread signal. The ∑ denotes a sum-and-dump or integrating function: the despread signal is summed at the bit rate over the period of so many chips as encode a single bit. In WCDMA the chip rate is 3.84MHz, and Richardson gives an interesting example. The bit rate for speech is 12.2 kilobits per second (kbps) and so the gain is 3.84e6/12.2e3 or about 314: in decibels that is 25dB. In such a case, the ‘processing gain’ effected by the despreading is 25dB on the desired signal, and if the other user’s signals are not despread they will amount to noise.

In WCDMA the way in which the spreading code and the scrambling code are allocated to each user and the purpose for which they are used is not quite the same on the uplink and the downlink, but each pair is unique to a single user/base station combination. Thus on the uplink, the scrambling code distinguishes UEs but on the downlink it distinguishes cells. It is part of the function of higher level control functions to allocate the actual codes, and with these functions this case is not concerned.

With this very brief introduction, it is possible to consider the earliest of the patents with which I am concerned,’610 and ’807. Both these patents are concerned with power control, and, broadly stated, are concerned with ensuring that the signals received from the UEs at different distances from the base station arrive with the same power. The claims are alleged to be infringed by, and the patent to be essential to, the mechanism prescribed by the Standard for controlling the power transmitted by the UE at the beginning of the process of starting up, for example when it is turned on.

The specifications of the patents are substantially identical as a matter of disclosure. The claims differ in one respect, and I shall concentrate on ’610, entitled ‘Adaptive Power Control for a Spread Spectrum Transmitter’. It claims priority from 16 November 1990, and so precedes the 3GPP standard substantially.

A patent specification is to be construed through the eyes of the skilled addressee, who possesses the common general knowledge. On this issue, Professor Purat gave evidence on behalf of Nokia, and Professor Marshall on behalf of InterDigital. It was established that there was at this date no working CDMA system publicly available. Nevertheless, the CDMA coding technique, including both the need for, and the basic techniques of, open and closed loop power control must on the evidence be taken to be common general knowledge. There was much discussion of open and closed loop systems: closed loop systems involve some form of feedback, whereas an open loop system has no feedback.

The problem with which the patent is concerned is described at column 1 line 40:

‘A cellular communications network using spread spectrum modulation for communicating between a base station and a multiplicity of users, requires control of the power level of a particular mobile user station. Within a particular cell, a mobile station near the base station of the cell may be required to transmit with a power level less than that when the mobile station is near an outer perimeter of the cell. This is done to ensure a constant power level at the base station, received from each mobile station. A representative power level control system adapted to address power control requirements within particular cell is that discussed in a scientific paper by R.F. Ormondroyd entitled "Power Control for Spread-Spectrum Systems”, pages 109 to 115 of the Proceedings of the Conference on Communications Equipment and Systems, held on April 20-22, 1982 in Birmingham, U.K. The Ormondroyd system is a closed loop system requiring feedback, namely, the power level of a mobile unit transmission is measured at the base station and, responsive to this measurement, the base station directs an increase or decrease in the mobile unit’s transmitter power.’

The specification then discusses the problem of moving between cells in which the mobile unit may be close to the base station, so requiring low power, on the one hand and distant from the base station, so requiring higher power, on the other. The objects of the invention are described thus:

‘An object of the invention is to provide an apparatus and method for automatically and adaptively controlling the power level of a plurality of mobile stations so that the power level received at the base station of each cell is the same for each mobile station.

Another object of the invention is to provide a spread-spectrum apparatus and method which will allow operating a spread spectrum transmitter in different geographic regions, wherein each geographic region has a multiplicity of cells, and cells within a geographic region may have different size cells and transmitter power requirements.’

The invention is defined by claim 1 as follows:

‘An apparatus for adaptive-power control of a spread-spectrum transmitter of a mobile station operating in a cellular communications network using spread spectrum modulation

characterised by

a base station for transmitting, on a continuous basis or on a repetitive periodic basis, a generic spread-spectrum signal and an APC-data signal;

a multiplicity of mobile stations, each mobile station having

an acquisition circuit (101,102,103) for acquiring and decoding the generic spread-spectrum signal;

a detector (104) coupled to said acquisition circuit for detecting a received power level of the generic spread-spectrum signal;

a decoder (105) coupled to said acquisition circuit for decoding the APC-data signal as a threshold;

a differential amplifier (106) coupled to said detector and said decoder for generating a comparison signal by comparing the received power level to said threshold;

a transmitter (112) for transmitting a transmitter spread-spectrum signal;

an antenna coupled to said transmitter; and

a variable-gain device (111) coupled to said differential amplifier and between said transmitter and said antenna, responsive to said comparison signal indicating an increase or decrease, for adjusting a transmitter-power level of the transmitter spread-spectrum signal from said transmitter.’

The corresponding method claim, claim 6, omits the hardware features entirely:

‘6. A method for adaptive-power control of a spread-spectrum transmitter of a respective one of a plurality of mobile stations operating in a cellular communications network using spread-spectrum modulation,

characterized by

a base station transmitting, on a continuous basis or on a repetitive periodic basis, a generic spread-spectrum signal and an APC-data signal, used by the plurality of mobile stations to adjust a transmitter-power level of a respective plurality of mobile station transmitters, said method comprising the steps, at each mobile station, of:

acquiring (702) and decoding (704) the generic spread-spectrum signal;

detecting (703) a received power level of the generic spread-spectrum signal;

decoding the APC-data signal as a threshold (705);

generating a comparison signal by comparing (706) the received power level to said threshold; and

adjusting (707) a transmitter-power level of a transmitter spread-spectrum signal from a respective transmitter responsive to said comparison signal.’

The general scheme of these claims is clear enough. The base station transmits two things: a signal whose transmitted power is known (the ‘generic spread-spectrum signal’) and a number for the purposes of comparison (the ‘APC-data’). The measured power of the received generic signal is compared in the mobile with the number, so providing a measure of the attenuation of the channel in the downlink direction. The assumption is made that the attenuation of the uplink will be the same, and the transmitter power is adjusted so that taking the measured attenuation into account the detected power of the uplink signal at the base station should have an appropriate value.

A number of observations need to be made on the claims generally. The method claim has no equivalent to the requirement in the apparatus claim that there be a variable gain device ‘responsive to said comparison signal indicating an increase or decrease, for adjusting a transmitter-power level’. All the claim requires is that the transmitter-power level be adjusted ‘responsive to said comparison signal’.

It is not clear that the method disclosed is for use during startup: at column 6 lines 48 to 50, the ‘threshold’ (i.e. the result of decoding the APC-data) is used either to adjust or set the threshold of the differential amplifier, the device which, in the preferred embodiment, is used to compare the ‘threshold’ with the measured power of the received generic signal. I did think that the use of the word ‘set’ is not really consistent with anything other than startup, but on reflection I think this is wrong. What matters is not what the threshold is, but what the transmitted power is to be. Were the starting power level of all the UEs standardised (all start at 10mW, say) this would not be a problem, because all use the same ‘threshold’, provided that the comparison process of the invention produced a number which could be used to set the attenuation/amplification of the transmitter. It seems clear that if the device were to be used for startup, the APC-data would have to be a representation of the transmitted power of the generic signal. But nowhere in the specification is the nature of the APC data described at all: it is left to inference.

With these preliminary observations, it seems sensible to turn to the 3GPP Standard to identify the points of construction that arise. Although one construes a claim ‘as if the defendant had never been born’, in any complex case it is essential to see where the shoe pinches so that one can concentrate on the important points. It is important nevertheless that the opportunity thus presented to construe the document with one eye on the infringement must be rejected, as far as possible. So when the claim calls for A, and the standard requires B, the right question is not whether A means B, or covers B, or might with hindsight be said to be another example of the genus of which B is also a member, but whether in its context in the specification the skilled man would appreciate that A in the claim encompassed B.

At an earlier stage in the action, I directed the parties to serve statements of case on essentiality. This had the merit that the relevant passages in the standard should have been defined before the evidence was exchanged, and so should have reduced the opportunity for surprise. The experience of this case suggests that such statements of case are essential.

The relevant parts of the 3GPP Standard are TS 25.214 § 6.1 (the ‘Physical Random Access procedure’) and §8.5.7 (the ‘Open Loop Power Control’ protocol). The protocol is employed during the Random Access procedure, which dictates how a UE makes itself known to a particular base station at the physical level. For present purposes, the essence of the transaction is that the UE transmits a ‘preamble’ signal at a power level which is computed in the UE over the physical random access channel (PRACH). If this preamble is not acknowledged by the base station, the UE increases the power and tries again. The invention specified in the claims is said to be essential to the setting of the power level for the first try at sending the preamble. Professor Purat’s summary of what the standard requires (page 66 of his report) was accepted to be correct:

‘…the mobile station transmits a first PRACH preamble with the preamble transmission power that is equal to the parameter "Preamble_Initial_Power". When the preamble is not detected or incorrectly received by the base station it sends nothing or a negative acquisition indicator (AI) on the AI channel (AICH) to the mobile station. After a specified time or on reception of the negative AI, the mobile station transmits a new preamble with an increased power level which is calculated as a sum of the initial power level "Preamble_Initial_Power" and a value "Power ramp step". This process is repeated until either a positive AI is received by the mobile station, indicating a correct detection of the AI by the base station (good case), or until a maximum number of preambles has been transmitted by the mobile station without positive acknowledgement by the base station (bad case). In the good case, the mobile station transmits the message data on the PRACH before ending the physical random access procedure. The message is sent with a power offset Pp_m relative to the last transmitted preamble. In the worst case, the physical random access procedure is exited without data transmission and random access control is given back to higher layers [programs controlling the base station] that will re-initiate the complete physical random procedure with the same message data after a certain random backoff time. The random access can be illustrated by the following example, see figure 11, where the first physical random access procedure is unsuccessful, and the message will be transmitted only during a second physical random access procedure. Also, in this example, the preamble initial power P_init was updated by RRC for the second physical random access procedure. Such an update may result for instance from an increased path loss that has been measured by the mobile station, as explained in the next paragraph.’

Professor Purat’s Figure 11, which I reproduce below, shows the whole process:

The question is how the power level P_{init, 1} in the diagram is arrived at. One of its components, CPICH_RSCP, is the Received Signal Code Power, measured by the UE, of a particular signal transmitted by the base station, the common pilot channel or CPICH. Carrying on with Professor Purat’s explanation:

‘According to this specification … the transmitter power level for the initial PRACH preamble transmission (Preamble_Initial_Power) is calculated from the values "CPICH_RSCP", "Primary CPICH TX power", "UL interference", and "Constant Value".

…

The value "CPICH_RSCP" is defined in TS25.215…[as]

Received Signal Code Power, the received power on one code measured on the Primary CPICH. The reference point for the RSCP shall be the antenna connector of the UE…

…

The values "Primary CPICH TX power", "UL interference", and "Constant Value" are transmitted on the broadcast channel so that all mobile stations can read them from this channel. However, the mobile station may take these values from an internal storage as well, if it has read and stored them before and the stored values are still valid, which basically means that they haven't become obsolete. A value may for instance become obsolete when the mobile station changes the cell, because the broadcast parameters are cell dependent. Furthermore, the network can notify the mobile stations that parameters have changed in a cell and shall be re-read from the broadcast channel.’

The source of the data and the calculation is prescribed in section 8.5.7 of TS 25.331 (‘IE’ stands for ‘information element’):

‘1> acquire valid versions of the necessary System Information IEs as follows:

2> if the UE has stored valid versions of the IEs "Primary CPICH Tx power" and "Constant value":

3> use the stored content of the IEs.

2> otherwise:

3> read and store the IE "Primary CPICH Tx power" and "Constant value" in System Information Block type 6 (or System Information Block type 5, if System Information Block type 6 is not being broadcast).

2> if the UE has a valid version of the IE "UL interference" stored:

3> use the stored content of the IE "UL interference".

2> otherwise:

3> read and store the IE "UL interference" in System Information Block type 7;

3> if the UE fails to read the lE "UL interference" in System Information Block type 7 due to bad radio conditions, the UE shall use the last stored lE "UL interference".

1> measure the value for the CPICH_RSCP;

1> calculate the power for the first preamble as:

Preamble_Initial_Power = Primary CPICH TX power - CPICH_RSCP + UL interference + Constant Value

Where,

Primary CPICH TX power shall have the value of IE "Primary CPICH Tx power",

UL interference shall have the value of IE "UL interference"; and

Constant Value shall have the value of IE "Constant value".

1> as long as the physical layer is configured for PRACH transmission:

2> continuously recalculate the Preamble_Initial_Power when any of the broadcast parameters used in the above formula changes; and

2> resubmit to the physical layer the new calculated Preamble_Initial_Power.’

There is thus one measured quantity (CPICH_RSCP) and three supplied quantities (Primary CPICH TX power, UL interference and Constant value). Semantically, the 3GPP Standard requires (Footnote: ¹) these numbers to represent physical quantities and factors, as follows: Primary CPICH TX power is a power in dBm (an absolute unit, i.e. power relative to 1 milliwatt expressed in dB, so that 3dBm is 2mW, 10dBm is 10mW, 20dBm is 100mW, 60dBm is 1kW and so on) (range -10 to 50); UL interference is a power in dBm (range -110 to -70); and finally Constant Value is expressed in dB (i.e. a factor expressed as a logarithm) (range -35 to -10).

The calculation is expressed in logarithmic quantities. So far as the linear physical quantities represented by those logarithms are concerned, the calculation is to be understood as taking the ratio of the supplied Primary CPICH Tx Power (the power of the base station transmission) and the CPICH_RSCP (the measured received power), multiplying the UL Interference by that ratio and multiplying the product again by the Constant Value: . I see this as setting the power of the transmitter in the UE so that the power of that signal at the base station is equal to the power of the interference at the base station times a constant. Professor Purat demonstrated that this expression can be derived directly from the fundamental consideration that there be a specified minimum signal-to-interference ratio at the base station and from the consideration that the uplink and downlink path losses should be identical. The Constant Value is, in fact, the required signal-to-interference ratio–see paragraph 75 of his statement.

Given this analysis, the questions on construction can be boiled down to the following (this generally follows Nokia’s list, but there was some agreement between the parties):

One preliminary observation needs to be made. The terminology of claim 1, and to a lesser extent that of claim 6 also, is analogue in its nature. By way of example ‘differential amplifier’ is a term of art, denoting a linear device whose output is proportional to the difference between two inputs. Of course the output may be arranged to swing to maximum and minimum values depending on whether the difference is positive or negative, but that does not affect the general point. On the face of it, no claim requiring a differential amplifier could possibly be essential to a Standard exclusively drafted in terms of digital processing. But there is a difficult passage at column 7 line 2 – difficult because of the problems in assessing its impact on the claim:

‘The APC circuit 110 of FIGS. 1 and 2 may be built on a digital signal processor chip. An analog to digital converter located at the output of the bandpass filter 103 would convert the received signal to a data signal. The envelope detector 104, decoder 105 and differential amplifier 106 may be implemented as part of digital signal processing functions on the digital signal processor (DSP) chip. The analog to digital converters may be included on the DSP chip.’

If claim 1 is read in the way this passage encourages, the hardware features expressly specified have to be considered to be functional limitations. The differential amplifier must just be viewed as some sort of subtractor, or an apparatus functioning to subtract two numbers representing its inputs, for example. I am doubtful as a matter of general principle whether this can be a valid approach to the interpretation of a patent, the more so since this specification dates from 1990 and digital signal processing had been well understood for many years before that, and the selection of a differential amplifier in the claim must have been deliberate. On the whole, nevertheless, I shall interpret the claim as a ‘means+function’.

This means that in the claim the words ‘comparison signal’ encompass a number representing a comparison between two other numbers, respectively a number representing the received power and a number representing the threshold. This is entirely consistent with column 3 line 17: ‘[t]he comparison signal may be an analogue or digital data signal’.

(i) What is the differential amplifier? It follows from the discussion above that in context the term ‘differential amplifier’ is apt to cover any device, or arithmetic function, which produces as a result a number representative of a comparison between two numbers. Classically this is a subtraction function, or a division, because the result of a subtraction is a number indicating the size of the difference between the two, and division is just repeated subtraction. It is what ‘differential’ means, in this context. The result of a division does not indicate the absolute size of the difference between the numbers, but only its relative size.

(ii) Initial setting. Nokia set great store by the answer given by Professor Marshall at transcript 586-7. I do not find it particularly helpful, because the Professor was not being asked about the passages in the specification (column 5 line 50-4, column 6 lines 48-50, column 7 line 36) to which I have referred above which refer to the use of the threshold to adjust and/or set the threshold of the differential amplifier. The comparison signal adjusts the transmitter power level in both claims 1 and 6, but ‘adjust’ can just imply that the power is set to the right value. What is determinative in claim 1 is the clear indication that adjustment is to take place in response to changes in the measured signal or the threshold (‘…said comparison signal indicating an increase or decrease…’). This cannot cover startup, at least without a number of additional requirements (a standard startup power or an APC which specified the transmitted power of the generic signal). Nokia are right that the focus of the specification is just open-loop power control as a general expedient but there is support in the specification for a claim which covers setting up the threshold value for the first time.

(iii) threshold. The meaning of threshold in its context is difficult. Normally, it would mean a number which represents a minimum or floor value of some quantity. I do not think this matters, because in the patent there is no doubt that the ‘threshold’ is just the value applied to one of the inputs of the ‘differential amplifier’ against which the measured power value is compared to give the comparison signal.

(iv) comparison signal. There was a considerable dispute between the experts whether the comparison signal as described in the patent was a two-valued signal (ie up/down) or not. This depends upon whether the differential amplifier is arranged to have a two-value output. Professor Purat considered that it did do so, on the basis that it was merely a comparator, but there is really no compelling reason why this should be so. The differential amplifier is thus capable, and the claim covers a digital device capable, of yielding an output equal to the difference between its two inputs.

(v) acquiring and decoding the generic spread-spectrum signal. The indicative reference numbers in the claim indicate that this function is carried out by the despreader. In the embodiments described in the patent, the skilled person is given the option of transmitting the APC-data (which is generic and intended for reception by all users within a given cell) encoded in the generic-spread spectrum signal whose received power is measured at the UE (column 4 lines 42-54) or in another signal. In the former case, obviously the generic spread-spectrum signal must be despread, i.e. decoded. The short passage at column 3 lines 30-35 is absolutely precise:

‘Each mobile base station performs the steps of acquiring the generic spread-spectrum signal transmitted from the base station, and detecting a received power level of the generic spread-spectrum signal. The steps also include decoding the APC-data signal as a threshold from the generic spread-spectrum signal or from a signal or channel separate from the generic spread-spectrum signal.’

A construction of the claims that required the generic signal to be despread limits them to the Figure 1 embodiment only. The claim is not well drafted and I am not willing to conclude that there is any requirement that the spread-spectrum signal must be decoded when it carries no useful information. The word ‘for’ does not usually introduce a limitation (it may mean ‘suitable for’ or ‘capable of as required’) and I consider that the claims should be interpreted as generally descriptive of the function of the acquisition circuit, but without imposing a technical limitation that is both unnecessary and inconsistent with the statement elsewhere in the specification that the Figure 2 embodiment is an embodiment of the invention (column 4 lines 2-3).

Interpreting the claim in this way leaves the question on essentiality limited to the question whether the standard can be said to specify a threshold signal, a comparison of detected power with the threshold, and the generation of a comparison signal, and (for claim 1) whether the comparison signal indicates and increase or a decrease. The last point apart, this depends entirely on the right view of the obligatory formation of the sum:

‘Preamble_Initial_Power = Primary CPICH TX power - CPICH_RSCP + UL interference + Constant Value’

whose equivalent in linear quantities is

Professor Marshall saw the required threshold value in the sum of the three quantities Primary CPICH TX power + UL interference + Constant Value, all of which are provided in the System Information Blocks broadcast by the base station. The difficulty with this view is that it would follow that the comparison which is made by subtraction of the logarithmic value of the measured received power, is taking a ratio, not to derive a signal which indicates an increase or decrease in transmitted power, but to state what the absolute transmitted power is to be, as Professor Purat observes in paragraph 114 of his first Report. This is not in any sense determining a gain–as it would be for the variable gain device–and I conclude that claim 1 would not be infringed by an apparatus which carried out the scheme of the standard.

The position of claim 6 is different. I think that the last feature of the claim

‘adjusting (707) a transmitter-power level of a

transmitter spread-spectrum signal from a respective

transmitter responsive to said comparison signal’

does read on to the process in which the comparison signal specifies the transmitter power which is to be used.

In my view claim 6 is infringed by carrying out the method specified in the standard of effecting open loop power control in the Physical Random Access procedure, and the invention is essential to the 3GPP Standard to that extent.

Claim 1 of this patent contains the features of claim 1 of ’610 that in my view preclude a finding of essentiality of that claim. For that reason, this claim is not essential. There is no claim of the same kind as claim 6, and I conclude that all the claims are inessential to the 3GPP Standard.

This patent, entitled ‘Modular Base Station with Variable Communication Capacity’, is concerned with the use of multiple pilot signals by a single CDMA air-interface base station. Its priority date is 1998.

Every base station has to transmit at least one pilot signal that will be received by all mobile units within range. The problem that the patent is primarily dealing with is an infrastructure in which base stations are made up of more than one ‘base station unit’ each one of which transmits global pilot signals. No part of the 3GPP Standard relates to such an infrastructure, but claim 13 is directed to a subscriber unit in communication with any base station transmitting multiple global pilot signals. Such an arrangement is said to be relevant to systems which do not have the scalable base station feature with which the patent is concerned.

The specification thus describes multiple global pilots as a consequence of the use of multiple base station units. The existing prior art and the problem to be overcome are described in those terms:

‘[[0005] If a coherent modulation technique such as phase shift keying or PSK is used for a plurality of subscribers, whether stationary or mobile, a global pilot is continuously transmitted by the base station for synchronizing with the subscribers. The subscriber units are synchronizing with the base station at all times and use the pilot signal information to estimate channel phase and magnitude parameters. For the reverse-link, a common pilot signal is not feasible. Typically, only non-coherent detection techniques are suitable to establish reverse-link communications. For initial acquisition by the base station to establish a reverse-link, a subscriber transmits a random access packet over a predetermined random access channel (RACH).

…

[0011] The present invention provides a base station architecture that is modular in configuration, lowering the initial cost of implementing a new CDMA telecommunication system for a defined geographical region while allowing for future capacity. The scalable architecture is assembled for a digital base station unit that is configured to support a plurality of simultaneous wireless calls connecting to a conventional public switched telephone network. For initial startup, two base station units are deployed for redundancy in case of a single failure. Additional base station units may be added when the need arises for extra traffic capacity. If sectorization is required, the base station units may be directionally oriented.

…

[0021] The scalable modular base station for a CDMA air interface requires a set of global channels to support operation. The global pilot supports initial acquisition by the subscriber and provides channel estimation for coherent processing. One or more global broadcast channels provide signalling information. Each BSU requires its own set of global channels. However, global channels use air capacity and [it] is therefore costly to assign a set of full strength global channels for each BSU.’

The problem with which the patent is concerned is therefore the need for the use of more than one global pilot signal so that the UE can acquire any one of the BSUs and the solution proposed is for each BSU to transmit a full-power global pilot signal for a period in turn. This is a form of time division multiplex of the global pilot signals. Professor Purat illustrated it like this (paragraph 528 of his report):

The diagram shows subscriber unit wake-up because this is synchronised with the high-power periods of transmission. A subscriber (mobile) unit which is ‘asleep’ to save energy must nevertheless check on a regular basis for messages (as, for example, incoming calls) addressed to it. So it must wake and, when it does so, it will acquire the BSU which is transmitting a high power signal at that moment. The ‘summary of the invention’ (column 2 line 50 – column 3 line 22) throws no light at all on the signalling protocols or format, and the only context for the claim is provided by the common general knowledge and by the single preferred embodiment of the invention described in the specification. This is reasonably clear. Once the idea of a scalable, modular, base station is understood, it must equally be understood that each module will act as a base station, it will be appreciated that each module (i.e. BSU) has its own set of global pilots, just like the set of global pilots possessed by a non-scalable, non-modular, base station.

‘[0045] Each BSU communicates independently with an assigned subscriber. As previously described, to accomplish this function each BSU 69 must have unique global channels for the global pilot 137, the fast broadcast channel 139 and the slow broadcast channel 141.

[0046] The unique global pilot 137 allows each subscriber 25 to synchronize with an individual BSU 69. The fast broadcast channel 139 provides a traffic light function to the subscriber 25 informing him [of] BSU 69 availability and power ramp-up status from the respective BSU 69. The slow broadcast channel 141 transports activity and paging information…

[0047] …if each BSU 69 global pilot signal is transmitted as in the prior art, sector or cell capacity availability would be severely affected due to the effect on air capacity. Unlike the prior art, each BSU 69 continuously transmits a weak global pilot signal 137 approximately one half of the signal strength of a standard 32kbps POTS channel.’

For some reason, the pilot signal 137 and the two broadcast signals 139 and 141 are not shown on the figures, or more accurately the index numbers have been omitted. The ‘per BSU’ nature of the global pilot signal is clear enough.

An elaborate system is proposed for systems in which subscriber units employ battery backup, involving sleep and awakening. In this system the subscriber unit ‘knows’ to which group of BSUs it is assigned, and by maintaining known timing it ‘knows’ which BSU is transmitting when it wakes up. This does not matter for present purposes. Perhaps a little light is thrown on the claim by [0023]:

‘Each subscriber unit 25 is assigned to a set of collocated BSUs and alternately acquires each one in sequence, once per wake up period. ’

There is no suggestion in the patent that the invention is to deal with any other configuration of system infrastructure.

Somehow, the examiner considered that this disclosure supported the following claim, claim 13, which is the only claim in dispute:

‘A subscriber unit for use in a bidirectional communication system using CDMA air interface between the plurality of subscriber units communicating with a base station which transmits multiple global pilot channel signals, comprising:

means for selectively receiving a predetermined number n of global pilot channel signals from the base station such that reception of each global pilot signal is in one of n discrete time intervals, each interval for receiving a different global pilot channel signal.’

Again it is helpful to consider the various contentions on essentiality in order to identify the troublesome points of construction.

Two distinct cases are advanced for essentiality of this invention. The first, which came very late in the evidence of Professor Manikas, turns on the treatment of the Primary Common Pilot Channel (P-CPICH). The second has two aspects, and concerns the synchronisation channel (SCH). The Synchronisation Channel has primary and secondary components, and I am concerned only with the Secondary Synchronisation Channel (S-SCH). Either the invention of the patent is said to be essential for the UE’s treatment of S-SCH or it is said to be essential for what is called Time-Switched Transmission Diversity (TSTD) used with the S-SCH. TSTD is not a mandatory feature of the Standard, but UEs must be capable of dealing with such a transmission technique when it is encountered.

P-CPICH is the primary common pilot channel of the 3GPP Standard. It is defined in TS25.211 §5.3.3.1. Its primary function is to provide a known phase reference to enable the demodulation of the downlink data signals, and for the synchronisation channels. It consists of a single, known sequence spread with a specified channelization (i.e. spreading) code, which is always used. It is scrambled by the primary scrambling code. It is defined as follows. The material relating to diversity will become relevant in considering the last patent, ’777.

The features of P-CPICH set out in 5.3.3.1.1 are what matters for present purposes. P-CPICH does not carry message data. It is accepted that P-CPICH is a global pilot channel in the sense those words are used in the claim. It is also accepted that the Standard permits a single base station to cover more than one cell. Each ‘cell’ is demarcated by a particular primary scrambling code.

P-CPICH is thus an unmodulated code channel, spread with a prescribed spreading code, or channelization code, with a spreading factor of 256. It is always scrambled using the primary scrambling code, and there is only one P-CPICH per cell. It provides a phase reference for most of the other downlink channels. Thus, for each cell (or sector) P-CPICH differs only in the primary scrambling code.

A given UE may be within range of a number of base stations at any one time. Professor Manikas’ first report sets out in detail why he says that the UE infringes claim 13. He bases himself upon the mechanism of cell search and measurement, which, so far as the Standard is concerned, is indifferent as to the location of the signal source or sources received by the UE. At this point an air of unreality creeps in. Although the Standard is indifferent as to the cell/sector location, claim 13, concerned as it is with a subscriber unit for use with a modular base station, is concerned only with multiple sources of pilot signals (co)located within a single base station. Hence the ‘number n of global pilot signals’ be ‘from a base station’.

Accordingly, the setup for the purposes of Professor Manikas’ analysis must include at least one Node B transmitting more than one P-CPICH. The presence or absence of such a Node B does not affect how the standard requires the P-CPICHs to be used, but it must be present for infringement. Professor Manikas describes the process as follows:

‘200. It is well known that the signal coverage area of one cell will usually overlap with signal coverage areas of adjacent cells (typically, there may be six relatively close cells whose areas overlap with each other). This 'cell overlap' is the basis for the softer and soft handover procedures.

201.

When operating within any given area, the signal environment at the UE's receiver will be constantly changing. Signal strength from any cell transmitter will decay with distance. It will also decay in an unpredictable manner due to physical variations in the environment surrounding the UE ('slow fading').

202.

However, variations in signal strength will not be the same as between signals originating from different cells, as each signal can experience fading independently of other signals. Therefore, the cell with the strongest signal at the UE will be constantly changing.

203.

The UE therefore needs to perform measurements on a number of signals (from different cells) in order that it can select the optimum cells in its vicinity to form radio link to (i.e. that it can use in softer/soft handover).

204.

This process is referred to in the Standard as 'Cell Selection' (when the UE first switches on) and 'Cell Re-Selection' (the on-going re-assessment of signal strength by the UE).

205.

The Standard defines a number of categories for categorising the cells in the area surrounding a particular UE:

(a)

Active Set: This defines the collection of cells to which the UE is currently in soft/softer handover. The UE will have a radio link with each of the cells in the Active Set.

(b)

Monitored Set: This defines the collection of cells (excluding the Active Set) that the UE has been requested to measure the signals from [by the RRC, part of the control software in the Node B)]. These cells could become members of the Active Set if their signal levels and quality reach the required level.

(c)

Detected Set: This defines the cells that have been detected by the UE, but which are neither in the Monitored Set nor in the Active Set.

206.

Consider initially that the MS has turned on its power and that a cellular network has been selected (TS 25.122 Section 4 page 16). Then, as described in TS 125.304 page 12) the initial Cell selection process takes place. This process allows the UE to select a suitable cell to camp on. Note that camping on a cell is necessary for access to available services.

207.

However, when camped on a cell, the UE needs to search for camping to a better cell according to some criteria based on measurements (see TS 125.304 Section 5.2.3.1.2, page 17, or page 23). In this process the UE may use stored information about the cells, kept in the "Monitor set" database.

208.

If the UE is in "idle mode" this procedure should be repeated again (cell reselection). For instance, in "idle mode" this allows the UE to select a more suitable (better) cell during its wake up period and camp on it.

209.

The selection/reselection process is based on measurements carried out by the UE on CPICH signals of the various cells in its geographical neighbourhood.

210.

Then, based on these measurements, the cell selection criterion "S" given in TS 25.305 (Section 5.2.3.1.2 p17, p18) is evaluated.

211.

Thus, according to TS 125.133 page 42 Section 8.1.2.2.2 or page 55 Section 8.4.2.2.2) the UE shall be capable of performing CPICH measurements, with the measurement period of 200 ms, for up to 8 identified cells.

212.

Furthermore, the "monitored set" of cells is checked periodically by selecting for evaluation, in discrete times, each of the received pilot signals of the cells, in order to camp on a better cell.

213.

As an example consider that the monitoring set has 3 cells (n = 3) and the UE is in idle mode. In this case the UE will have to

•

monitor,

•

synchronise to a common synchronisation channel of the cell and then

•

make measurements by receiving the CPICH channel of the 1st cell in the set.

214.

This sequence of steps would then be repeated for the 2nd and 3rd cells with the whole process repeated periodically. It is important to realise that the UE cannot make these three measurements in parallel.

215.

Thus over n discrete interval all n different pilot signals have been selected, each time a different pilot, for carrying out measurements.

216.

In a "Cell" Selection/Reselection, a subscriber unit (UE) will continuously monitor the power level of the received pilot channels CPICHs transmitted from the various neighbouring Cells/BSs (covering the current position of the UE) including those in the Active Set of the UE.

…[facsimile of standard 25.304 omitted]

217.

The "Cell" selection/reselection criteria [in fact CPICH power ≥ -95 dBm] in FDD mode are described in TS 25.304 (Sections 5.2.3.1.2 and 5.2.6.1.4) and involve the power of the CPICH channels.

Conclusion: The UE is selectively receiving a number of CPICH signals and the selection criterion is associated with the power level of CPICH signals. Thus, accordingly the Standard requires means for selectively receiving a predetermined number of global pilot channel signals. Furthermore, the Standard requires that each global pilot channel signal is received in one of a number of discrete time intervals, each interval for receiving a different global pilot channel signal.’

This analysis depends for its validity on Professor Maniakis being right to say that this is both a process involving a predetermined number N of pilot signals, and that these are selectively received in one of N discrete time signals. These are the first two issues of construction.

“Predetermined” In the patent, we are concerned with one base station only, transmitting a number of global pilots with high power portions in predetermined order. This is plain from [0023] set out above:

‘The BSUs are preprogrammed to specify which BSU is selected to send its pilot at high power and which is selected to send its pilot at low power.

[0029]…a BSU pilot is always programmed to be strong when a subscriber unit wakes-up…’

In the patent, the subscriber unit selects one favoured BSU from the Igroup. Plainly the claim is only referring to the phase before an access channel is obtained (see [0032]). At this stage, the order of global pilot signals is fixed by the preprogrammed order of BSU transmissions. ‘Predetermined’ does not, in context, mean dynamically identified from the entire field of signals available to the UE. The position is not helped if one considers the use of the Monitored Set and the Active Set (see Professor Manikas’s §205 above). These are cells already identified by the UE as either in use or to be measured. They are determined initially by the UE but subsequently by RRC, the Radio Resource Controller associated with the Node B. In addition to these, there is the Detected Set, which is the list of cells actually identified. All are identified by measuring CPICH. It is certainly true that the UE must be capable of detected and reporting on 8 CPICHs within 200ms, but which these are is not predetermined.

“Selectively receiving” This is really the complementary phrase to ‘predetermined’. The global pilots which are not selected are those which are not predetermined. So in my view this feature is not present in the 3GPP Standard.

“Discrete time intervals…for receiving a different global pilot channel signal” Again, in the patent the concept of discrete time intervals is closely tied into the predetermined global pilot channel signals, so each is identified implicitly by the time at which it is received.

‘[0027] The time of day…is converted to the identity of one BSU…Both the BSU and all subscribers of Igroup know which BSU will be broadcasting at a specific time…’

So far as the Standard is concerned, the only feature which is said to satisfy this requirement is the requirement that 8 P-CPICHs will be measured in the 200ms period and the argument is that it is likely that if this is done by a single correlator the measurements will be done in series; and, if the measurements are done in series each measurement will take a finite time; and finite time intervals of a particular duration are ‘discrete’ (the definition advanced of a discrete time interval is one ‘with a defined start and end’). I think this deprives ‘discrete’ of any of the sense it possesses when the claim is read in context. In context, a ‘discrete’ period is one during which a particular global pilot signal was transmitted at high power. This has its explanation in the final words of the claim ‘for receiving a different global pilot channel signal’. There must be a relationship between the discrete time interval and a corresponding global pilot signal: to one discrete interval belongs a particular global pilot signal. There is nothing remotely like this called for by the 3GPP Standard.

The argument on P-CPICH fails. I turn to the alternative argument, on the synchronisation channel, SCH.

SCH is a synchronization signal, and it is described thus in the Standard:

‘The Synchronisation Channel (SCH) is a downlink signal used for cell search. The SCH consists of two sub channels, the Primary and Secondary SCH. The 10ms radio frames of the Primary and Secondary SCH are divided into 15 slots, each of length 2560 chips. Figure 18 shows the structure of the SCH radio frame.

The Primary SCH consists of a modulated code of length 256 chips, the Primary Synchronisation Code (PSC) denoted c_p in figure 18, transmitted once every slot. The PSC is the same for every cell in the system.

The Secondary SCH consists of repeatedly transmitting a length 15 sequence of modulated codes of length 256 chips, the Secondary Synchronisation Codes (SSC), transmitted in parallel with the Primary SCH. The SSC is denoted c_s^i,k in figure 18, where i = 0, 1, …, 63 is the number of the scrambling code group, and k = 0, 1, …, 14 is the slot number. Each SSC is chosen from a set of 16 different codes of length 256. This sequence on the Secondary SCH indicates which of the code groups the cell's downlink scrambling code belongs to.

The primary and secondary synchronization codes are modulated by the symbol a shown in figure 18, which indicates the presence/ absence of STTD encoding on the P-CCPCH and is given by the following table:

’

Neither of the two SCH channels is scrambled. SCH is not referred to as a pilot channel by the Standard. The priority date of the patent is comparatively late, but the first question is what is meant by ‘global pilot channel’ in the claim. I shall assume, without deciding, that this phrase may cover SCH, largely because the argument is sterile, given that each of the ac_p shown in the Primary SCH would have to be equated to the global pilot signals of the claim. They cannot be so equated: even if the SCH is to be considered a pilot, it occupies the whole of a frame. So instead InterDigital concentrate on the Secondary SCH, where the code fragments in the slots are each said to be global pilot channel signals.

In more detail, the primary SCH is just a repeated sequence of 256 chips, identical in every slot. This enables a degree of synchronisation with the cell without any knowledge about the cell at all. The secondary SCH enables the code group used by the base station to be identified and frame synchonisation to be obtained. At the beginning of each frame, there is one of 16 possible 256-chip codewords, and it is the order of these codewords that defines the code group by enabling 64 different code words to be transmitted in the frame. CPICH (which is spread with a known spreading code) is then used to identify the actual scrambling code from the 8 possible members of the group.

I do not think that the phrase ‘global pilot channel signals’ can be used to refer to the SSCs, that is, the individual codewords, because it is not these alone which are the signal. The Secondary SCH depends upon both the order and the selection of the synchronization codes. As the Standard says,

‘Each SSC is chosen from a set of 16 different codes of length 256. This sequence on the Secondary SCH indicates which of the code groups the cell's downlink scrambling code belongs to.’

It is artificial to consider the Secondary SCH as being anything other than a single code repeated on a per-frame basis, which is after all how the Standard describes it. I really regard the illustration used by InterDigital to illustrate their contention as extremely misleading.

This is intended to show the similarities between the successive transmission of different global pilot signals in the patent (BSU1 etc) with the SSCs at the beginning of every frame. But the SSCs are part of a sequence which sends a broadcast message to every UE which decodes it. They are not distinct pilot signals, if they are pilot signals at all. They are a single signal, carrying information.

Finally, I must consider the impact of Time-Switched Transmit Diversity. Diversity is a topic to which I shall return in more detail in considering ’777, but for present purposes it is only necessary to explain that diversity means that the signal of interest is transmitted from more than one antenna sufficiently spaced apart. The Standard only contemplates antenna diversity involving two antennae, and the kind of antenna diversity under consideration is Time-Switched Transmit Diversity, or TSTD. The figure from the Standard demonstrates the point:

‘5.3.3.5.1 SCH transmitted by TSTD

Figure 19 illustrates the structure of the SCH transmitted by the TSTD scheme. In even numbered slots both PSC and SSC are transmitted on antenna 1, and in odd numbered slots both PSC and SSC are transmitted on antenna 2.

’

This argument fails for the same reason as the argument on SCH alone failed. The Primary SCH signal does not consist of more than one global pilot channel signal, if it can be described as a pilot channel at all, and the position is not changed by its being transmitted by the same transmitter through more than one antenna. I do not understand why it matters that the UE might, in principle, be able to detect which slot came from which antenna. But the SSCs together still constitute one single signal, repeated from frame to frame. Thus, to identify the primary scrambling code group, all 15 of the slots of Secondary SCH need to be processed, so that P-CPICH can be descrambled and phase synchronisation be obtained. Half of them will come from antenna 1 and half from antenna 2.

In my view, therefore, claim 13 of this patent is certainly not essential to the Standard.

This patent is entitled ‘Transmission using an antenna array in CDMA communication system’. It is concerned with antenna diversity. Its priority date is 10 September 1999.

The nature of the case being advanced by InterDigital on this patent changed substantially during the hearing. Indeed there was a very large amount of discussion on figures 7-11 of the patent, which are hardly referred to in Professor Manikas’s report, and the significance, which I shall discuss below, of the opening words of Section 7.1 of TS 25.214, on which a considerable amount was made to depend, did not appear at all. Ultimately this caused a number of serious problems. Neither Professor Manikas nor Professor Purat discussed channel estimation, the subject of the point, in any detail in their respective reports. Both were experienced speakers of English, but both used English in a slightly non-standard way, and it was particularly important that the points of dispute were clearly identified in writing, and that written evidence addressed to the particular points could be delivered. In the end, Professor Manikas gave a lot of mathematical evidence from the witness box. I only allowed this to proceed because there was no objection to it. In the end a further statement of case was filed by InterDigital accompanied by a further report (his fourth).

In short, the patent is concerned with a base station which uses an antenna array to transmit its signal to the UE. Each of the antennae transmits a ‘version’ of a pilot signal. Every UE receiving these pilot signal ‘versions’ weights them and combines them, setting the weights to give a higher quality of received signal than one of them would provide on its own. The weights so obtained are applied at the UE or at the transmitter to the data signals to improve their quality of the data signal destined for a particular UE. At the UE each of the received versions of the signal is weighted (if not weighted at the transmitter) and combined with the others, to provide a signal of improved quality.

The task of construing the specification is further complicated by the fact that it is terse and superficial, is admittedly in error in certain respects and contains no clear indication whether the time division duplex embodiments (TDD) are within the claims at all. It is difficult to construe the claims so that they are coherent and cover the described embodiments.

The specification begins ([0002]) with a description with reference to figures 1 and 2, with a description of prior art CDMA communication systems. This contains what appears to be a conventional description of CDMA itself, but in [0004] deals briefly with timing synchronisation by use of a pilot signal transmitted from the base station. After a description in [0006] of adaptive power control, [0007] introduces ‘sectorization’:

‘[0007] Although adaptive power control reduces interference between signals in the same bandwidth, interference still exists limiting the capacity of the system. One technique for increasing the number of signals using the same radio frequency (RF) spectrum is to use sectorization. In sectorization, a base station uses directional antennas to divide the base station's operating area into a number of sectors. As a result, interference between signals in differing sectors is reduced. However, signals within the same bandwidth within the same sector interfere with one another. Additionally, sectorized base stations commonly assign different frequencies to adjoining sectors decreasing the spectral efficiency for a given frequency bandwidth.’

After a summary of the cited prior art, the introductory part of the specification concludes at [0011] with the statement that there exists a need for a system which further improves the signal quality of received signals without increasing transmitter power levels.

The ‘Summary of the Invention’ at [0012] is merely a summary of the claim. It is convenient to set out the claim with the features labelled as proposed in exhibit X2. These labels were used throughout the trial to refer to its features.

A.

A method for use in a spread spectrum communication system having a plurality of transmitting antennas (48-52), the method comprising the steps of:

B.

transmitting from each transmitting antenna (48-52), a pilot signal having a pseudo random chip code sequence uniquely associated with that antenna (48-52);

C.

receiving at the receiver all of said transmitted pilot signals;

D.

filtering each said transmitted pilot signal using that pilot signal’s pseudo random chip code sequence;

characterized by

E.

weighting each said filtered pilot signal by a particular weight;

F.

combining said weighted pilot signals as a combined signal;

G.

adaptively adjusting each said pilot signal’s particular weight based in part on a signal quality of the combined signal;

H.

transmitting a data signal such that different spread spectrum versions of the data signal are transmitted from each antenna (48-52), each version having a different chip code identifier for the respective transmitting antenna (48-52); and

I1.

receiving the data signal via filtering each version with its associated chip code and combining the filtered versions,

I2.

wherein the different data signal versions are weighted in accordance with the adjusted weights associated with the pilot signal of the respective antenna (48-52).

This is a method claim. There is a corresponding apparatus claim, claim 13, but it is more difficult to construe while raising no new points of substance. All the discussion at trial revolved around claim 1.

The invention is concerned both with transmission from an array of antennae at a base station of a pilot signal and a data signal, and the manner in which the data signal is to be treated having regard to measurements carried out on the pilot signal. The figures are of particular importance. Figure 3 shows a transmitter of the invention, with a single data signal source. This is not particularly realistic where mobile telephony is concerned, and Figure 4 shows a corresponding transmitter with M data sources, which might be capable of accommodating M users.

In the transmitter of Figure 3, the pilot signals produced by the pilot signal generators 56-60 are spread with different chip code sequences. The data signal is spread by a different chip code sequence (D₁..D_N) for each antenna, so there are as many chip code sequences as there are antennae. The spread signals are modulated using some appropriate technique onto carriers and supplied to the antennae. The reason for using multiple antennae is described in the specification:

‘[0015] By using an antenna array, the transmitter utilizes spa[t]ial diversity. If spaced far enough apart, the signals radiated by each antenna 48-52 will experience different multipath distortion while travelling to a given receiver. Since each signal sent by an antenna 48-52 will follow multiple paths to a given receiver, each received signal will have many multipath components. These components create a virtual communication channel between each antenna 48-52 of the transmitter and the receiver. Effectively, when signals transmitted by one antenna 48-52 over a virtual channel to a given receiver are fading, signals from the other antennas 48-52 are used to maintain a high received [signal-to-noise ratio]. This effect is achieved by the adaptive combining of the transmitted signals at the receiver.’

Figure 4 of the patent is a version of Figure 3 but capable of transmitting multiple data signals. Each data signal is separately spread for each antenna, and so if there are N antennae and M data signals N×M separate spreading codes are required. The signal radiated by each antenna is a combination of M data signals, each modulated by a different spreading code.

The description then moves to reception of the pilot code. Figure 5 is the relevant diagram. [0017] describes this diagram. Before considering what it discloses, it is helpful to consider what is meant by RAKE in the boxes 82, 84, 86. RAKE is not an acronym, but the name for a particular type of receiver used for despreading a CDMA signal when multipath effects are present, which they always are. Multipath effects are the effects caused by multiple reflections along the path between the transmitter and receiver, the transmission path. The reflected signal can interfere with itself; and the result is that there may be sudden and rapid fading and phase distortion. RAKE receivers were at the priority date a common general knowledge technique for receiving CDMA signals with multipath distortion. The underlying idea is to treat the received signal as a direct undistorted signal to which have been added a number of delayed versions of itself, the multipath components. The RAKE receiver extracts the delayed components of the signal, removes the delay and combines them coherently. The relative contributions of the recovered multipath components are important to the noise performance of the receiver.

Such a description of the distorted channel suggests that the signal may be recovered by delaying the received signal and adding the delayed versions. This is illustrated in three figures taken from Richardson, The WCDMA Design Handbook.

First, Figure 6.8 shows the effect of multiplying a single spread signal by the spreading code. In the delay domain, there is one component. In the time domain, the spread signal is a series of +1 and -1. In the frequency domain, the spectrum is well spread out. The correlator (often shown as ⊗) multiplies the spread data as received with the spreading code properly aligned: the result is a signal with a narrow frequency spread but a higher power.

Figure 6.9 shows what happens if two versions of the identical signal are received, one delayed by a time dt with respect to the other. Both have the same low peak power wide spectrum. If the despreading code is aligned on the first it will be decoded as one would expect. The second is not decoded, and merely contributes to the noise level. It is not shown in the diagram, but each correlator is followed by a sum-and-dump circuit or integrator (paragraph 13 above), which will remove the unspread noise component.

If two correlators are used to decode the two signals and the despreading signals supplied to the correlators are correctly aligned, then, if the outputs of the correlators are combined coherently (i.e. in phase) then there will be two noise components as a result of the correlation between the despreading codes and the ‘other’ signal, but the two correctly despread signals combine to make a high peak power narrow spectrum output. The noise is again reduced by the sum-and-dump circuit leaving contributions from both multipath components.

The method of combining which adds the two signals coherently has a particular advantage. One approach to the problem of multipath is merely to select the detected component which has the greatest amplitude at the correlator output—this is called ‘selection diversity’. But if coherent combining of the despread signals is used, an advantage in signal to noise ratio can be obtained. The mathematics doesn’t matter: but it can be shown that in the best case the addition of two equal noisy signals, where the signals are coherent (i.e. correlated with each other) and the noise is not, doubles the signal to noise ratio (i.e. improves it by +3AB). This is called equal gain combining. Further improvement may be obtained if instead of combining the signals equally they are combined in proportions that reflect the some quality of the signals, perhaps their amplitude as received. This is called maximum ratio combining. Professor Manikas identified an article in Proc IRE from 1959 which set all this out: I think that it is either common general knowledge or at least something which any competent radio transmission engineer confronted with the problems of multipath ought to be aware of. A simplified diagram of a RAKE receiver as used in WCDMA employing maximum ratio combining of multipath components is Figure 6.13 from Richardson:

This deserves a little explanation. r(t) is the incoming received signal. The boxes labelled t₁..t_n represent delays of duration t_i. Each ‘finger’ of the rake except the first accordingly processes a delayed version of the received signal. First, the signal is correlated with (multiplied by) u*(t), which is a combination of the spreading and scrambling codes. Then it is multiplied by a weight factor , which is obtained by ‘estimating’ the channel. Because in general both amplitude and phase are affected by the channel, the impulse response is conveniently modelled by the use of complex quantities, and the weight factor is also a complex quantity. The * signifies that u^* is the complex conjugate of the combined spreading and scrambling sequence and a^* the complex conjugateof the channel impulse response component corresponding to delay t_I (a). The complex conjugate must be used to obtain the magnitude of the complex signal. The box at the end combines (+) and sums-and-dumps () the combined signal.

Channel estimation is the last aspect of this. The ‘impulse response’ of the channel describes the effect that the channel has on a transmitted pulse (the narrower the pulse, the more frequencies it contains). Here, if one transmits a spread simple, known signal (say all 0s) and despreads it using a RAKE arrangement without applying a weight, the output of the fingers will be the impulse response of the channel. In fact, the estimating process is complex because the delays between the multipath components are not known and must therefore so far as possible be determined as well. But ignoring this aspect (which is irrelevant to this patent) the function of channel estimation is to determine the weight factors for maximum ratio combining to be used.

I can now return to the specification of the patent. [0017] describes Figure 5. Its language is confused, but it seems plain that it is contemplating that in the pilot receiver of Figure 5 either a second weighting device may be used on the output of each RAKE, or that the weighting devices necessary for the RAKE may be employed also together to weight the signals from the different antennae.

‘[0017]…Each of the transmitted pilot signals is received by the antenna 80. For each pilot signal , a de-spreading device, such as RAKE82-86 as shown in the Figure 5 or a vector correlator is used to despread each pilot signal using a replica of the corresponding pilot signal’s [PN]. The despreading device also compensates for multipath in the communication channel. Each of the recovered pilot signals is weighted by a weighting device 88-92. Weight refers to both magnitude and phase of the signal. Although the weighting is shown as being coupled to a RAKE, the weighting device preferably also weights each finger of the RAKE.’

As the specification makes clear, the purpose of the use of multiple antennae is to obtain spatial diversity, or transmit antenna diversity. The principles underlying transmit antenna diversity are very similar to multipath diversity. In each case, the received signal has components from a number of paths that will differ one from another in phase and amplitude. They are demodulated to separate them and their relative phase difference is removed so that they can be coherently combined (if selection diversity is not to be used). In a multiple transmit antenna system, it may be necessary to distinguish between the signals from the different antennae at the receiver. In the preferred embodiments of the patent, there is always either a per-RAKE weight w_i and the ‘internal’ RAKE weights w_i,j are just the same as the multipath weights for that single antenna (figures 10 and 11), or there are only weights w_i,j which have to do double duty, both multipath and per-antenna weighting (Figures 7 and 9).

This fact led to a passage in cross-examination on Figure 7 that was not technically clear and has resulted in a criticism of Professor Purat which I do not consider justified.

Mr Birss concentrates on a particular answer given by Professor Purat at transcript 664 and says it is ‘simply wrong’ adding that

‘Indeed elsewhere in his oral evidence, when his eye was not on fig 7, he seemed to have no difficulty with the idea that generating and refining a set of weights from pilot signals—in other words generating the impulse response—is indeed channel estimation.’

Professor Purat accepted this because it is true: but it is not what is in Figure 7, in which the weights w_ij are affected not only by the channel characteristics associated with the particular antenna with which the particular RAKE is associated (i.e. antenna number i) but by the characteristics of other channels as well. This is made clear by X12, a copy of Figure 7 with additional lines showing the flow of weighting information from the ‘Weight Adjustment Device’ 170 to the weighting multipliers (Footnote: ²) which was approved by Professor Manikas. In my judgment, Professor Purat was not wrong: he knew exactly what channel estimation was. This appears from transcript 666:

‘Q. I want to consider two implementations, both of which are seeking to perform channel estimation. A. Right.

Q. OK, so far so good, obviously. The objective in both cases is to derive an estimate of the channel which will be used to receive a data signal. OK? A. OK.

Q. Do you accept that as a reasonable working basis to move forward? A. Yes, then we are not talking again now about only channel estimation; we are talking about receiving a data via other channel, OK.’

I think it is right to view this as Professor Purat being careful about what is shown in Figure 7. That is all.

Turning now to the determination of the weights, it is clear that the manner of their determination is a standard one:

‘[0017]…After weighting, all of the weighted recovered pilot signals are combined in a combiner 94. Using an error signal generator 98, an estimate of the pilot signal provided by the weighted combination is used to create an error signal. Based on the error signal, the weights of each weighting device 88-92 are adjusted to minimize the error signal using an adaptive algorithm, such as least mean squared (LMS) or recursive least squares (RLS). As a result, the signal quality of the combined signal is maximised.’

The per-finger weights and global per-antenna weights (if used) so determined are applied to the internal mixers or multipliers in the RAKEs and to the per-RAKE multipliers (if any) of the corresponding data receivers.

Figure 9 is a data receiving circuit used with the pilot signal receiver of Figure 7. It is expressly stated to be for use in a base station (col. 6 l. 15) and there is no reason why a single UE should receive more than one data signal. In figure 7, it is the spreading codes that are delayed to detect the multipath components in the pilot but in the receiver the received signal is delayed: this means that the delay line signal flow should be reversed from what is shown.

Professor Purat thought it would not work and Professor Manikas thought it would (slightly surprisingly). The principle of operation, however, is clear enough, and it can be transposed to a single user equipment by despreading only using the set of despreading codes for a single user i and M antennae, i.e. D_i,j, j=1..M.

Figure 10 raises a different problem.

As I have indicated, this is the embodiment of a pilot signal receiver in which both the individual RAKEs have internal weights and their outputs are weighted before they are combined. Each of the RAKEs uses the same amplitude and phase reference (this is shown as 1+j0, a complex number whose imaginary part, denoting phase, is zero) at the input to the three subtractors which provide input to the ‘Weight Adjustment Device’ to fix the weights. Because they all use the same amplitude and phase reference, the outputs of the combiners 162, 164 and 166 will be in phase with each other, and there will, in Professor Purat’s view, be nothing for the final weights (w₁, w₂ etc) to do. Professor Manikas disagreed. I asked him a number of questions to try and obtain an explanation for this. My conclusion is that if each of the RAKEs uses the same phase reference, then their outputs will be brought into phase and there will be nothing for the second weights to do. They will all have the value 1+j0. If the phase references are not the same, derived for example from the strongest multipath component from the particular antenna, then phases between different antennae, which may additionally be in different cells if the system has soft handover, will have to be compensated for and the second weights then become meaningful.

Figures 12 and 13, which are essentially simplified circuits for the case in which the antennae are so close together that their multipath environment, and hence the internal RAKE weights, are the same do not, I think, add anything to the discussion. Nor do I think that there are any general observations which need to be made about the beam-forming examples. In the beam-forming case, the pilot receiver derives the per-RAKE weight, which is then transmitted to weight the antenna to which that RAKE corresponds. This will tend to ensure that the UE is located at a maximum in the interference pattern resulting from same-frequency transmissions from more than one antenna.

At this point, it is helpful to summarise what has been disclosed. The preferred embodiment uses RAKE receivers to receive the pilot signal, one per transmitting antenna. The outputs of the RAKEs are themselves weighted to eliminate phase and amplitude variation and combined. The weights are determined using algorithms with which the addressee must be familiar, since they are not described in detail. Alternatively, there is no second weighting, the weighting taking place at the separate fingers of the RAKEs only, the outputs of the RAKEs being immediately combined (Figure 7). Either way, the weights so determined are applied to the corresponding multipliers of a data receiver complementary in structure to the (Figures 11 and 9).

Again, it is convenient to consider the 3GPP Standard before considering the particular points of construction that arise. TS25.211 §5.3.1 is the only part of the Standard which refers to weights in the context of downlink transmit diversity, and this refers to Closed Loop Mode 1, to which InterDigital’s case is limited. Closed Loop Mode 1 is optional for a Node B transmitter, and the ability to deal with it is accordingly mandatory for the UE. The relevant part of the Standard defining Closed Loop Mode 1 is 25.214 §7, which begins as follows:

The general transmitter structure to support closed loop mode transmit diversity for DPCH transmission is shown in figure 3. Channel coding, interleaving and spreading are done as in non-diversity mode. The spread complex valued signal is fed to both TX antenna branches, and weighted with antenna specific weight factors w1 and w2. The weight factor w1 is a constant scalar and the weight factor w2 is complex valued signal. The weight factor w2 (actually the corresponding phase adjustment) is determined by the UE, and signalled to the UTRAN access point (i.e. cell transceiver) using the FBI field of uplink DPCCH.

DPCH is the dedicated data channel, that is, the per user data channel, and uplink DPCCH is the uplink Dedicated Physical Control Channel. DPCH has two parts, DPCCH and DPDCH (Dedicated Physical Data Channel).

‘For the closed loop mode 1 different orthogonal dedicated pilot symbols in the DPCCH are sent on the 2 different antennas.

Then, the following appears in section 7.1, General Procedure:

‘The UE uses the CPICH to separately estimate the channels seen from each antenna.

Once every slot, the UE computes the phase adjustment, φ, that should be applied at the [Node B transmitter] to maximise the UE received power.’

Finally, it should be noted that the structure of downlink DPCH includes pilot bits, (§25.211 5.3.2) :

In closed loop mode 1, there may be 4, 8 or 16 pilot bits, and the patterns from the two antennae shall be orthogonal:

Accordingly, the Standard requires that the UE be capable of dealing with a transmitter with two antennae, sending a dedicated physical channel in which the signal from each antenna differs in the pilot bits in every slot, accompanied from every antenna by CPICH₁ (all zeros) and CPICH₂ (the prescribed pattern). The spreading and scrambling codes for the data at each antenna are the same. Both the scrambling and spreading code for the secondary CPICH may differ from that of P-CPICH. Weights are not applied at the transmitter to the CPICHs.

The feedback element of the system consists of the supply of power and phase weights in the prescribed form of command message sent in uplink DPCCH by the UE. The way in which these weights are derived, their form, and the way in which they are processed in the base station so as to control the transmitter is not directly material. It is necessary to note that the weight applied to antenna 1 is constant (1/√2) and the only weight that is adjusted is w₂, which is adjusted according to a formula

It is now possible to return to the claim. I set it out again for convenience:

A.

A method for use in a spread spectrum communication system having a plurality of transmitting antennas (48-52), the method comprising the steps of:

B.

transmitting from each transmitting antenna (48-52), a pilot signal having a pseudo random chip code sequence uniquely associated with that antenna (48-52);

C.

receiving at the receiver all of said transmitted pilot signals;

D.

filtering each said transmitted pilot signal using that pilot signal’s pseudo random chip code sequence;

characterized by

E.

weighting each said filtered pilot signal by a particular weight;

F.

combining said weighted pilot signals as a combined signal;

G.

adaptively adjusting each said pilot signal’s particular weight based in part on a signal quality of the combined signal;

H.

transmitting a data signal such that different spread spectrum versions of the data signal are transmitted from each antenna (48-52), each version having a different chip code identifier for the respective transmitting antenna (48-52); and

I1.

receiving the data signal via filtering each version with its associated chip code and combining the filtered versions,

I2.

wherein the different data signal versions are weighted in accordance with the adjusted weights associated with the pilot signal of the respective antenna (48-52).

The claim is concerned with the processing both of the pilot signal and of the data signal, and with the manner in which the weights derived from measuring the pilot signal are applied to the data signal. I have attempted to show this by marking the appropriate parts of the claim. The following points arise when considering Closed Loop Mode 1 Transmission Antenna Diversity.

Another way of looking at these problems is this. Both experts were agreed that the 3GPP Standard has very little to say about the internals of a UE, prescribing signal structures and timing relationships together with performance standards. So, when the 3GPP Standard says, in the opening words of 25.214 §7.1, that the UE uses the CPICH to separately estimate the channels seen from each antenna, is it assuming that the UE will use some way of approximating the individual channels using CPICH as a pilot so as to have available per finger weights? It is assumed by the 3GPP Standard that a UE will use a RAKE receiver, but I am not satisfied that this is mandated. Professor Manikas said that they gave you an indication of what kind of architecture is needed. What is mandatory is the performance which has to be achieved. Does the Standard mandate any particular treatment of the pilot signals in the receiver, and if so does that method satisfy features E, F and G of the claim? How does the Standard require the data signal to be processed?

The central part of the whole dispute lies with features E, F and G. In context, I consider that their meanings are perfectly clear. Feature E talks of weighting filtered pilot signals. The word ‘filter’ is apt to describe a RAKE or a vector correlator: see the description in [0017] and [0032]. Whether or not this is performed in a separate weighting device, it is the combined signal (Feature F) which is used to provide the error signal: that is, the signal at the input to the error signal generator 96 in Figure 5 is the ‘combined signal’ ([0017]), and the signal at the input of the subtractor 168 in Figure 7 is the combined signal ([0020], col 6 line 1). In Figure 10 the despread pilot signals are weighted in 256-260 ‘prior to combining. After combining, the combined signal is compared with the ideal value’, obviously in 168 again. In Figure 12, the despread pilot signals are combined in 94 after the sum-and-dump circuits 308 and 310. Figure 12 incidentally shows that despreading may take place before or after weighting in the RAKE fingers.

It follows inevitably that the weights of Features E, F and G are not concerned with channel estimation of the individual channels between the receiver and the transmitting antennae so as to deal with per channel issues such as multipath distortion. The weights allow for the difference between the pilot signals viewed as recovered signals, as [0017] makes clear in referring to weighting the ‘recovered pilot signals’.

I have found the argument to the contrary extremely difficult to understand. It may be that it is best summed up in the question set out in InterDigital’s skeleton argument, ‘What is the purpose of the combination of pilot signals?’. The answer is, to take advantage of antenna diversity. It is crucial that the receiver be able to discriminate between the different signals it is receiving if it is to combine them using maximum ratio combining, just as in the multipath example described above the multipath components are distinguished by their different delays and so the phase of the despreading code which detects them. Even if the recovered pilot signals of features E, F and G are already in phase, it is because they were transmitted using different spreading codes and so the pilot from Antenna i appears at the output of RAKE i. So weight w_i will be adjusted so as to produce maximum ratio combining and produce a maximum signal-to-noise ratio.

If the weights obtained for the pilots are to be used for the data signals, the receivers must be similar: the weight w_i must be applied to the data signal appearing at the output of filter (RAKE) i, so that the data signal recovered ‘via filtering’ is weighted in accordance with feature I2 of the claim.

The means, in turn, that the data signals will have to be marked with their origin so that the data signal from Antenna i goes to filter i in the receiver. The way this is done in the patent is to use chip codes which identify both the data signal and its antenna—it will be remembered that the preferred embodiments call for NM different spreading codes for N data signals and M antennae (column 4 lines 43-4).

The foregoing consideration provides what I think is a reasonably sensible interpretation of the final part of feature H of the claim, ‘each version [of the data signal] having a different chip code identifier for the respective transmitting antenna’. So far as this invention is concerned, the use of one filter per antenna necessarily involves some method of maintaining the association, and the method adopted is to use separate chip codes to identify the different antennae. This feature of the claim is very odd in its phraseology, being used nowhere else in the specification or claims. I do not think ‘chip code identifier’ is a noun phrase. Rather, ‘identifier for the respective transmitting antenna’ qualifies ‘chip code’.

There was a long dispute as to whether different chip codes were in fact needed to be applied to the different versions of the data signal. I accept that if the symmetry of treatment of the pilot and the data code may be broken, as in a closed loop method of transmission where the intention is to eliminate phase differences between the signals received by the receiver from the different antennae, it would not be necessary to use different spreading codes for the data signal. This is because the w_i will have been sent to the base station and the power and phase of the signals supplied to the antennae adjusted accordingly. But I do not understand how the individual internal rake weights (necessary to compensate for multipath distortion) can be applied to the received signals if they are derived from separate measurements on each incoming pilot unless corresponding distinct filters for the data signals are present in the receiver. What the standard calls the ‘antenna weight’ has to be derived and applied separately from what I shall call the multipath weight.

Returning now to the Standard and ignoring features E to G of the claim, it seems to me clear that features I1 and H are not present or mentioned in the 3GPP Standard, and cannot be essential. These features require the presence of the chip code. The orthogonal sequences in the DPDCH (paragraphs 115 and 116 above) were pressed into service for the purpose of suggesting that at least there was something there which identified the antenna, which might with a fair wind be viewed as a chip code identifier, but these are no such thing. The Standard does require them to be used for antenna verification, and states that they may be used for verifying frame synchronisation. They have nothing to do with identifying the signal for any other purpose, let alone identifying a chip code. One only has to consider an antenna transmitting two DPCHs destined for two different UEs. These will be transmitted with different chip codes, but these orthogonal pilot codes will be the same. Which chip code do they identify?

For these reasons alone, I am quite satisfied that this claim (and all the claims dependent from it) are not essential to the 3GPP Standard.

The question of features E, F and G is rather more difficult. As the 3GPP Standard is drafted, it seems to me that features E, F and G are not present because no weights are ever applied to the pilot signals in the transmitter. In the receiver, all the Standard says (I shall assume this is mandatory) is that the primary and secondary CPICHs are separately employed to estimate the channel. What if the process of estimation involves separate weighting of the recovered pilot signals from the two antennae and their combination as part of the process of determining the impulse response of the transmission channel, which for this purpose includes both antennae. This is close to the process suggested in Appendix A2:

In non-soft handover case, the computation of feedback information can be accomplished by e.g. solving for weight vector, w, that maximises.

where

and where the column vectors h1 and h2 represent the estimated channel impulse responses for the transmission antennas 1 and 2, of length equal to the length of the channel impulse response. The elements of w correspond to the adjustments computed by the UE.

This expression for the power P is, or may be, a substantial expression. Professor Manikas was of the view that as a practical matter the necessary maximisation could not be achieved without maximum ratio combining in the pilot receiver. This would involve taking the initial channel estimate and fixing the weights by some iterative technique based on some criterion, although he does seem to have accepted that maximum ratio combining was possible without adaptive adjustment of the weights, so potentially rendering feature G unnecessary. Professor Purat considered that it was possible to take an initial channel estimate but do no further processing, or possibly some processing without feedback: his figure 63 in his fourth report shows such a system. He says that iterative adjustment of pilot signal weights is optional. I think that Professor Manikas’s argument was really one of practical necessity: but the performance criteria in the 3GPP Standard were not referred to, and I do not know whether they are relevant or not.

As Mr Birss made clear in opening this part of the case, a great deal of this argument depends on the opening sentence of TS 25.114 §7.1. I think that if it was to support the great weight which was ultimately put on it, it was necessary clearly to state that compliance with these words necessarily involved using the features E, F and G of the claim. My clear impression was that this was not an inevitable mandatory requirement, but at most a question of practical necessity, and I was not satisfied of its practical necessity. The effective silence of the 3GPP Standard on processing within the UE makes it obligatory to produce such an argument faced with features in the claim like this, and I cannot hold that these features of the claim are essential either.

I consider that three out of the four patents in issue before me are not essential to the standard. As I have indicated, I consider on the authorities that the only question is whether the evidence establishes that a declaration to this effect, coupled with the refusal of InterDigital to step up to the base in respect of a number of the patents in issue, would have practical utility. This is a matter for decision on the evidence adduced at trial: what was said in the Court of Appeal in the judgment of 5 December 2006 was not, I accept, evidence, although I am satisfied that it reflects the common sense of the situation.

The only witness to give evidence under oath was Mr Richardson. As he said, and as one would expect, the factors taken into account in a licensing negotiation are manifold and there would be little point in trying to list each of them appropriately weighted. The real question, as it seems to me, was whether findings by a single judge sitting in England after a full trial with competent experts will be relevant either to the scope of any licence or, given that a licence is negotiated for geographical areas encompassing many independent states, whether those findings would be a factor which will be taken into account in fixing the royalty to be paid.

One point can be taken out of the discussion immediately. This is so-called ‘Limb 2’ essentiality, which arises when the only way of complying with the Standard in a particular respect involves the use of either one or another patented invention. In such a case, both patents are to be taken to be essential. I was not told whether this has ever arisen. Nor was it suggested, however faintly, that it may be the case with any of the patents in this case. It is clear that InterDigital were conscious throughout this action of this possibility yet declined to raise any positive case. I have already said in interlocutory hearings that ‘Limb 2’ must be raised in this action. It is now too late to employ the mere possibility of a ‘Limb 2’ case as a ground for refusing declaratory relief.

The next question is the question of evolving standards. Whether the standards have evolved in any relevant respect is a matter I should have expected to be ventilated. Version 7 of the Standard was in Court but was not referred to. This point too should be disregarded.

Finally, there is the question of infringement by particular constructions of mobile phone. In respect of the two patents I have held to be inessential, this is not because the features I have identified as inessential might nevertheless be used, but because they are not consistent with the Standard. (This might not be the case in respect of features E, F and G of ’777: but these are not central to my finding of inessentiality.)

The remaining matters raised by InterDigital—opening of floodgates of litigation, the impact of declaratory relief upon ETSI, the tactical nature of the litigation—have all been considered by the Court of Appeal in declining to dismiss the action. They are not sensitive to the facts established in a particular action, and I will not consider them again.

So I come to Mr Richardson’s evidence. I thought he was a frank and straightforward witness, although his experience of the particular type of licensing underlying the dispute between the parties to this action was not wide. He had never negotiated a licence against a background of obligatory licensing. Mr Thorley QC submits that his evidence established the following:

Of these, perhaps the most directly relevant are number vi and number viii.

Mr Watson suggested these answers, and particularly the second, had been given upon a false assumption as to the relative sizes of the relevant markets in the UK and Europe on the one hand and the US on the other, where not only WCDMA but other CDMA systems had also to be considered. I do not think that this is primarily a size of market question, although that is relevant. The notifications with which I am concerned are ETSI notifications, which involve the prospect of an obligatory licence. I do not even know if that is also the position in the US. As a practical matter, of course the parties will take a general approach, but once it is established that the UK market is very far from insubstantial I think Mr Richardson’s answers are what you would expect. The decision is material, but may not be determinative. For the UK market alone, of course, subject to what may happen in Scotland, it is determinative. So I think that the declarations proposed are genuinely useful, and I shall hear counsel as to their form.

The action succeeds to the extent indicated. I wish to add a word about the witnesses. Both Professors Purat and Manikas spoke English, a second language, well, and their expertise was wide. Both were heavily criticised. I do not think the criticisms were justified, although I have preferred one or other on occasion. I am grateful to them both and to Professor Marshall, and I am grateful also to Mr Wiffen who introduced me, as I have said, to the subject.