England and Wales High Court (Patents Court) Decisions
You are here: BAILII >> Databases >> England and Wales High Court (Patents Court) Decisions >> Adaptive Spectrum and Signal Alignment Inc v British Telecommunications Plc [2013] EWHC 3768 (Pat) (03 December 2013)
URL: http://www.bailii.org/ew/cases/EWHC/Patents/2013/3768.html
Cite as: [2013] EWHC 3768 (Pat)
[New search] [Printable RTF version] [Help]
CHANCERY DIVISION
PATENTS COURT
7 Rolls Buildings Fetter Lane London EC4A 1NL |
||
B e f o r e :
____________________
| Adaptive Spectrum and Signal Alignment Inc |
Claimant |
|
| - and - |
||
| British Telecommunications PLC |
Defendant |
____________________
Roger Wyand QC and Hugo Cuddigan (instructed by Bird & Bird LLP) for the Defendant
Hearing dates: 23rd to 25th October 2013
30th October to 1st November 2013
5th November 2013
____________________
Crown Copyright ©
- In this action the claimant (ASSIA) contends that the defendant (BT) has infringed two of its European patents (UK). The first is EP (UK) 2,259,495 entitled "Adaptive DSL margin and band control using historical operational data" with an earliest claimed priority date of 7th December 2003, a filing date of 2nd December 2004 and a date of grant of 9th November 2011. The second is EP (UK) 1,869,790 entitled "DSL state and line profile control" with an earliest claimed priority date of 3rd March 2005, a filing date of 28th February 2006 and a date of grant of 23rd May 2012. The defendant denies infringement and contends that the patents are invalid.
- The claimant is represented by Iain Purvis QC and Joe Delaney instructed by Wragge & Co and the defendant is represented by Roger Wyand QC and Hugo Cuddigan instructed by Bird & Bird.
- Both patents relate to methods for controlling the way in which a digital subscriber line (DSL) operates. The term DSL is an umbrella term that refers to various ways of sending digital information over telephone lines. A well known DSL technology is Asymmetric Digital Subscriber Line (ADSL) that is used by many communications providers to supply broadband internet access to consumers. The infringement claim relates to the operation of the BT Dynamic Line Management System (the DLM System) which is responsible for controlling the operation of the DSL lines that make up BT's broadband access network. In particular it concerns the operation of a device in the BT DLM system known as the Rate Adaptive Monitoring Box (RAMBo) which monitors and controls various parameters of the DSL lines connected to multiple telephone exchanges.
- BT's DLM system operates on three networks known as 20CN, 21CN and NGA. These refer to BT's "20th Century Network", "21st Century Network", and "Next Generation Access" network. There are separate RAMBos for each of these networks and each operates differently. The underlying type of DSL technology used on each of those networks varies. 20CN is an ADSL1 network, 21CN is an ADSL 2/2+ network and NGA is a VDSL2 network. The expressions ADSL 1, ADSL2, ADSL 2+ and VDSL refer to different types of ADSL system governed by different standards. The relevant standards are promulgated by the International Telecommunication Union (ITU).
- The acts of alleged infringement are the making, keeping and use by BT of the DLM system and in particular the RAMBo elements and the use and offer for use of the method used in the DLM system to control BT's DSL lines.
- The only claims which I need to be concerned with are the ones which ASSIA contends are independently valid and are infringed by BT. These are:
- ASSIA contends that BT's 20CN and 21CN networks infringe claims 1, 8 and 15 of the 495 patent. ASSIA does not contend that BT's 20CN or 21CN networks infringe any of the other independently valid claims of either patent.
- ASSIA contends that BT's NGA system infringes claims 1, 6, 8 and 15 of the 495 patent and claims 1, 10 and 13 of the 790 patent.
- BT denies all these infringement allegations.
- In relation to validity, BT contends that claim 1 of the 495 patent lacks novelty over United States Patent number 6,587,502 (Hendrichs) which was published on 1st July 2003. BT also contends that claims 1, 6, 8 and 15 of the 495 patent are obvious over three references. They are Hendrichs, European Patent Application 1,337,062 A2 (Rahamim) which was published on 20th August 2003, and an article appearing in the IEEE Communications magazine in September 2003 entitled "Advanced DSL Management" by Kenneth Kerpez and others (Kerpez).
- In relation to the 790 patent BT contends that claims 1, 10 and 13 of the 790 patent are obvious over Hendrichs and over US patent 6,266,348 published on 24th July 2001 (Gross). At an earlier stage in these proceedings BT also relied on two further references against the 790 patent. One was International Application PCT/US 2006/001131 published as WO 2006/076518 A2 (Wu) which was relied on for lack of novelty under section 2(3) of the 1977 Act (article 54(3) EPC). The other was US 6,567,464 (Hamdi) which was relied on for obviousness. Both were dropped during trial.
- The relevant claims of the 495 patent are:
- The relevant claims of the 790 patent are:
- ASSIA is a US corporation based in Redwood City, California and was founded in 2003 by Professor John Cioffi. The technical concepts that now form the foundation of the global DSL industry emerged in the late 1970s at least in significant part from work done by Prof. Cioffi at Bell Labs. In 2010 Prof. Cioffi was awarded the IEEE Alexander Graham Bell Medal in recognition of this pioneering work and is apparently credited as being the "father of DSL". Of course, the fact that Prof. Cioffi did this work and was credited in this way in 2003 and the fact that he is named as one of the inventors on both of the patents has no bearing on the issues I have to decide. Some of the things clever people do will still be obvious to a person skilled in the art.
- BT is the well known provider of technology communication services in the United Kingdom. At one point in the argument BT's counsel were anxious to point out that if BT is held to infringe, this may have an impact on all the home broadband connections provided by BT across the whole country. Although this may underline the commercial importance of these proceedings, it has no bearing on the technical issues which I have to decide in this judgment.
- When ADSL is used for a broadband internet connection, the way it works is as follows. There will be an ADSL modem in the customer's home and another one situated centrally, at the exchange. The two modems are connected to each other over the pair of twisted copper wires which form the typical cabling of what is called the Plain Old Telephone Service or POTS. Many twisted pairs, sometimes thousands, are bundled together in a single cable. The twisting of the pairs was invented by Alexander Graham Bell in order to minimise the electromagnetic interference from external sources, allowing many voice circuits to share a single cable. The conventional telephone system is also sometimes referred to as PSTN or Public Switched Telephone Network.
- On a conventional telephone line the voice communication uses the frequency band of 300 Hz to 3.4 kHz. This is referred to as the voice band. Originally, computers were connected to the internet using dial up analogue modems. These modems made use of the voice band frequencies but this meant that the data rate was limited and also meant that a separate line would be required in order to conduct a voice call and a data session simultaneously.
- In fact the copper pairs in the telephone network can carry frequencies well above 3.4 kHz and DSL modems make use of these higher frequencies to carry data at much higher bit rates than are possible using the voice band. In addition this allows telephone lines with a broadband service to carry simultaneous voice calls and data sessions. At the local exchange the voice calls and the data sessions are separated, with the voice calls connected to the PSTN and the data connected to a data network.
- DSL began with basic rate access to the Integrated Services Digital Network (ISDN) which, when it was first introduced, allowed systems to operate at 160 kbit/s to provide two fixed rate 64 kbit/s data channels, one in each direction, for each customer. ISDN was conceived in 1976.
- In 1999 the ITU recommendation for ADSL was published. In ADSL the data rate towards the customer (downstream) is much greater than the data rate of the network upstream which is why it is called "Asymmetric" DSL.
- The relevant forms of ADSL commonly use a system called DMT (Discrete Multi-Tone) to send signals in the frequency bands above the voice band. In DMT the data is sent by means of a stream of "symbols", whereby each symbol is comprised of multiple modulated tones. Each tone is modulated by a number of bits of data. The symbols are sent at a rate called the signalling rate. The rate of data transfer, i.e. the bit rate, depends on the signalling rate, the number of bits in each tone and the number of tones per symbol. The medium through which a signal is transmitted is known as a channel. The DMT symbol structure allows the transmitted signal to be tailored to the characteristics of the channel.
- In order for a received symbol to be demodulated correctly, it must be sent with sufficient power relative to any noise present in the channel. Signals also attenuate during their passage through a medium. Thus for a given medium, a higher power will be required to successfully transmit a signal to a receiver further away than a receiver that is closer. Higher power can be achieved by an increase in gain (amplification) in the transmitter. Attenuation can be a function not only of distance but also frequency. In DMT gain can be applied to individual tones.
- The signal to noise ratio (SNR) is a commonly used measure of the power of a signal. If the SNR is insufficient, the noise will corrupt more of the data. Since noise can increase and decrease randomly, one cannot be sure to eliminate the effects of noise but if the SNR is large enough, few errors will be caused. Even then there can be short bursts of noise called impulse noise which are of such high energy that the SNR cannot protect against them.
- The idea of an SNR Margin is used to specify a margin which the signal power must have above a minimum signal power for a specified error rate. BER stands for Bit Error Rate and is the number of uncorrected errored bits divided by the total number of bits transferred in a specified time. BER cannot be measured directly. SNR, SNR margin, bit rate, signal power and noise power are interrelated. So for example if signal power is lowered then, in order to maintain SNR, the bit rate must be reduced. For a given SNR, high power and low noise allow for higher bit rates.
- Power spectral density (PSD) refers to the transmission power distribution across the frequency band used in the system. This concept is used in DMT where transmission occurs simultaneously on multiple tones and the transmission power of each of the tones may vary.
- There are ways of detecting and correcting errors in digital communications. A particular relevant technique is Forward Error Correction (FEC). This works by adding redundancy to the transmitted information which in turn allows the receiver to detect and correct a certain number of errored bits without the need for retransmission. In DSL FEC is implemented using Reed-Solomon coding.
- Interleaving is a technique used to enhance error correction when impulse noise is a cause of errors. It works by spreading out the bytes of several code words so that the code words are interleaved. The number of code words across which the data is spread is called the interleaving depth. This means that if there is an impulse noise event which results in errors at one point in the stream of data, when the stream is de-interleaved at the receiver the errors end up in different places and each one can be corrected by FEC. Interleaving does not itself correct errors but enhances the burst error correcting capacity of FEC. The disadvantage of interleaving is increased latency. Interleaving causes a delay between the data transmission and the time the data has been decoded and made available.
- The incorporation of twisted pair copper lines from many homes into a single cable or binder can give rise to an important source of noise known as crosstalk. Crosstalk involves the signals on one twisted pair interfering with the signals on another twisted pair. Owing to the low frequency of the voice band, crosstalk between lines is not a serious problem with conventional telephone communications but at the much higher frequencies used in ADSL, crosstalk between lines can occur and can be a problem. The crosstalk caused by a given line is related to the power at which the signals are being transmitted and so crosstalk caused by a line can be reduced if the power of the signals on that line is reduced. Equally if a line is experiencing crosstalk, its impact may be mitigated if the line increases its own power.
- The basic elements of the typical ADSL system are shown below:
- This is figure 1 of one of the patents (both figures 1 are essentially the same). A very similar diagram appears as figure 5-1/G.997.1 "system reference model" in the May 2003 edition of ITU standard G.997.1 entitled "Physical Layer Management for Digital Subscriber Line (DSL) Transceivers".
- In the figure, the equipment on the left hand side is in the customers' premises. On the right hand side is the equipment at the operator's network (called the central office or CO). In the diagram interfaces are shown as dotted lines.
- Starting from the left, the boxes marked TE refer to terminal equipment. That is, for example, a home computer. The home computer is then connected to a home network shown as item 112. The home network is connected to a box marked NT (Network Termination). In a typical example in a home this box is the thing referred to as a home broadband router. In fact inside the typical home broadband router is a router which allows the home network to operate and a modem. The physical connection to the home network is marked "PHY" in the NT box while the ADSL modem is marked ATU-R, which stands for ADSL Transceiver Unit - Remote. The word "remote" refers to the home. The ATU-R is the modem (modulator demodulator) which connects the home network to the twisted pair telephone line marked as "loop" in the diagram. The modem modulates the outgoing signals so that they are in a suitable format to be sent across the channel formed by the telephone line and demodulates the received signals to extract the data encoded in them.
- Inside the NT box is an oval labelled "Management Entity". This does not represent a physically distinct thing but represents the control functions on the computer inside the NT box which manages the behaviour of that piece of equipment. The connection between the home and the central office is via the telephone wire marked as "loop".
- At the Central Office there is an Access Node (AN) which corresponds to the NT at the home. There is a corresponding modem called ATU-C which stands for ADSL Transceiver Unit - Central office. This connects to the telephone wire and communicates with the ATU-R. The connection between the ATU-C and the wider broadband network is shown on the right hand side. As with the NT box in the home, the AN has a Management Entity which is not necessarily physically separate but manages the AN.
- An important interface is the interface between the Management Entity in the Central Office Access Node and the Network Management System or NMS. This is the Q interface. The Network Management System controls the operation of a number DSL lines and interacts with the individual modems over the Q interface. The Q interface is defined in ITU Standard G.997.1. It is sometimes called the G.ploam. The term ploam being an acronym for Physical Layer Operations Administration and Maintenance. This standard is also sometimes called the DSL Management Standard.
- Another important interface is the U interface (U-R and U-C) which determines how the two modems talk to each other. The different DSL standards differ in the specifications of their U interfaces.
- The diagram shows a double arrowed dashed line (132) between the Management Entities in the home and the Central Office. This is not a physical connection between the entities but represents a logical connection in the sense that the entities can communicate with each other. Physically those communications occur over the loop along with all the other data which passes across the loop.
- The home telephone handset is not shown on the diagram above. In a system in which the user's telephone shares the line with the ADSL system, a splitter or filter would be inserted between the handset and the line carrying the ADSL signals. This provides some electronic separation between the electrical effects of the telephone and the ADSL signals. This is shown in figure 3 of Dr Adams' report:
- In this diagram the ADSL modem at the central office is shown in a DSLAM or DSL Access Multiplexer. This is a unit in which a number of ATU-C modems, one for each customer line, are put together and controlled together.
- An ITU standard, G.992.2, specifies the U interface for a kind of ADSL known as "ADSL Lite". This is a kind of DSL technology that does not require the use of such splitters. This will be discussed further below.
- In order for the two modems to successfully communicate, they must have a shared communications protocol. To achieve this they must start by negotiating the appropriate values for a number of standardised parameters. This initialisation occurs during a training phase of operation which is specified in the standard. In this action the term "transmission parameters" has been used to describe the set of successfully negotiated/ mutually shared parameters that are needed to ensure successful communication between the modems. "Transmission parameters" is not a term of art but both experts found it a useful one to describe these sorts of parameters. Following the training phase the modems start working normally to transmit data upstream and downstream. This normal operation phase is called SHOWTIME. For various reasons the modems may undergo a "full retrain" in which they completely renegotiate the transmission parameters.
- The transmission parameters will include individual settings for each of the DMT tones being employed. These settings specify the number of bits to be encoded in a given tone and the gain (power) used to transmit that tone. So for example if the channel has a high attenuation in a certain narrow frequency range, the tones transmitted in that frequency range may have a higher gain and/or lower number of bits.
- During normal operation (i.e. during SHOWTIME) the modems can adjust the transmission parameters somewhat. The bits and gains for individual tones can be adjusted by the modems while maintaining the same overall bit rate. This is called bit swapping.
- Another expression used in the evidence is "configuration parameters". These are the values given to the modems by the operator of the line, ahead of initialisation. These types of parameters are not used directly by the modems during SHOWTIME but they set the boundaries or requirements that the operator wishes the modem pair to comply with in operation. "Configuration parameters" is not a term of art but both experts found it a useful one to describe these sorts of parameters. An example of a configuration parameter specified by the DSL Management Standard is MAXNOMPSD. This specifies the maximum nominal transmit power spectrum density in a given direction (upstream or downstream). Its units are dBm/Hz. The standard specifies the range of possible values and requires that they take any value in the range in 0.1dB steps.
- At the priority date (and for this purpose there is no relevant difference between the position at the priority date of either patent) any change to configuration parameters used in a particular DSL line was typically done through a process of manual human evaluation and intervention at the central network operations office.
- ASSIA relied on the expert evidence of Professor Stephen McLaughlin. Professor McLaughlin is a Professor of Signal Processing and Head of School in the School of Engineering and Physical Sciences at Heriott Watt University, Edinburgh. With a BSc in Electronics and Electrical Engineering from the University of Glasgow and a PhD from the University of Edinburgh he was awarded a chair in the school of engineering and electronics at the University of Edinburgh in October 2002 and moved to the Chair of Signal Processing at Heriott Watt University in October 2011. After his undergraduate degree, from 1981 until he was appointed a lecturer at Edinburgh University in 1988, Professor McLaughlin worked in industry first for Barr & Stroud in Glasgow and then for MEL Ltd a subsidiary within the Philips Group in Dunfermline. Part of that work involved the design and development of data communication equipment operating over the high frequency band and signal processing techniques for the military. His PhD was entitled "Adaptive Estimation and Equalisation of HF (High Frequency) Communications Channels". His general research interest is in the study of signals and their interaction with systems with a particular focus on signal processing.
- Professor McLaughlin gave his evidence fairly and clearly sought to assist the Court in understanding the technical issues in this case. The defendant submitted he was out of his depth and referred to a number of errors he had made, some of which were acknowledged before trial and some of which emerged in cross-examination. It is clear that Prof McLaughlin had indeed made some errors in his assessment of the details of the BT system. However the fact that these errors were made does not undermine the Professor's evidence in general. I do not think the errors indicate any lack of care or lack of expertise on the Professor's part. They are rather a symptom of the sometimes byzantine complexity and abstraction of some of the arguments on infringement.
- The defendant relied on the evidence of Dr Peter Adams. Dr Adams is an electronic engineer and currently works as a senior standards consultant. When he left school in 1966 he became a student apprentice at BT (which was then the Post Office) before going to university. At university he remained a student apprentice at BT and after graduating took a job with BT carrying out research and development work relating to digital signal processing. The particular focus of his work in the 1970s was on digital filters used in BT's digital exchanges. He continued to work for BT working in the speech band modem group and the local group transmission group and the copper access section. Finally between 1992 and 2002 he was the manager of the advanced operational solutions unit at BT. This involved working with access broadband and internet protocol and international networks and at that time he also chaired the operations working group at the DSL Forum which is now called the Broadband Forum. In 2002 he retired from BT and worked as a consultant. Between 2003 and 2009 he consulted part-time for BT relating to broadband access architecture. While he was a student he commenced a part-time PhD at the University of London which was awarded in 2011 entitled "Fault Demarcation and Electromagnetic Interference Identification for Broadband DSL Access". He had lectured on broadband access and carries out research on broadband diagnostics at Queen Mary, University of London.
- The claimant submitted that although Dr Adams was now nominally independent of BT he was not in a position to give objective unbiased evidence in a case in which BT was the defendant and had made no effort to do so. Dr Adams was criticised for a change he had made to his first report about configuration and transmission parameters, for spontaneous evidence he gave in the morning on day 6 of the trial and for being unwilling to accept what the claimant contended were self evident propositions. I reject the submission that Dr Adams was biased or made no effort to give objective evidence. It is true that at times Dr Adams had a tendency to argue, but in my judgment this arose from his enthusiasm for the issues rather than from any bias. The change to his first report and the spontaneous evidence on day 6 were the actions of an expert who was conscious of his duty to the court and was seeking to fulfil that duty.
- On the technical issues about which this case relates, I thought Dr Adams was seeking to assist the Court just as was Professor McLaughlin. I am grateful to both gentlemen for their evidence.
- By closing, there was no dispute that the relevant person skilled in the art would be a team of electronic engineers with knowledge and experience of DSL transmission. Dr Adams' view, which I accept, is that by that time DSL was well established as a distinct field.
- By closing, there was nothing between the parties as to the content of the common general knowledge of the skilled person. It would include all the matters set out above in the background section from the heading ADSL to the end. As I have said, the expressions "transmission parameters" and "configuration parameters" were not terms of art but were employed for these proceedings. The skilled person understood the distinction between parameters set by the network operator which provide boundaries or requirements that the modems must comply with (configuration parameters) and the parameters negotiated by the modems during training and actually applied during SHOWTIME (transmission parameters).
- A specific question was whether the content of the Kerpez article represented common general knowledge. Dr Adams's view was that a member of the skilled team would have read the Kerpez article itself when it was circulated. Whether or not that is so is not something I have to decide because I understood Prof McLaughlin's evidence to be that the content of the article reflected what those working in the DSL field itself knew about and had been discussing for some years at the time. It seems to me that this supports the defendant's case that the content of the article represented common general knowledge and I find that that is the case. Whether the existence of the article itself was common general knowledge does not matter if its contents represent what the skilled person already knew.
- Kerpez discusses management of ADSL lines including the concept of Dynamic Spectrum Management (DSM). Prof McLaughlin explained that DSM was an umbrella term referring to ways of automatically enhancing DSL by compensating for impairments like crosstalk. The basic idea is that by centrally and simultaneously controlling a number of DSL operational parameters the adverse effects of crosstalk can be mitigated and data rates improved. A number of different categories of DSM were recognised in the art. They were known as DSM level 0, 1, 2 or 3. Kerpez sets out these four levels of DSM. Although in other places a different characterisation of the four levels has been specified, nothing turns on the different ways of characterising the levels. The various DSM levels can be summarised as follows:
- The defendant also put a textbook from 2003 called DSL Advances by Starr, Sorbara, Cioffi and Silverman to Prof McLaughlin. This was a text Prof McLaughlin had described as thorough and accurate in his first report. Chapter 11 of that textbook relates to DSM. It is a detailed chapter and I do not have to decide whether the whole of it represents common general knowledge. What is clear on the evidence in my judgment is that it is further proof that DSM was common general knowledge.
- However it is important not to misunderstand what was common general knowledge about DSM. DSM was something which had been talked about for years. The idea of automatically controlling the parameters applicable to a particular ADSL line was something the skilled person was familiar with as a proposal, but at the relevant date it was just that, a proposal. The Kerpez article bears this out. It treats the automatic management of ADSL lines as something which will happen in the future. That is an accurate reflection of the common general knowledge.
- Finally I should mention the standards. They were common general knowledge. This includes the standards specifying the U interface for the various forms of ADSL such as G.992.1, G.992.2, G.992.3 etc. and also the DSL Management Standard G.997.1.
- The 495 Patent relates to methods, systems and apparatus for managing digital communication systems and in particular for providing adaptive control of the parameters used in DSL communication systems. Paragraph [0001] refers to the control of "transmission parameters" but it is clear that this term is not being used there in the same sense as that term has been used in the evidence in this case.
- The patent explains the purpose of the invention in the section from paragraph [0002]. Paragraph [0003] explains how DSL lines can suffer from problems with accuracy of the data transmitted as a result of impulse noise and other sources of error. Paragraphs [0004] to [0006] explain how these problems are addressed by selecting a transmission power of a modem which ensures an SNR above a given margin. This is typically done by an individual pair of modems in the initialisation phase. Once the normal operation has started, i.e. during SHOWTIME, the modem pairs are limited in the adjustments they can make. The modems can adjust the bit-loading and gains of their individual sub-channels.
- The patent then points out that the restrictions placed on operational limits by the various standards are commonly disregarded, so that modems can use excessive power. Excessive power in one line can cause serious problems of crosstalk in neighbouring lines. Between paragraphs [0009] – [0014] the various adjustments which are possible within the modem pairs under the different DSL standards are discussed and in paragraphs [0015] to [0017] the patent explains that in many DSL systems modem operational characteristics have been set in a static mode to accommodate worst-case scenarios in the system. Excessive power can create particular crosstalk problems. Impulse noise is also mentioned as a significant problem. Paragraph [0017] concludes with the statement that the lack of compliance with the standards (or lack of compliance by neighbouring modems) causes many users to suffer a poor or less than optimal service.
- There is then a short section reviewing certain prior art citations and at paragraph [0024] the patent states as follows:
- The next section, entitled "Brief Summary of the Invention" contains a consistory clause in terms of claim 1. Some further aspects and embodiments are set out in this section up to paragraph [0031] and the specification then continues with a description of the figures and a detailed description of embodiments of the invention (paragraphs [0033] to [0156]).
- In summary, the essential idea of the invention is to monitor the line, analyse the data, generate new parameters which ought to produce an improvement and then tell the system to operate using these new parameters. The method allows the operator to dynamically control those parameters of the operation of a modem pair which affect the required SNR margin (margin-related parameters). This is a different approach from the known approaches which typically established a static set of parameters for the modem pair and relied on the modems' dynamic adjustment of bit-loading and gains.
- The issues to be addressed in relation to the construction of claim 1 are:
- I will deal with them in turn.
- A major dispute between the parties is about the meaning of this word in claim 1. The claim calls for "a method in a controller communicably attached with a Digital Subscriber Line modem pair". The claimant contends that the skilled reader would understand that although the controller required by the claim need not be physically distinct from the ADSL modem, it must be logically distinct.
- Essentially the claimant's case is that the skilled reader would see on reading the specification that the function of the controller is to set configuration parameters (such as MAXNOMPSD) rather than transmission parameters (such as the bits and gains for individual DMT tones). All the parameters set by the controller described in the specification are configuration parameters and that reflects the true nature of the controller required. It is not operating at the level of setting the detailed transmission parameters negotiated between the known ADSL modems. The claimant's case involved a close analysis of the specification to support its argument.
- The defendant does not agree. Its case is that the distinction sought to be drawn by the claimant is not to be found in the 495 patent. In particular it points to the express disclosure of a "smart modem" as a possible embodiment in paragraph [0028], paragraph [0052], claim 13 and figure 12. The figure shows a unit having a controller comprising a processor and memory integrated with a DSL modem.
- The word controller is an ordinary English word. It has no special meaning in this art. The skilled reader would not expect such an expression to be used in a particularly limited sense so as to exclude some things which exercise control without a clear indication that this is so. If it were not for the smart modem, then the words "communicably attached with a Digital Subscriber Line modem pair" might be taken to indicate some sort of physical distinction but the smart modem embodiment rules that out. A controller is still a controller even if it is integrated into the processor in an ADSL modem.
- It is true that in the detailed description of embodiments in the specification the parameters produced by the controller to be used by the modems are configuration parameters and not transmission parameters. However focussing on what sort of parameters the controller provides as an output takes one to a different part of the claim. The claim defines what the controller produces in order to instruct the modem how to operate as simply a "margin-related parameter set", without distinguishing between configuration parameters and transmission parameters.
- At one stage in this action the claimant appeared to be arguing that "margin-related parameter set" had to be interpreted as limited to a set of configuration parameters but that is plainly not correct. In closing the claimant disavowed any such argument. The specification makes it clear that margin-related parameters includes some parameters which are indeed configuration parameters such as MAXNOMPSD but also others which are transmission parameters such as bit loading. A margin-related parameter set is simply a set of margin-related parameters. Thus there is nothing in claim 1 to indicate to the skilled reader that the draughtsman has sought to place limits on the kind of parameter which might form part of the margin-related parameter set which is the output from the controller. Indeed there is no good reason why the reader should think they had.
- The skilled reader would not think the patentee, by drafting its claim broadly so as to cover a set of any margin-related parameters, intended to exclude a method which used a controller (of whatever kind) which actually set transmission parameters. The fact that a physically separate controller built according to the current standards and communicating across the Q interface could not do that is not determinative since the invention is not limited to systems compliant with any particular standard. Pointing out that the controller in the specific embodiments only sets configuration parameters does not help given the wide meaning of "margin-related parameter set". The specific embodiments do not prove that the controller of claim 1 has to be something distinct from the modem.
- The claimant also referred to the words of claim 1 which are said to favour the claimant's construction. In the claim the controller must be "communicably attached with a DSL modem pair", and must collect data "from the DSL modem pair" and must be capable of "instructing the DSL modem pair" to operate in a certain way. This is a better point than the argument based on the specific embodiments but in the end I do not think it is sufficient to justify reading words into the claim which are simply absent.
- I cannot reconcile the specification of the 495 Patent as a whole with the claimant's construction of "controller" and I reject it. To interpret the word as the claimant contends is an exercise in putting a limitation into an ordinary general word which is simply not present. A controller is something (hardware or software) which exercises control. It must be communicably attached with a Digital Subscriber Line modem pair but as long as it exercises the required control, it will be a "controller". It does not have to be distinct from the modem. Any narrower interpretation is ruled out by the smart modem and the broad meaning of margin-related parameter set.
- At one stage there was an argument that data about channel attenuation was not operational data but this was not pursued by the claimant and nothing turns on the ambit of the term. It will include data concerning the characteristics of the relevant line during operation.
- There was no real dispute about what the terms current and historical referred to. Their meaning is illustrated by the DSM Management Standard. This provides that ATU-Rs and ATU-Cs will each have an internal database called a MIB (Management Information Base). The MIB is a repository of the configuration parameters, data from measurements made by the modem and other information. It is a store of operational data. The standard specifies that certain data are collected and stored for 15 minute evaluation periods and kept in stacks. When data for a new period is put on the stack, data for the period at the bottom of the stack is discarded. Some data for a 24 hour period is also kept in a similar way. Thus the standard includes a diagram (fig 7-1/G.997.1) which represents the way in which successive blocks of data for successive periods are stored:
- The skilled reader would understand the reference to current and historical data in the 495 patent in the same sense as the terms current and historic data are used in this figure. The current data is the data block in the store at the top. The historic data is data which relates to earlier evaluation periods.
- Given the temporal nature of the concepts, it is necessary to ask at what moment during the performance of the method steps is the data to be considered current or historic. The claimant contended the relevant point was when the data was collected. After all, the claim refers to current and historical data in the clause reading "collecting operational data … wherein the operational data comprises current operational data and historical operational data".
- The only alternative would be to judge the current or historical nature of the data at the moment the data is analysed. However that is not what the claim says. Moreover it would mean that a system collecting 15 minute blocks of current and historical data would only infringe if it completed the analysis step in 15 minutes. If the system waited any longer there would be another block of current data in the MIB and, on the defendant's case, the data which was current when it was collected would then have to be regarded as historic. This does not make sense. In my judgment the question of whether data is current or historic relates to the moment of collection.
- The step of "collecting operational data from the DSL modem pair" refers to a process in which the data is gathered up from somewhere in order to be used by the controller. It would include the act of a program running in a processor in the DSL modem retrieving operational data from the MIB but it was common ground that it does not cover the conventional recording in the MIB of operational data by the modem.
- The claim includes the step of "analyzing at least a portion of operational data". It was common ground but is worth noting that the claim does not require that both current and historical data are analyzed. Both must be collected but the claim would cover a method in which only (say) historical data was analyzed.
- The next step in the claimed method is "generating a margin-related parameter set based on the operational data analyzed". The parties did not agree what "generating" encompassed. The defendant argued that the skilled reader would understand it to mean that the set of margin related parameters had to be created anew and so, for example, a system which selected an existing set of parameters from a group of pre-stored parameter sets would not infringe. The defendant pointed out that in the parallel opposition proceedings in the European Patent Office brought by the defendant, the claimant has argued exactly that in seeking to distinguish the claim from the Gross prior art. There the claimant draws the contrast between generating a margin related parameter set and merely using a look-up table to reference pre-stored parameter sets in the prior art.
- Before me the claimant contended that the claim covered both creating a parameter set and using a table of pre-stored parameter sets so that, for example, the best match profile from a stored collection could be used. In effect the claimant's construction requires the controller to generate a set of parameters in the sense of producing them to be used, regardless of where they came from.
- As regards the EPO, Mr Purvis submitted that his client was not bound by the interpretation of claim 1 it had advanced in the opposition proceedings and since Mr Wyand did not contend to the contrary, I do not have to decide on file wrapper estoppel. It seems to me that the reference to what was said in the EPO is relevant only to this extent. It shows that the claimant cannot credibly argue that the construction advanced by the defendant on this point is untenable.
- The defendant contended that the specification was consistent with its construction of "generating" in that it discloses the idea of calculating a new value for the relevant margin-related parameter. The relevant parts are paragraph [0055] (in which the word calculated appears) and paragraph [0063] referring to Fig 7 which talks abut modifying a parameter to generate a modified parameter value (box 740). These passages are indeed consistent with the defendant's construction but the defendant's argument there is similar to the claimant's argument on the word "controller" in that it is seeking to rely on a narrow disclosure in the embodiments section of the patent to try to narrow down the claim.
- The defendant also contended (a) that the draughtsman clearly had in mind the difference between selecting and generating since the word selecting is also used in the patent elsewhere and (b) that the term "generating" was used elsewhere in the patent with the defendant's proposed meaning such as in the context of a signal generation module (paragraph [0027]) and generating an alarm (paragraph [0023]). I think neither is a strong point but the latter point is rather better than the former.
- The claimant referred to paragraph [0057], which mentions the possibility of using databases when large numbers of lines are being managed in order to manage the volume of data generated by the lines and the controller.
- An issue debated in the evidence was whether there were technical reasons why the claim could or should be read one way of the other. As the claimant established with Dr Adams, (i) the difference between the two options is simply one of programming, and (ii) as far as the modem is concerned, it makes no difference whether the margin-related parameter set it is provided with is one which was created afresh or selected from a pre-stored group. The claimant also pointed out that under the DSL Management Standard the values which configuration parameters might take are fixed and are quantised (in other words they must take certain given values). Thus at least in theory the set of all possible values of all possible parameters is finite and could be stored. Although the number of permutations could be very high, the claimant pointed out that the claim covered the case in which a single parameter was set and in such a simple case there was no difference between selecting and creating a value.
- A distinct point arose from the common general knowledge. The common general knowledge included the idea that at the priority date DSLAMs stored a number of profiles (i.e. sets of configuration parameters) and network operators manually decided which of those profiles were to be used on a particular line. The claimant argued that the skilled reader would understand that the benefits of the patented method extended to making a decision to optimise the line in a similar way in place of a human operator manually reacting to a fault.
- There is no conclusive argument for either party's construction of this part of the claim. The word is not one with a special technical meaning. I think its natural meaning refers to the idea of creating a bespoke set of parameters, tailored to the particular line conditions, rather than simply selecting from a pre-existing "off the peg" collection. That is how I read the words.
- Although there are reasons why one might think the patentee could have intended the claim to be read more broadly, there are also reasons supporting what I regard as its natural meaning. There are also limits to how far one can go in imagining how a skilled reader would speculate about the patentee's intention in this respect. Perhaps the word was used to distinguish from the process of selection of known profiles stored in the DSLAM which was part of the common general knowledge. The skilled reader cannot know.
- While I can see that it makes no difference to the modem whether it is told to use a bespoke set of parameters or one selected from an off the peg collection, it will make a difference to the controller. The controller has to conduct a different kind of process to reach the end result in each case albeit that in some circumstances (a single parameter) the difference between the two kinds of process will be effectively non-existent. It will also make a difference to the operator since a bespoke set of parameters may be able to be configured more closely to the actual line conditions. That is consistent with the objectives of the invention to improve the management of DSL lines by adjusting and adapting the relevant parameters dynamically in response to changing conditions.
- I reject the claimant's wide construction of generating. It has the narrower meaning proposed by the defendant.
- A margin-related parameter set is simply a set comprised of margin-related parameters. Margin-related parameters clearly have to relate to the margin (i.e. SNR margin) but they may be transmission parameters or configuration parameters. It is common ground that a "set" in this context may consist of a single parameter.
- Having dealt with the detail on claim 1, it is worth standing back and reviewing the overall result. Properly construed claim 1 is a relatively broad claim to a method of adaptively managing any margin-related parameters which relate to a DSL line. Its key limitations are that the sets of parameters must be generated (i.e. created) rather than selected and that both historic and current operational data must be collected by the controller.
- At some points in the argument the claimant submitted that if the skilled person thought a particular construction of the claim would mean it covered conventional DSL modems, then that construction would be rejected since the invention is obviously supposed to be something more than a conventional DSL modem. Although this kind of argument cannot be taken too far it is worth considering in this case at least as a cross check, since conventional modems do adaptively manage margin-related parameters during normal operation by bit swapping, as the patent acknowledges in paragraph [0002].
- The point was advanced to support the claimant's construction of "controller". However despite the broad nature of claim 1 as I have construed it, the skilled reader would see that it does not cover a conventional DSL modem. A conventional modem operating in accordance with the DSL Management standard will store operational data in the MIB. It was common ground that this conventional process did not comprise collecting current or historical operational data. Since conventional modems did not collect such data from MIBs, they cannot fall within the claim. Thus there is no reason why the argument about coverage of a conventional modem requires a narrower interpretation of "controller".
- Claim 6 adds to claim 1 by requiring the step of analysing the operational data to include determining what margin-related parameter value will cause the DSL modem pair to meet a performance target or a target threshold. The reference to "target threshold" can be read as "performance threshold". These are criteria against which the modem pair's performance can be judged. The words target and threshold bear their ordinary meanings.
- The claimant submitted that claim 6 requires the actual value of a parameter to be determined which will meet the relevant criterion and therefore excludes a method in which the control system simply considers whether the modems' profile should go up or down one level. I accept this submission.
- Claim 8 is another claim dependent on claim 1 which calls for the analysing step to work in a particular way. The claim is drafted to require that the analysing step includes one or both of two possible steps defined in the claim. Both steps require a comparison to be undertaken between a collected margin-related parameter value and a corresponding threshold. The objective is to see if the threshold is satisfied. The difference between the two possible steps is that one relates to doing this with current operational data and the other with historical operational data.
- Claim 15 is a product claim. Broadly it claims a controller which is defined so as to have modules which are configured to perform the method of claim 1: collecting operational data, analysing a portion of the collected data, generating a margin-related parameter set based on the analysis and instructing the modem pairs to operate in accordance with that set. To that extent the language in claim 15 would clearly be understood in the same way as the corresponding language in claim 1, albeit bearing in mind it is a claim to a product with components defined in functional terms to carry out the tasks.
- A key difference between claim 15 and claim 1 is that the controller in claim 15 is one to monitor a plurality of DSL subscriber line modem pairs whereas the controller of claim 1 collects data from a single modem pair. Both claims cover a controller which collects data from and instructs a single DSL modem pair. The difference is that claim 15 also requires the controller to monitor more than one DSL modem pair.
- The DLM systems in the three BT networks which are alleged to infringe (20CN, 21CN and NGA) are different from each other. Each network is controlled by a RAMBo, that is a network management system. Each network offers several "products" to service providers which in turn offer broadband services to customers. The "products" have pre-determined sets of profiles. A profile is a set of configuration parameters for a DSL modem pair. The RAMBo instructs the modems on a given line to use a particular profile selected from the relevant set of available profiles.
- Each profile is identified by a unique name or ID number and contains differing values for things like data rates, margins and interleaving parameters. The 20CN, 21CN and NGA profiles contain values for levels of interleaving, cap and/or margin (depending on the network). The role of the RAMBo is to determine from analysis of data from the modems whether it is necessary to change the profile on a line and if so, to select a new profile. To do this the RAMBo applies some logic, which is different between the three networks, to decide what to do. Once per day the RAMBo receives a data file containing operational data about the operation of the line in the last 24 hours. From this the RAMBo assesses the line's performance and decides whether it is such as to require a change of profile. In the 20CN and 21CN networks a DSLAM at any time will alert the RAMBo if the modem pair re-initialises. If this occurs too often the RAMBo will run a process to determine whether to change the profile. This alert in the 20CN and 21CN system is called a re-train alarm.
- In general the logic used by the RAMBo is in three steps. First there is an assessment of the 24 hour line data to determine if the profile should be changed. This leads to the allocation of Indicated Line Quality (ILQ). The ILQs are characterised by colour. Next, if the ILQ indicates a change is required, a further assessment takes place to determine what it should be. Finally, once a new profile is selected a signal is sent to the DSLAM modem to cause the modem to re-initialise using the new profile.
- Taking the features of claim 1 in turn:
- The defendant admits that the RAMBos of all three DLM systems are controllers irrespective of the argument on construction. Each RAMBo is physically distinct from the modems.
- The defendant admits that the 20CN and 21CN systems satisfy this feature on any view of construction since they have retrain alarms.
- The defendant does not admit that the NGA system satisfies this feature because of the way the data is collected. Part of this dispute turned on the issue of construction about what "current" and "historical" data means and when that it is to be assessed. I have rejected the defendant's construction. The question remains whether the NGA system infringes on the construction I have accepted.
- The first point is as follows. In the NGA system the operational data is collected every 24 hours. This will include data from the various 15 minute bins. It includes data in what was the current 15 minute bin at the moment the data is collected. However the RAMBo only analyses the data, and decides whether a new profile should be selected, a number of hours after the data is collected. Thus at the moment the RAMBo performs the analysis, the data in what was the current 15 minute bin is no longer current. However that is irrelevant as I have construed the claim. The claimant's case is that at the time the data is collected by the controller, the data comprises current and historical data. That would be enough to satisfy the claim.
- During trial a pleading point arose about whether the claimant had limited its case to exclude an element of the defendant's NGA system called Yukon. The point was that on one view it was the Yukon not the RAMBo which collected the data and so, by the time the RAMBo itself collected the data, it was indeed not current. However that point fell away when the claimant amended its pleading to include the Yukon as well as the RAMBo as the controller. The defendant did not in the end object to this amendment since the claimant could legitimately point to ambiguities in the defendant's Product and Process Description in this respect (which I am sure were not intentional).
- Finally the defendant submitted that the Yukon did not collect current data either because although when the modem offered it to the Yukon the data was current, it still took one or two hours before the data was stored by the Yukon. However it is not clear what the defendant means by saying that the modem is "offering" data to the Yukon. I did not have my attention drawn to any expert evidence on the point and this way of characterising how the NGA system works is not in the same terms as the defendant's description in its PPD paragraph 20.2. The words actually used in the PPD are "The [data] are sent hourly to the Yukon […]". Similarly the Amended Response to a Request for Further Information (paragraph 21 of the Amended Response dated 9th August 2013) states "The element managers sends the [24 Hour line data] hourly to Yukon which then takes 1-2 hours to load into the database." It seems to me that the description in the PPD and in the Amended Response means that the Yukon is being sent current data. When the data is sent to the Yukon, it is current. The descriptions do not explain in any more detail. Perhaps the data is in a buffer at the Yukon or somewhere else in memory but the description does not say so. The fact the Yukon takes 1-2 hours to put the data in a database makes no difference. If the defendant had wanted to make this point good, it could and should have provided a clearer description.
- Accordingly I find that the NGA system satisfies this aspect of the claim.
- No issue arises in relation to this feature for any of the three systems. It is satisfied.
- Although not all the parameters in the relevant profiles for all three systems are margin-related, for example interleaving parameters are not margin-related, there is no dispute that every profile contains some margin-related parameters. Thus this aspect of these parts of the claim is satisfied.
- The claimant's infringement case in relation to these features depended on its case on construction of the word "generating" because in all three of the defendant's systems, once the RAMBo has identified which profile it is going to send to the modem, the set of margin-related parameters making up that profile are selected by looking up the levels in its database. As I have construed the claim, this will not fall within the claim. On this basis neither the 20CN nor 21CN systems infringe.
- There was a further point on the NGA system as follows. In certain circumstances, the logic employed in the NGA system in order to decide which profile it wants to select performs this task by calculating a rate called the safe rate. The safe rate is (0.7 x actual rate + 0.3 x maximum attainable rate). So if the actual rate on the line in question had been 50Mbps and the maximum attainable rate in the system was 80Mbps, the safe rate would be 59 Mbps (=35+24). Thus the safe rate is a rate somewhat higher than the actual rate but never more than the maximum attainable rate. The safe rate is then used to identify the correct profile to be selected. It may allow the line to change to a profile with a somewhat higher maximum rate. Each profile has a different "cap level". The cap level corresponds to the maximum data rate for that profile. Lower cap level profiles have lower maximum data rates.
- The profile chosen by the logic in this circumstance will be the one with a maximum data rate just higher than the safe rate. In other words, ranking the profiles in order of rate, if for a given profile (say cap level 20) it has a maximum data rate above the safe rate and if the next lower profile (cap level 19) has a maximum data rate below the safe rate, cap level 20 is selected. The safe rate is in between the maximum data rates of these two profiles.
- The point is that this therefore involves the NGA system in "generating" a parameter and using it to decide which profile to use. The defendant objected to this argument as not having been pleaded and it is true that the safe rate is not identified in the claimant's Statement of Case on Infringement as a relevant margin related parameter. The margin related parameter set relied on for NGA comprises four values, none of which are the safe rate. Mr Purvis submitted that the cross-reference in the relevant part of the Statement of Case on Infringement to paragraphs 122 to 149 of the PPD was sufficient but I do not accept that since the relevant sentence referring to those paragraphs characterises them as referring to "that set" of parameters, i.e. refers to the already defined set of parameters. However I will not decide this on a pleading point. In my judgment although it is true that the NGA system does generate the safe rate as a value in its logic and does use that to determine which profile, i.e. which set of margin-related parameters, it will choose, the NGA system is still only selecting the parameter set from a pre-existing table. The profile actually sent to the modem comes from the pre-existing table of profiles. The fact that a safe rate was generated as part of the decision logic does not turn the selection of that pre-existing profile into the creation of a new profile. Although I accept that in some cases the safe rate will happen to be the same as the maximum rate in the profile selected, in other cases the actual value of the safe rate will be different from the maximum rate in the profile selected. This does not mean the system infringes sometimes and not others, rather it shows that the system is not using the safe rate itself in the profile. Accordingly even if the case had been pleaded properly by the claimant, I would not accept it.
- I find that claim 1 is not infringed by any of the BT DLM systems.
- Since claims 6 and 8 are dependent on claim 1, they cannot be infringed if claim 1 is not infringed. Nevertheless I will deal with them in case the matter goes further.
- The point of claim 6 is to require the system to actually determine what value will meet a given threshold rather than simply come to a conclusion that a parameter should be increased or decreased by a preset amount. In this respect the 20CN and 21CN systems differ from NGA. In 20CN and 21CN the system decides to increase or decrease a parameter by one or two steps and so the claimant did not argue that this infringed claim 6. However the claimant did contend that the NGA system, by using the safe rate calculation approach described above, fell within claim 6. The argument is that the cap level for a given profile, which corresponds to a maximum rate above but closets to the safe rate, is a "performance target" within claim 6.
- The defendant did not agree. It argued that the cap level was not a performance target. I think this is semantics. The cap level corresponds to the maximum data rate for a given profile. When the NGA system checks whether a profile is the one with a maximum data rate above but closest to the safe rate, it is determining what margin-related parameter value will cause the DSL modem pair to meet a performance target. That is what claim 6 requires. If claim 1 had been infringed by the NGA system, then I find the system also satisfies claim 6.
- The defendant did not dispute that if claim 1 was infringed, all three systems would also infringe claims 8 and 15.
- An additional free standing point arose on claim 15 relating to the NGA system. Although the NGA system does not use the re-train alarm which is present in the 20CN and 21CN systems, that is apparently because, as the defendant puts it in its Amended Response to the Request for Further Information (paragraph 44.9) "this functionality is currently de-activated in the NGA network and has never been operational". The claimant submitted that although this meant the method claims 1 et seq were not infringed, the product claim 15 was infringed because the RAMBo is able to collect the relevant data and so satisfied claim 15.
- Neither side addressed this point on claim 15 in any depth. I do not accept it. If, as the defendant states, the "functionality" is switched off then the system as it stands is not able to perform the relevant task and does not fall within claim 15. It is not alleged that supplying such a system to a third party was a "means essential" under s60 (2) nor is it alleged that the presence of some switched off software in a system could be construed as a threat to switch it on and thereby infringe in future.
- The defendant contends that claim 1 lacks novelty over Hendrichs. It was not in dispute that Hendrichs was published before the priority date.
- A form of ADSL is known as ADSL-Lite. In ADSL-lite no splitter is used to separate the POTS aspect of the system (i.e. the conventional telephone) from the ADSL parts. These are sometimes called "splitter-less" forms of ADSL. The lack of a splitter leads to relatively large changes in transmission environment occurring when the telephone is taken off-hook in the customer's premises. The term "off-hook" derives from the early days of telephony when the mouthpiece was on a hook and was taken off the hook to start a call. The off-hook event and other things like the telephone ringing cause changes in the line's characteristics and can increase the noise on the line. For all the time the telephone is off-hook (in effect during a call) the line will have different characteristics from those applicable when the line is not being used in that way. In order to deal with these sudden changes, spitterless ADSL methods employed a technique called fast-retrain, in which the modems would select a set of previously determined transmissions parameters which have been stored in the modem as profiles. The only alternative when such line changes take place would be for the modems to undergo a full re-train – in other words to completely re-run the training procedure used at initialisation to find appropriate transmission parameters for the line in its current state. The disadvantage of a full retrain is that it is slow, requiring approximately 40 seconds. A fast re-train, in which the modems simply switch to a pre-existing profile, is much quicker than a full re-train.
- Hendrichs was published in July 2003. It is a disclosure of a fast-retrain system. As it states in the Abstract:
- The way Hendrichs works is as follows. At the first initialisation the system generates a set of transmission parameters for the then current line conditions. These parameters are stored in a profile in the modem. The first time an off-hook event occurs, the system has no choice. All it can do is recognise that the line conditions have changed and undergo a full retain. The result of that full retrain is a new set of parameters, tailored to the new line conditions. These parameters make up a second stored profile. Also stored with each profile is information identifying the line conditions which were in existence when that profile was in use.
- Now consider a time in the future when an off-hook event occurs. This time, when the off-hook event is detected, the system has a choice because it has some profiles stored and available. The system compares the new current line conditions with the information it has about the stored profiles and decides whether any available profile is suitable. If so, the system causes a fast re-train to take place to bring the modems to the suitable stored profile. If no suitable profile is in the store, then a full retrain will be required. That will cause delay but completed it will also produce another new profile to add to the store. After a while the system will have built up a collection of stored profiles which it can choose from.
- The claimant contended that the controller in Hendrichs was not logically or physically distinct from the modem pair. This depended on the claimant's construction of "controller" which I have rejected. In my judgment the Hendrichs' system has a controller which satisfies claim 1. The fact the decision making element is inside the modems is irrelevant. The fact that the parameter sent are transmission parameters is also irrelevant.
- In any case I am not convinced that what is disclosed in Hendrichs would be outside claim 1 even on the claimant's construction. The processing required for Hendrichs is run on a processor in the modem but the software which would implement Hendrichs' system is distinct. While not physically distinct, it seems to me that whatever the precise ambit of the claimant's construction, the control in Hendrichs is exercised by something which is "logically" distinct. The processing disclosed by Hendrichs exercises control over the modem and instructs it to act in a particular way. "Smart modem" is a reasonable way of describing Hendrichs.
- By the closing it was common ground that Hendrichs' method involves collecting some operational data, i.e. the data representing the attenuation characteristics of the line. The claimant submitted that this cannot be both current and historical. Whatever it is, it is only one type of data.
- The defendant submitted that if one considers the later time when the system has encountered a off-hook event and is considering whether to use the stored profile, the line attenuation data corresponding to that stored profile represents a past line state and not the current line state. The defendant also submitted that Prof McLaughlin had accepted in cross-examination that Hendrichs collects historical data and contended that this part of claim 1 was satisfied. The argument was that since Hendrichs does a comparison between current line conditions and stored profiles which represent a previous line condition then the claim is satisfied. Both current and historical data are involved.
- I reject the defendant's argument that Hendrichs satisfies the requirement of collecting current and historical data for the following reasons. The issue to which the cross-examination and the expert evidence is relevant is the disclosure of Hendrichs and the manner in which the Hendrichs system works. On that there is no debate. Hendrichs records the attenuation data for the line in a given circumstance and stores that data along with the relevant profile. This data is then used later to decide whether to use the profile to which it corresponds. The issue is whether a system operating in that manner falls within the claim. Prof McLaughlin's testimony in cross-examination has a bearing on the first question (how the system operates) but not on the second (whether that falls within the language of the claim).
- In my judgment to characterise this overall method as one in which the controller collects both current and historical data is not fair or correct. When the controller in Hendrichs stores a profile with the then relevant attenuation data, that seems to me to amount to the step of collecting the data. And the data collected is current data. At that stage nothing further is done with the data by the controller. No analysis step takes place at this point but the data has now been collected by the controller. Later on, when the controller in Hendrichs needs to make a decision about whether to move to a stored profile (or trigger a full retrain instead) the stored data has to be considered and compared to current line data. At that point the stored data does of course represent the conditions in the past but the fact remains that it was current when it was collected. Unless there is another collecting step at this comparison stage then the fact the data may be old is irrelevant. I do not think the comparison of stored data with the current data counts as a collecting step within the terms of the claim even though no doubt the stored data has to be read. It seems to me that this step is the analysis step. Thus the claim is not satisfied.
- It was also suggested that an implementation of Hendrichs would have stacks of data for a parameter called "errored seconds" in the MIB which are current and historical data but I will deal with that issue in relation to inventive step, where I believe it properly belongs.
- The analysis step of claim 1 is satisfied by Hendrichs. The next issue is "generating" a parameter set. The process of selecting a stored profile does involve selecting a margin-related parameter set but does not involve generating such a set as I have construed that term. However the defendant submitted that when the Hendrichs logic requires the system to undertake a full retrain because it cannot find a suitable stored profile, that process generates a profile and so satisfies the claim. The claimant argued that if this sort of full retrain process satisfied claim 1 then the claim would cover a conventional DSL modem pair but, as I have already said, in my judgment claim 1 does not do so. The fact that one step in a claimed method may read onto one step in a conventional method is irrelevant. I prefer the defendant's argument in relation to generating. I find that the controller does generate a relevant parameter set when it triggers a full retrain.
- There is no issue over the final step of claim 1 over Hendrichs.
- Accordingly I find that the method disclosed in Hendrichs does not satisfy claim 1. The claim is novel. All the claims dependent on claim 1 are therefore also novel. No issue arose in relation to claim 15. It contains in effect the same limitation as claim 1 albeit in a product claim (a module to collect current and historical operational data). It also requires a controller to monitor more than one modem pair, which Hendrichs does not disclose either.
- The defendant contends that the relevant claims (1, 6, 8, 15) of the 495 patent are obvious over Hendrichs, Rahamim and Kerpez. The only issue of principle arising on obviousness was an argument about whether the defendant's case over Kerpez was truly an impermissible attack over common general knowledge alone. I will address that in context.
- The structured approach to the assessment of obviousness was set out by the Court of Appeal in Pozzoli v BDMO [2007] EWCA Civ 588, [2007] FSR 37. It is:
- I have identified the skilled person and the common general knowledge above. I have also construed claim 1 and since neither side advanced an inventive concept for that claim, I will not invent one.
- The difference I have found to exist between Hendrichs and the claim is that the method of Hendrichs does not collect both current and historical data.
- The defendant submitted the following were obvious things for a skilled person to do in the light of Hendrichs and bearing in mind their common general knowledge:
- The first point is irrelevant because I have found Hendrichs is within the claim anyway.
- The second point relates to errored seconds. Prof McLaughlin accepted that the DSL Management Standard mandates storing certain parameters in the MIB and one of them is errored seconds. Errored seconds are a way of gauging the error rate in the transmission. The parameter is a count of the number of seconds in a given time period for which the system detected that there was at least one error in the data stream. A second is declared to be an errored second if there was at least one error during that second. The fact there might have been multiple errors in the same second makes no difference.
- Given the terms of the standard, I cannot see that it would have required an inventive step to implement Hendrichs in a system which complied with the standard and stored errored seconds in the MIB but I do not see how that produces a method within the claim. For one thing the parties agree that conventional storage of this information in the MIB is not collecting operational data within this claim. For there to be collection of such data by the controller in Hendrichs, the controller has to collect the errored seconds data from the MIB.
- Simply because modems compliant with the standard would store data in the MIB does not mean it was necessarily obvious for the modem to extract that data from the MIB and use it in a Hendrichs type system. I was not persuaded that a system produced starting from Hendrichs would collect errored seconds data from the MIB. The MIB is a database to be used by network management entities. Hitherto modems never accessed the MIB to retrieve data stored there. There is no reason for a skilled person to think of doing that either from reading Hendrichs itself or from the common general knowledge.
- In my judgment claim 1 is not obvious over Hendrichs. The same therefore follows for claims 6, 8 and 15.
- Rahamim was published in August 2003. It describes a system for link adaptation in DSL lines with a view to improving the quality of the link by reducing crosstalk and avoiding bridged taps. Crosstalk has been described above. The details of what a bridged tap is do not matter. They can cause problems and in particular they can lead to an impedance mismatch which can effect line quality.
- The method described in Rahamim consists of the following steps. The ATU-C (called the Central Office (CO) modem in Rahamim) evaluates certain parameters on the DSL line such as SNR and error rates. Impairments on the link are identified such as crosstalk and bridged taps. The ATU-C works out what adjustment the ATU-R (called the Customer Premises Equipment (CPE) modem in Rahamim) should make to improve line performance. The ATU-C sends an instruction to the ATU-R to dictate the adjustments to make and then the two modems establish an adjusted communications link. The two particular proposals in Rahamim are for the ATU-C to dictate a new carrier frequency for the upstream band and to dictate a new power level.
- There are two putative distinctions between claim 1 and Rahamim: no controller and no collection of current and historical data. The first point is no distinction on my construction of "controller".
- On the second point, it is clear that as described, the data evaluated by Rahamim is current data. The defendant's obviousness argument was that it was obvious to put Rahamim into practice in such a way that the controller would in fact evaluate (and therefore collect) historical data as well.
- At paragraph [0046] Rahamim states that:
- As written there is no reason to regard any of these parameters as anything other than current data, i.e. data representing the current line conditions when the data is collected. That is why Rahamim does not deprive claim 1 of novelty on any view. The defendant's argument focussed on the reference to bit error rate (BER) and how one would implement such a system. The starting point is that it is very difficult to measure BER.
- The defendant's argument was that instead of measuring BER, the skilled person would use the data stored in the MIB in accordance with the DSL standard to perform the analysis. The standard does not require the storage of bit error rates but it does require storage of errored seconds (see in relation to Hendrichs above) and another similar parameter – severely errored seconds. The defendant submitted that Dr Adams' view was that one could get the error data mandated by the standard and use that to evaluate the line. The defendant also relied on Prof McLaughlin's evidence in cross-examination that although using the data mandated by the standard (errored seconds etc.) would give a poor estimate of the actual BER, that data would be a reasonable means of evaluating the performance of the line. So the defendant submitted the claim was obvious because this would therefore involve collecting data from the MIB. Since the data stored in the MIB would include both current and historical data, both kinds of data would be collected.
- I do not accept this argument for the following reasons:
- I find claim 1 is not obvious over Rahamim. The same conclusion will follow for claims 6, 8 and 15.
- The Kerpez article was published in the IEEE Communications journal. The IEEE is the Institute of Electrical and Electronics Engineers. The IEEE publishes numerous journals. The Communications series contains high level papers providing general overviews of areas of technology and Kerpez is one such.
- Kerpez was published in September 2003. The article describes the management of DSL lines. It suggests that current DSL lines are conservatively managed and refers to the problem of crosstalk. The article proposes ways of increasing bit rates and decreasing failure rates. The abstract states as follows:
- After a reference to the advantages which are said to flow from this, the abstract also refers to Dynamic Spectrum Management (DSM) as a way of addressing crosstalk.
- In the body of the article, the current method of provisioning DSL is described, referring to the conservative practices and the problem that many DSL modems are set to transmit at higher power than necessary. A "vision" of advanced DSL management is presented in which the properties of loops are stored in a database which can be accessed by something called an "analysis engine". The article then contrasts static and dynamic spectrum management as methods of addressing crosstalk and refers to the various levels of DSM (0, 1, 2 and 3) which were common general knowledge. At p118 (RH side) is a reference to DSM being administered by a spectrum management centre (SMC) associated with the service provider or with a DSLAM or "the DSL modems themselves could be allowed to autonomously adapt to their crosstalk environments".
- Next there is a section on diagnosing DSL impairments, which also refers to the DSL analysis engine idea and to how to remedy the problems identified ("remediation"). Amongst other things, Kerpez states that many parameters such as power and bandwidth can be varied. The types of remediation for different classes of impairment is summarised in Table 1. The remediation can be "plant remediation", i.e. physical changes to the equipment ("plant"), or "electronic remediation", i.e. changes to electronic settings. So for example plant remediation for crosstalk may be swapping pairs (in other words physically swapping lines) while electronic remediation may be by reducing the cross talker's power or by implementing DSM or performing joint spectral optimisation. The electronic remediation proposed for impulse noise is to increase interleaver depth. The idea of electronic remediation involves signals coming from the DSL manager (which contains the analysis engine) to the modems. This is illustrated by the double headed arrow line drawn between the advanced DSL manager and the DSL modems in figure 2.
- The next section addresses data collection, which refers inter alia to the MIBs in DSL modems as a potential source of information to be extracted. This section is followed by a section on the "bottom line", considering whether the advanced DSL management proposal is worth the effort in terms of cost and the likely improvements which will flow from it. A summary at the end of the article proposes that:
- There is no question that Kerpez is written at a high level of generality and contains a wide variety of ideas. Also I am sure the skilled reader would see it as an advertisement both for advanced DSL management as a technique and the skills of the authors, who all work for Telcordia Technologies Inc., as experts who will be able to help someone implement an advanced DSL management system. However these points do not alter the fact that Kerpez is a technical article written to those skilled in the art and appearing in a technical journal. It would be taken seriously by the skilled reader. The fact that Kerpez reflected the common general knowledge at the time does not mean the skilled person would ignore the article or give it less weight.
- Comparing the disclosure of Kerpez with claim 1: the article discloses the concept of using a management system remote from the DSL lines to gather and store information about those lines and to analyse and use that information to control the lines automatically. The concept is a proposal rather than an existing system which has actually been built and is not the only thing proposed in Kerpez. I have asked myself whether it is an exercise in hindsight selection to focus upon this concept in the article. In my judgment it is not. The concept is clearly articulated in the paper and would be seen as such by the skilled reader.
- In terms of the language of claim 1: the proposal involves using a central controller communicably attached with DSL modem pairs. The controller is physically remote from the DSL modems and so would satisfy either party's construction of "controller". It would collect operational data from the DSL modem pair, such as data from the MIBs. It would have an "analysis engine" which will analyse the collected data and make decisions. The idea of managing parameters which are margin-related is disclosed as is the idea of electronically instructing the DSL modem pair to operate in accordance with the parameters set by the controller.
- The differences between claim 1 and the disclosure of Kerpez are (i) the disclosure is a proposal and work would need to be done to put it into practice; (ii) there is no explicit reference to the idea that the operational data to be collected should be both historical and current operational data, and (ii) there is no explicit description of how the parameters are produced. Kerpez does not mention "generating" a parameter set as I have construed the term.
- The claimant's case is that it was not obvious to do anything over Kerpez. It is a "chatty" document. No blueprint for an actual system is provided. The DSL analysis engine which is at the heart of Kerpez's vision is based on the idea that it is possible to break down the received noise on a line into its separate elements (impulse noise, electromagnetic interference (EMI), crosstalk, background noise) and to go further and identify individual sources of crosstalk. However Kerpez does not explain how this would be done and to do so would be very difficult and would require a "research project". The claimant submitted that Dr Adams accepted the points just made and also referred to Prof McLaughlin's evidence that to carry out such a task was very difficult.
- The defendant's case was that the claim was obvious over Kerpez. It argued that Kerpez was selling the idea of DSM and the skilled person would be well aware of the potential of such a system. The skilled person would know the information which was available from the MIB as a result of the DSL Management Standard. The defendant relied on Dr Adams' evidence. He distinguished between identifying sources of crosstalk (which was hard) and identifying sources of noise more generally (which he thought was less difficult).
- The defendant also argued that identification of sources of noise was not a necessary component of the Kerpez disclosure and even if that was too difficult, it was obvious to produce a DSL management system which did not do that. However I was not convinced by this. Given the prominence in the article to noise identification, the argument that it was obvious to produce a DSL management system over Kerpez without embarking on that exercise is based on hindsight.
- The claimant objected that the defendant was seeking to put an unpleaded case of obviousness over common general knowledge alone to Prof McLaughlin, based on a mosaic of the standards and the DSL Advances textbook. It argued that it would be unfair to permit such a case to be run since a case based on common general knowledge alone was expressly disavowed in a Part 18 response a long time ago. Moreover the court should be wary of such attacks having regard to the warning expressed by Floyd J in Abbott v Evysio [2008] EWHC 800 (Pat) at paragraph 180 (and in other cases).
- If I thought the defendant was running an unpleaded obviousness case based on common general knowledge alone, I would not permit it. However the defendant did not accept that its case over Kerpez was such a thing. I agree. The fact that Kerpez contains or embodies the common general knowledge does not mean an obviousness case over Kerpez is an unencumbered attack based on common general knowledge of the kind referred to by Floyd J.
- The major issue is the debate about the references to identifying sources of noise. One of the ideas is to identify the sources of crosstalk. Dr Adams accepted this was difficult but as I have said he distinguished between identifying sources of crosstalk and distinguishing between crosstalk and impulse noise. In his view the latter was a simpler problem and was within the capabilities of the skilled person to solve with the data he knew was available in the MIBs. This aspect of Dr Adams' evidence came out most clearly in his oral testimony given spontaneously at the start of the court morning on Day 6 after he had been asked about identifying sources of noise in cross-examination the previous afternoon. The claimant criticised Dr Adams for this and contended it showed a lack of objectivity. I have already rejected that. It is clear that Kerpez describes both identifying sources of crosstalk and more generally distinguishing between crosstalk and impulse noise and I can understand why Dr Adams wanted to ensure that the distinction was clear in his oral testimony.
- Against Dr Adams' opinion that distinguishing between crosstalk and impulse noise was within the capability of the skilled person I have Prof McLaughlin's opinion to the contrary. The claimant submitted that Prof McLaughlin's opinion was that implementing Kerpez would be problematic due to the difficulties in separating sources of noise and referred to paragraph 149 of his second report and two passages in cross-examination. Prof McLaughlin's clearly expressed opinion was that the issue of distinguishing sources of noise would be extremely problematic and would not be pursued (e.g. Day 4 p540 line 11-14). Nevertheless it was also clear that he drew a similar distinction to Dr Adams. He thought identifying individual sources of crosstalk was very difficult but he did accept that one could surmise that a burst of errors was caused by impulse noise, albeit that this was surmise and the cause may not have been impulse noise but a rapid change in crosstalk.
- The claimant pointed out that Prof McLaughlin's paragraph 149 was not traversed in cross-examination. This is true but Mr Wyand put his client's case to Prof McLaughlin fairly and fully. It cannot be necessary to traverse every sentence of every paragraph of an expert's report on the issue of obviousness. The fact this passage was not directly challenged goes to its weight but does not mean I am bound to accept it.
- It is clear to me that if the skilled person would think reading Kerpez that the concept proposed required the identification of individual sources of crosstalk, then the defendant's case must fail. To embark on trying to do such a thing would involve an inordinate amount of research with a highly uncertain outcome. The claim would not be obvious in this case.
- However I reject the idea that the skilled reader would see identifying individual sources of crosstalk as a necessary aspect of the concept proposed. The proposal includes the still not trivial but less ambitious proposal of distinguishing between crosstalk and impulse noise. On this I must decide between Dr Adams and Prof McLaughlin's opinions.
- Standing back, I prefer the evidence of Dr Adams on this point, for the following reasons:
- I find that the skilled person would not be put off from seeking to implement Kerpez in the manner I have described by the fact that the system operated by identifying sources of noise. The work involved would not be trivial but it would not involve an inventive step to set out to produce and to actually produce a system based on Kerpez in this way.
- Kerpez does not expressly require the collection of both current and historical data but in my judgment it would be obvious to collect both. That is because the skilled person would see that one of the sources of data specifically identified by Kerpez is the MIB in each of the modems. The skilled person would know that the MIBs mandated by the DSL Management Standard will contain both current and historical data which is there in order for it to be collected and used for management purposes. Collecting this data involves no inventive step.
- If generating bears the meaning contended for by the claimant then there is no distinction over Kerpez in this respect since it does not matter how the parameter set is produced – the controller can create it or select it.
- The only point advanced by the defendant in support of obviousness over Kerpez on the narrow construction of "generating" was the following. Dr Adams expressed the view that the DSL management system proposed by Kerpez would set the configuration parameters for the DSL modems. I agree. He also thought that this meant that the claim would be satisfied on the narrow construction because the modem pair will then create (i.e. generate) their own transmission parameters during initialisation based on the configuration parameters provided. However that is not within the claim since the claim requires the generating to be something done by the controller and to come before the step of instructing the modem. This is unlike the arrangement in Hendrichs in which a controller inside the modem triggers a full retrain. Following Kerpez the controller will be physically distinct from the modem. The margin-related parameter set which the modem is instructed to use by the controller in this instance is the set of configuration parameters. It is not the transmission parameters generated by the modems after a retrain.
- I do not see why it was obvious for the controller to create the configuration parameter set anew. It could simply select from a pre-existing group of profiles and choose one of those. No-one suggested generating such a set was obvious. I find that claim 1 is not obvious on the narrow construction of "generating".
- Before leaving claim 1 I will consider its position on the wide construction of the term generating. On that basis there would be no difference from Kerpez arising from the manner in which the parameter set to be sent to the modem was produced.
- Kerpez is a very different kind of proposal from Hendrichs and Rahamim. Although I have rejected the claimant's construction of claim 1 on the controller issue, that does not mean that there is no difference between systems in which the control is located at and associated with the modems themselves (like Hendrichs and Rahamim) and arrangements such as that proposed by Kerpez and at least some embodiments of the 495 patent (not the smart modem) in which the control is not undertaken at the modem but is centralised and separate (e.g. fig 3 of the 495 patent). Equally the distinction between configuration parameters and transmission parameters is a real one, even if it is not reflected in the claim language. What Kerpez discloses is very close to the ideas shown in figure 3 of the 495 patent. That is a remote DSL management system which can read data from the MIBs of the modems and can set configuration parameters for the modems to use.
- The claimant's best point is that the concept I have identified as disclosed in Kerpez is not the only idea in the document. The document is in very general terms and discloses all sorts of concepts which, if implemented, would not fall within claim 1. Examples include setting parameters which are not margin-related (such as interleaving depth) and taking manual remediation steps. There are others. However I think claim 1 would be obvious over this article on the wide interpretation of "generating". The claim on that basis is a broad claim to an automatic DSL manager. It is limited to using margin-related parameters but they are disclosed and would be obvious, and it is limited to collecting both current and historical data but the standard already provides for this sort of data to be stored in order for it to be collected and used for managing DSL lines. Claim 1 read broadly is little more than a claim to automating the management of DSL lines which was already being conducted manually. A narrower claim to a particular technique would stand in a different light but in my judgment, having preferred Dr Adams' view about the feasibility of the idea to identify noise sources over that of Prof McLaughlin, there is nothing inventive in claim 1 as it is construed by the claimant.
- Over Kerpez claim 15 stands or falls with claim 1. That is because Kerpez clearly contemplates managing multiple modem pairs. So claim 15 is valid. However if claim 1 was obvious claim 15 would be obvious as well.
- Since claim 1 is valid, so too are claim 6 and 8. I will consider claims 6 and 8 on the wide construction of "generating", starting with claim 8.
- Claim 8 simply comes down to requiring that the system will compare a margin-related parameter value reported by the modem with a threshold to see if a target is met. Dr Adams' view was that since Kerpez suggested analysing the impact of various performance affecting impairments such as noise, the comparison required by claim 8 would almost certainly be required. Prof McLaughlin did not disagree. I accept Dr Adams' view. I think claim 8 would be obvious if claim 1 was obvious.
- Claim 6 requires that the system should not simply compare a margin-related parameter value with a threshold (like claim 8) but should actually determine the value of such a parameter which would cause the modem pair to meet a performance threshold. In other words claim 8 considers the data coming from the modem while claim 6 goes further and tries to determine what the effect of a value to be sent to the modem might be. Dr Adams' view was that the analysis proposed by Kerpez would necessarily include carrying out the method of claim 8. I did not have my attention drawn to any evidence from Prof McLaughlin to the contrary but the claimant did point out in argument that there are simpler ways of proceeding than complying with claim 6. A much simpler system would be to move up and down a list of possible profiles as the defendant's 20CN and 21CN systems do. It seems to me that the claimant is right. The fact that the 20CN and 21CN systems do not infringe claim 6 shows that it is not inevitable that a DSL management system has to operate this way. The 20CN/21CN method is a simpler approach than claim 6. I bear in mind that this point was not put to Dr Adams, nevertheless, it shows that performing the test required by claim 6 would not be inevitable over Kerpez. Since that is the only basis on which claim 6 is said to be obvious, the argument falls away. Thus I find that even if claim 1 was obvious, claim 6 would not be.
- I find that none of the defendant's systems (20CN, 21CN and NGA) infringe the 495 patent. The patent is valid. However if the claims were wide enough to cover the defendant's systems, claims 1, 8 and 15 would be obvious over Kerpez while claim 6 would not be. Thus on the wide construction of the 495 patent, only the NGA system would infringe a valid claim.
- The 790 patent is concerned with managing line profiles in a DSL system. The ambit of the term line profile or profile was in dispute and I will address it below. At this stage it is fair to note that the example of a line profile set out in paragraph [0003] shows a set of configuration parameters for a given line.
- The 790 patent is concerned with the problem of how to choose which profile from a set of possible line profiles for a given line should be selected. Moving from one profile to another is referred to as a transition. Paragraph [0006] states:
- The patent uses two concepts which are key to the method of selecting profiles it discloses. These are priority and feasibility. Neither expression is a term of art and the skilled reader's understanding of them would come from reading the patent itself. Essentially the idea is that the system will extract out from a set of possible profiles which a line could adopt, a set of feasible profiles. That is a set of profiles which would provide acceptable operation of the line in the given circumstances. The set of possible profiles is divided into those which are feasible and those which are infeasible. This still presents a problem since the system has to have a means for selecting a particular profile to use from the group of profiles which are determined to be feasible. This is the point at which the priority concept plays a part. The possible profile transitions have an allocated priority which is independent of line conditions. It is an inherent property of each transition from a given profile to each target profile. The transition to a feasible profile with the highest priority is the one which is selected. The DSL line is then operated using this profile.
- To carry out the method disclosed, the patent describes the use of a profile state transition matrix. The profile state transition matrix is a matrix which records the state transitions which are allowed and contains the priority of each possible transition. An example is matrix T in figure 4 as follows:
- In this matrix each column relates to a given starting profile. So if the current profile for the DSL line in question is profile 1, the first column is the relevant column to use. If the current profile was profile 2, the second column would be the relevant one, and so on. Thus although T is a matrix, when one is considering a given current profile the data is really just a single column of values, i.e. a list. The numbers in the list represent the target profiles. Thus starting with profile 1 as the current profile, profile 2 has the highest priority since it is at the top of the list. Profile 6 has the next priority, then profile 1 itself and then profile 5. The presence of an entry for the current profile in the list shows that this arrangement would prioritise changing to profiles 2 or 6 (if they were feasible) ahead of staying put at profile 1. On the other hand if profiles 2, 6 and the current profile 1 are found not to be feasible but profile 5 is feasible then profile 5 should be selected.
- Claim 1 of the 790 patent has been set out above. The meaning of virtually the entire claim was in dispute and many of the issues were interrelated. Some of the issues can be addressed distinctly but some are best seen in the context of the infringement arguments. I will address the points which can be dealt with distinctly first. They are:
- After dealing with these I will turn to infringement.
- It is convenient to start with line profile. The claimant contended that the skilled reader would understand that a line profile was a set of configuration parameters and not transmission parameters. The claimant referred to paragraph [0003] of the patent and the example of profile given there. Paragraph [0003] states that a "line profile" specifies parameters such as data rate, power spectral density (PSD), margin, FEC parameters and a carrier mask for a particular DSL line attached to a DSLAM. The example profile consists of configuration parameters. I agree that based on this passage, a line profile appears to be a set of configuration parameters. However the defendant, supported by Dr Adams, referred to claim 3 which lists a number of parameters which form part of a profile. Dr Adams' view was that some of those parameters were transmission parameters rather than configuration parameters.
- The three parameters referred to were data rate, forward error correction coding (FEC coding) and power spectrum density (PSD). In fact these terms are rather ambiguous. For example although the specific transmission data rate on the line at a given time is not a configuration parameter, the maximum and minimum data rates are configuration parameters. Similarly PSDMASK is a configuration parameter. The defendant's best case was about the FEC coding. In the context of FEC, there are coding values which are set and known as N, K, S and D. These values are undoubtedly set by the modems and are transmission parameters. Dr Adams' view is that these are what claim 3 referred to, particularly by using the word "coding".
- However the claimant contended that there are configuration parameters which can be set in order to influence the modems' choice of actual FEC settings and that is what claim 3 was referring to. One example is MAX DELAY, which has an effect on interleaving. Another is INP (impulse noise protection). The claimant also referred to paragraphs [0046], [0053], [0064] and [0103] which showed that the specification contemplated specifying configuration parameters in order to influence FEC levels. On the other hand, as the defendant pointed out, the last two parameters in the list in claim 3 (impulse noise protection and delay) are in effect the configuration parameters the claimant is relying on for this argument, so why read them into the reference to FEC coding?
- Despite the defendant's argument, I accept the claimant's submission. Apart from claim 3 the patent is clearly not talking about setting transmission parameters. Claim 3 is slightly ambiguous but I do not find the ambiguity sufficient to displace the overall view which I believe a skilled reader would form reading the specification as a whole. A line profile in the 790 patent is a set of configuration parameters.
- The argument about "DSL controller" in the 790 patent is closely related to the argument about the word "controller" in the 495 patent. Again it is an ordinary English word. Both sides made the same submission as before. They also contended that the term "controller" in the 790 patent meant the same thing as it meant in the 495 patent although they both accepted that in principle this did not necessarily have to follow. I suspect that each side said the meaning was the same in both patents for their own forensic reasons. The defendant's better case was on 495 and wanted that conclusion to follow for 790. The claimant's better case was on 790 and wanted the reverse to happen.
- Importantly in the 790 patent the only reference to a smart modem is less clear than the disclosure in the 495 patent. There is nothing in the 790 patent which shows that the patentee specifically intended to cover an arrangement like the smart modem of the 495 patent. In the 790 patent, the only clear representation (either in words or a diagram) of the physical relationship between the controller and the ATU-C is in figure 3A of the 790 patent. It shows a controller physically distinct from the ATU-C. The only reference to a smart modem at all in the 790 patent is in paragraph [0020]. The 790 patent has no text corresponding to paragraph [0052] of the 495 patent, no diagram depicting a smart modem like figure 12 of 495, and no claim referring to a smart modem.
- Given that a line profile is a set of configuration parameters, the controller of claim 1 of the 790 patent must be distinct from the modem pair operating the DSL line. To that extent I accept the claimant's submission although I am still not sure I understand the ambit of the claimant's term "logically distinct". The controller is distinct and its function is to provide a set of configuration parameters (i.e. a line profile). Something which exercises control merely by changing the transmission parameters on the line is not exercising the kind of control the patent is talking about. It is not a controller as that term would be understood in the patent and in the claim.
- There is no doubt what a profile state transition matrix is. For a given current profile, a list of profiles indicating the priority of the transitions would satisfy the claim.
- The debate can be best understood by considering the claimant's infringement case and the meaning of the following phrase extracted from claim 1:
- The claimant's infringement case was put in two ways. One case was based on what the defendant called reverse engineering. The other case was based on identifying a table (the profile cap level table) which is actually used by the NGA system and asserting that it is a profile state transition matrix.
- The first case focussed upon the overall logic used for the NGA system to govern the transitions between profiles. This overall logic has been expressed in flow charts. Although the logic does not actually use a matrix, the argument was that a profile state transition matrix could be derived from it. That derivation process was undertaken by Prof McLaughlin and the profile state transition matrix he derived was set out in his evidence. The claimant's case was that the claim would be satisfied by a system which could be represented in this way even if the software logic did not actually run that way. The claimant argued that the term "implemented" in claim 1 means that it does not matter how the profile state transition matrix is implemented and a system like the defendant's will infringe. The claimant also referred to paragraph [0132] of the patent which distinguishes "between the method of operations in operating a computer and the method of computation itself". In other words the claimant argued that a system whose logic can be shown to be the mathematical equivalent of logic based on a profile state transition matrix will fall within the claim even if the system does not actually have such a matrix in it.
- This first case involves a point on the construction of claim 1. Does the claim cover a system in which no actual profile state transition matrix is used in the software or hardware but in relation to which a profile state transition matrix can be derived?
- I reject the claimant's submission on construction. The skilled reader would see that using an actual profile state transition matrix is at the heart of the disclosure and is at the heart of the claim. The fact that the profile state transition matrix, as well as the various sub-rules, threshold tables and overall rule, can be implemented in hardware or software or a combination of both does not mean that they can be ignored. Nor does paragraph [0132] mean one can ignore the method by which the system actually operates. The invention claimed in claim 1 is a specific method which works in a particular way. How it works is important. The fact that two different methods can be shown logically to produce precisely the same outputs for the same inputs does not mean the methods are the same or can be said to work in the same way. One may be faster than the other. One may be more flexible than the other. The skilled reader of the patent would understand that there may be advantages to implementing the logic using an identifiable profile state transition matrix (whether in hardware or software). In particular using a matrix has the advantage of flexibility. By actually setting out the priorities in a matrix, it would be a simple matter to change the priorities by replacing one matrix with another. That is not the case for a system in which the logic is hard coded into the software even if that logic could be shown to be mathematically equivalent to a matrix.
- A further virtue of this construction of claim 1, requiring an actual matrix to be identifiably present, is clarity. Prof McLaughlin's attempts to derive a profile state transition matrix for the NGA system by a process of reverse engineering showed how difficult this was. His matrix exhibit went through a number of changes and in the end the cross-examination demonstrated that it was not correct. I do not think the skilled reader would read the 790 patent as contemplating that a complex exercise of that kind would be required to know whether a particular system employed the claimed method.
- The claim refers to a plurality of target line profiles on a number of occasions. A critical issue of construction arises from this phrase. The claimant contended it does not exclude a system in which the claimed method operates only in respect of a subset of all possible transitions. The fact that there may be other possible target line profiles than those identified in the profile state transition matrix is immaterial on the claimant's construction. The defendant contended to the contrary. Only a system in which every possible transition is within the plurality of target line profiles would be within the claim. The claim excludes a system in which the claimed method operates only in respect of certain transitions. Such an arrangement would make a nonsense of the claim. The defendant also submitted that both experts agreed a complete list of possible transitions had to be within the matrix.
- It is correct that Prof McLaughlin believed that the profile state transition matrix had to have all possible transitions in it. He accepted it in cross-examination. I will take it that Dr Adams had the same view although the defendant did not draw a particular passage to my attention in closing either orally (see day7 p869 and p921-922) or in writing (see paragraph 39ii of the defendant's closing). However this point is a matter of construction and is for the court and not the witnesses.
- I prefer the claimant's submission. It is true that the specification never mentions a system in which the method of the invention is used only for a sub-set of all possible transitions but that does not mean the claim should exclude such a method. I can think of no reason why the skilled reader would think the patentee intended to exclude a method which used the profile state transition matrix and the other steps in claim 1 to choose between a given subset of target line profiles in some circumstances but used a different method in other circumstances or for other profiles. The method would still be taking the benefits of the claimed method in the circumstances in which it was using it. The language "a plurality of target line profiles" is apt to cover such a case.
- Possible transitions are transitions which a DSL line operating in a current profile is, in theory, allowed to make. In the profile state transition matrix T shown above the four profiles in the left hand column are the possible transitions from profile 1. Feasibility is an assessment, made by the logic of the system itself, about whether a given target line profile is likely to give suitable performance. The possible target profiles are assessed as being feasible or infeasible by the operation of the sub-rules which evaluate (but do not finally determine) feasibility, the threshold tables which emphasise or de-emphasise individual sub-rules and the overall rule that makes the decision which assesses a given profile as feasible (or not). That decision is what the claim calls determining the feasibility or infeasibility of a line profile.
- The method selects one of the plurality of target line profiles in which to operate the DSL line. The selected profile is whichever of the current profile and target line profiles has the highest priority and is not infeasible. Thus priority is the means by which the system chooses between feasible profiles. The priority of a transition is independent of the particular line conditions on the current profile.
- The claim requires the rules to be based on this sort of data. "Reported data" is data directly observed from the operation of the line in a given state whereas "estimated data" is data that is estimated by the controller relating to the performance of the line. The reported and estimated data can relate to performance either of the current or target line profiles.
- The defendant contended that the claim required that the calculation of feasibility or infeasibility had to be carried out for each of the possible target line profiles, i.e. all of them. The claimant did not agree on two grounds. First the claimant relied on the opinion of Prof McLaughlin that the defendant's construction would make no technical sense and second the claimant argued the point as a matter of construction. I can dismiss the first point. Dr Adams did not agree with Prof McLaughlin and Prof McLaughlin's opinion on this point did not survive cross-examination.
- The claimant's second ground started with the observation that the claimed method refers to both "evaluating" feasibility and "determining" feasibility. The claimant contended that "determining" the feasibility did not mean the same thing as "calculating" and so, since the purpose of the feature was to select the profile which has the highest priority in the matrix which is not evaluated to be infeasible, if the system was able to arrive at the correct profile without checking every profile then there was no reason to require every profile to have its feasibility actually calculated. The claimant referred to paragraph [0052] in which the patent contemplates deeming the profile with the lowest priority to be feasible without examination.
- I think the claimant is correct. In appropriate circumstances, deeming a profile to be feasible without examination is contemplated and so a calculation for every profile is not necessary. There would be no reason to examine the feasibility of a profile which had a lower priority than a given feasible target profile.
- However I have reached this construction in the context of my interpretation of the claim overall in that it requires an actual profile state transition matrix to be used. Such a matrix will specify a priori the relevant priorities. The context in which this argument actually arose was concerned with aspects of the logic of the defendant's method which did not involve the profile cap level table and which depended on the claimant's argument that a profile state transition matrix could be reverse engineered from the logic. If the claim is broad enough to cover such a method, in which there is no actual profile state transition matrix which expressly specifies the priorities of the target profiles, then it is not clear to me how the argument I have accepted would work. Without a table which already specifies the priorities of a set of profiles it is not clear to me how it can fairly be said that a profile which is never evaluated has a lower priority. Happily however I do not have to resolve that issue on my interpretation of claim 1.
- Only the defendant's NGA system is alleged to infringe the 790 patent. The 20CN and 21CN systems are not said to infringe. As with the 495 patent, what is alleged to infringe is the method employed by the RAMBo to decide whether to change the profile for a given DSL line.
- As mentioned in relation to the 495 patent, the logic used by the RAMBo is in three steps. First an ILQ is set which indicates line quality. Next, if the ILQ indicates that a change may be required, logic is run to select a new profile (or stay with the existing profile). Flowcharts were provided for red profile selection logic and for green profile selection logic. There are separate profiles and separate flowcharts for the upstream and downstream lines. The logic is slightly different as between upstream and downstream but nothing turns on that. Broadly the red logic may reduce the cap rate or increase interleaving whereas green logic may increase the cap rate or reduce interleaving. Finally if a new profile is required, appropriate signals are sent to the DSLAM.
- Taking the language of claim 1 and dividing it into labelled features:
- I will address infringement in two parts, first regarding the profile state transition matrix issue and second regarding rules and thresholds.
- Much of the time at trial was spent poring over the flow charts. The three key flowcharts are one which represents the overall logic flow which decides what action to take if any based on the ILQ and two flow charts representing downstream red logic and green logic. As I have mentioned already, the claimant had two infringement cases. One was based on the evidence from Prof McLaughlin that the whole of the defendant's process could be represented by a profile state transition matrix he had reverse engineered from considering the logic. The other was based on the profile cap level table which is used as part of the NGA processing in certain circumstances.
- On my construction of claim 1, this argument fails in any event. However I will consider the point assuming the claimant's construction of the claim. The arguments on this were complex but I will address only two: first the reverse engineered profile state transition matrix, and second the argument that a priority between interleaving and cap rates could be derived from the green logic.
- A critical aspect of the claimant's reverse engineering case was that a unique priority for every possible transition permitted by the defendant's NGA system could be determined independently of line conditions and then set out in a profile state transition matrix. Prof McLaughlin said that the NGA system could be considered as a "finite state machine", in other words a machine which could adopt a finite set of states. All the transitions between the finite number of states could be identified and a priority allocated to them.
- However in cross-examination it emerged that Prof McLaughlin had made an error relating to the fact that in certain circumstances the red profile selection logic is run twice. The relevant circumstance is not important (it was when the "ILQ(R)" is red (ILQ(R) is a retrain ILQ)). He accepted that this meant that his final attempt at specifying a profile state transition matrix (exhibited as SML23) was wrong. He was not able to produce a profile state transition matrix for the NGA system. Without such a reverse engineered profile state transition matrix this limb of the claimant's case on infringement must fail even on the claimant's construction of the claim. I do not doubt that the NGA system can be considered to be a finite state machine, as the Professor explained, but in my judgment the Professor's difficulty shows that it does not follow from this that unique priorities, independent of line conditions, can be allocated to all transitions. To show that the NGA system can adopt a finite set of states is not sufficient to establish infringement. Moreover Prof McLaughlin's difficulty shows that it does not follow that any method which uses multiple profiles must be one for which there can be said to be a profile state transition matrix.
- Since Prof McLaughlin's error did not arise from the green logic aspect, the claimant contended that despite the error, the reverse engineering case could survive based on the green logic alone. I am not convinced that that is right but in case it is I will now address a second aspect of the reverse engineering case, focussed on the green logic.
- In some circumstances the ILQ indicates that green logic should be run. The green logic has a number of steps. The first important decision step in the downstream logic (step B) is an operation which provides a YES or NO answer. If the answer is YES the logic branches one way whereas if the answer is NO the logic branches another way. The YES branch may reduce the interleaving level while the NO branch may increase the cap level. In fact the NO branch includes some further questions (Step C and within step G) which might send the flow back to the interleaving branch but that is a detail and makes no difference to this analysis.
- Importantly on the YES branch, the interleaving level may either be reduced by one or left as it is but whether the interleaving changes level or not, there cannot be a change to the cap level. On the other hand on the NO branch (subject to steps C and G) the cap level may be increased (or left alone) but if the cap level is to be increased, the interleaving level cannot be altered. So the green logic can lead to three possible outcomes overall: no change, reduce interleaving or increase cap level. These are the only possibilities. The logic cannot change both the interleaving and the cap level at the same time. The decision whether to actually change interleaving or cap level depends on line conditions.
- A key part of the claimant's infringement case here involved arguing that this logic showed that the system gave a higher priority to increasing cap level over reducing interleaving. The defendant did not agree. One aspect of the claimant's argument and Prof McLaughlin's evidence was based on the defendant's marketing materials but I reject that. The priorities, if they exist, must be derived from the logic of the method used and not from assessments about the subjective views of the operator. Considering the logic itself, I fail to see how it is possible to state meaningfully that this green logic gives a higher priority to increases in cap level over interleaving. The logic simply chooses which branch to take in certain circumstances. The choice depends on line conditions. In this logic there is no priority between interleaving and cap rates.
- Accordingly for these two distinct reasons, I reject the claimant's case that the NGA logic as a whole, or the entire green logic as a part of it, infringes the claims.
- The infringement case based on the profile cap level table is much simpler. The branches concerned with changing interleaving are irrelevant. In certain circumstances the logic will try to increase the cap rate (on a green ILQ) or decrease the cap rate (on a red ILQ). In order to do this the system runs a loop which I have already described in relation to the 495 patent. The safe rate is calculated and used to identify the correct profile to be selected. The correct profile is one with a maximum data rate just higher than the safe rate. In order to find it the NGA system uses a table of profiles, sorted in order of cap level. The logic starts at the highest cap level profile and works out if that should be selected using the criterion I have described above in the section on infringement of the 495 patent. If the criterion is not satisfied then the logic repeats the test but with the next highest cap profile. This is the loop. The process continues moving down the profiles until the criterion is satisfied.
- The claimant contended that the profile cap level table was a profile state transition matrix. The defendant pointed out that it did not cover all possible target profiles. This is true. It takes no account of interleaving. However for the profiles which it addresses, the profile cap level table satisfies all the requirements of the profile state transition matrix in claim 1. The table indicates "a plurality of possible transitions from the current line profile to one of the plurality of target line profiles, and a priority value specifying the priority of transitions for each profile". The priority value indicated by the profile cap level table arises from the fact that the profiles are ranked in order of data rate. The result is that the higher data rate profiles are prioritised. When the loop is operating, the table represents all the possible transitions and the table indicates the priority of all those transitions. For the line profiles to which the profile cap level table applies, the elements of claim 1 are satisfied and since I have construed the term "plurality of target line profiles" to allow for a case in which some but not all possible targets are dealt with, the fact that the table does not prioritise every possible transition in all circumstances does not matter.
- In operating in this way the NGA logic will also satisfy feature (l) of claim 1. It selects one of the plurality of target line profiles in the profile cap level table in which to operate the DSL line. Furthermore the selected line profile will comprise whichever of the current line profile and the plurality of target line profiles in the table has a highest priority in the profile state transition matrix (i.e. is highest in the table), and is not evaluated to be infeasible (because its rate is too high).
- The claimant advanced a number of different infringement arguments in relation to these features. Given that I have rejected the case based on the overall logic but accepted the case based on the profile cap level table, I need to bear that in mind when considering the other elements of the claim.
- The claimant relied on two "sub-rules" said to be employed by the green logic. The first is the rule governing the decision at Step B of the green downstream logic. Part of the decision in step B depends on whether the minimum margin is less than or equal to the minimum margin threshold. If that is so then the logic will go down the YES branch which may change interleaving but will not go down the NO branch which changes cap level. The claimant submitted this was a feasibility test. I agree. If the YES branch is selected, the logic will not use the loop logic or the profile cap level table at all. In other words the target profiles in the table with higher cap levels will have been evaluated to be infeasible.
- The second sub-rule relied on is the rule governing the test applied in the loop which runs through the profiles in the profile cap level table. The claimant submitted that this evaluates the feasibility of the individual target profiles in turn. I agree. Thus I find that feature (g) is satisfied.
- This is about threshold tables. The claimant relied on the threshold values used at the ILQ allocation stage. A table of thresholds used at this stage by the NGA logic was set out in tables in Appendix D to the Product and Process Description. They include thresholds for "24 hr retrains", "MTBE" (mean time between errors and "MTBR" (mean time between retrains). The point the claimant made is that the threshold values for these parameters in the table at Appendix D determine which ILQ logic will apply and therefore emphasise or de-emphasise the green or red logic paths.
- In this part of the argument the claimant described the green logic as a sub-rule and the red logic as a sub-rule. However these are not the sub-rules the claimant relied on which I accepted in relation to feature (g). Although the defendant did not take the point, the threshold tables relied on need to satisfy the claim based on the sub-rules I have accepted.
- The threshold tables play a part in determining whether the green or red logic paths are taken. This must amount to the emphasising or de-emphasising those paths. In doing so this inevitably has the effect of playing a part in determining whether the sub-rules I have identified previously (Step B of the green logic and the test in the loop) are brought into operation. That seems to me to amount to emphasising or de-emphasising the relevant sub-rules and satisfies feature (h).
- This relates to the overall rule. The claimant submitted the whole NGA logic is the overall rule. The defendant did not agree. It submitted that the requirement for a set of sub-rules and overall rule in the claim meant that the claimed method had an inherently parallel character. For example in a case with three possible profiles and three sub-rules, each sub-rule will be applied to each target profile and evaluate its feasibility by reference to a corresponding threshold. The evaluation could be "GOOD", "NOT GOOD" or "NOT ENOUGH DATA" for each profile as against each threshold. On that basis an overall rule will be needed to reconcile these results in order to decide which profiles will go forward as profiles determined to be "feasible" so that the processing then chooses the feasible profile with the highest priority. The defendant supported this construction by reference to the detailed disclosure in the patent pages 15-17. The sub-rules operate in parallel with each other and the overall rule reconciles the result of these parallel assessments. By contrast the defendant submitted the NGA system was essentially linear. There may be a number of sub-rules but they operate in series and there is no overall rule reconciling the results. The NGA method only ever produces a single selected profile at the end.
- First, the fact that the NGA system only ever produces a single profile output does not show it is different from the claim. The claimed method is supposed only to produce a single profile at the end. Second, although the NGA system only ever produces a single profile as an output, in reaching this result it will, if necessary, evaluate multiple profiles to decide which one to choose. This takes place in the loop processing which uses the profile cap level table. Unless the first profile tested passes the test, the loop method will evaluate at least two profiles. The use of loop logic means that the testing is carried out sequentially but I do not see why that matters. Whether the evaluations are performed in parallel or in series cannot be relevant. There is nothing in the language of the claim which the skilled reader would think mandated only parallel logic. Third, as I have held already on construction, the claimed method does not need to test every possible profile. I have rejected the defendant's interpretation of "each" in feature (k). The method can stop once it has found the highest priority feasible profile.
- Although it is fair to describe the method described in the embodiment described in the patent as having a parallel nature to some extent, for the reasons set out above, I reject the defendant's submission that the claim is limited to parallel logic. The claim requires the overall rule to be one that "utilizes the one or more sub-rules to determine the feasibility or infeasibility for each of the plurality of target line profiles". Thus a rule which selects which sub-rules to apply will be an overall rule. That is what the overall logic of the NGA method does. It satisfies feature (i).
- This requires the sub-rules and overall rule to be based on reported and estimated data. The rules in the NGA system are obviously based on reported data. The processing is all carried out using data reported by the modems.
- The defendant argued that the method of the 790 patent requires the controller to ask, for each target profile: how would the line perform if the system transitioned to that target profile? Whereas the NGA system is different in that it decides how it wants the system to perform and then chooses a profile which will conform to that decision. I will not decide whether these characterisations of the method in the patent and the NGA system are correct because they have lost sight of the language of the claim. Having decided that the rules are based on reported data, the only question I have to decide is whether they are also based on estimated data.
- The estimated data relied on by the claimant is the safe rate used in the loop logic. The claimant argues that an estimation is a rough calculation of something, to be contrasted with a measurement. I agree. It is clear that the safe rate is not a measured rate, it has been calculated. I think that calculation can fairly be called an estimate. The safe rate is a data rate which is estimated based on the actual rate but is higher than it by a simple factor of the highest possible rate. The safe rate represents a rough calculation of a rate which is higher than the current actual rate but not too high so as to cause too many errors.
- The defendant also submitted that it was necessary for the reported data and the estimated data to be profile specific. However there is no such limitation in the claim. I find that feature (j) is satisfied.
- The claimant also relied on the fact that the allocation of the red logic takes account of another kind of estimated data, that is the number of unforced retrains estimated from 24hr line data. I can see that this is estimated data but it is not data on which are based either of the sub-rules relied on (which at least relate to the only profile state transition matrix I find to be used by the NGA system). This point was not focussed on by the defendant but I am not surprised given the extensive permutations of the points on infringement. I doubt this estimated data is relevant to the only infringement case I have so far accepted and I will not rely on it.
- The defendant submitted that the feasibility tests must leave a pool of available profiles and because of that, each and every possible profile transition must be examined, otherwise the pool would not be complete and the top priority transition may not be chosen. This argument is another way of putting the construction point about the word "each" which I have rejected. The claim does not require lower priority profiles to be evaluated if the highest priority feasible profile has been identified.
- The NGA loop processing stops once it has identified the right profile and does not go on to test the lower priority profiles. The defendant argued that since the loop logic never evaluates a profile further down the list once it has stopped, it is not accurate to describe those lower priority profiles as feasible (or deemed to be feasible) but I do not agree. It is no different from the idea expressly contemplated in the patent of deeming the lowest priority profile to be feasible without testing it. When the infringement argument is focussed on the loop, it seems to me that feature (k) is satisfied.
- For the reasons set out above I find that claim 1 is infringed by the NGA system.
- The defendant accepted that if, which it denied, claim 1 was infringed, then claim 10 was also infringed. Nor was there a separate point on claim 13.
- I conclude that the NGA system infringes claims 1, 10 and 13 of the 790 patent.
- The two citations relied on by the defendant are Hendrichs and Gross. I have not described Gross in detail. It is sufficient to say that it is similar to Hendrichs in that it is a splitterless fast retrain system, albeit it is concerned not only with addressing off-hook disturbances but also crosstalk.
- Strikingly Dr Adams' view was that the 790 patent was not obvious over either citation. That was on the basis that he understood claim 1 to be narrow and limited to the use of an actual profile state transition matrix and not covering a method for which a profile state transition matrix could be reverse engineered. It was common ground between Dr Adams and Prof McLaughlin that there is no profile state transition matrix disclosed in Gross. As for Hendrichs, although it does disclose the use of a matrix called a transition probability matrix, both Dr Adams and Prof McLaughlin explained that this was not a profile state transition matrix.
- Dr Adams expressed the view that it would be possible to perform the same reverse engineering exercise on Hendrichs and Gross which the claimant had been relying on for infringement by the NGA system as a whole. Thus the defendant's invalidity case had the nature of a squeeze: if the NGA system infringes, then the prior art is pertinent and renders the claim obvious. This squeeze was focussed on the claimant's infringement case based on the reverse engineering argument (which I have rejected).
- Since I have found that the claim requires an actual profile state transition matrix, this leaves the defendant with insurmountable difficulties advancing its obviousness case over Hendrichs and Gross. Dr Adams' view is that a profile state transition matrix is not obvious over either citation. Prof McLaughlin did not say otherwise. It is impossible to see how the claims could be obvious over Hendrichs or Gross.
- I will mention two further points briefly:
- I conclude that none of the claims of the 790 patent are obvious over either Hendrichs or Gross.
- I find that:
Mr Justice Birss :
| Topic | Para |
| Introduction | 1 |
| The issues | 3 |
| Background | 14 |
| Witnesses | 46 |
| Person skilled in the art | 51 |
| Common general knowledge | 52 |
| The 495 patent | 58 |
| 495 Claim construction | 64 |
| 495 Infringement | 103 |
| 495 Novelty (Hendrichs) | 127 |
| 495 Obviousness | 142 |
| Hendrichs | 145 |
| Rahamim | 152 |
| Kerpez | 161 |
| Conclusion on the 495 patent | 194 |
| The 790 patent | 195 |
| 790 Claim construction | 200 |
| 790 Infringement | 226 |
| 790 Obviousness | 261 |
| Overall conclusion | 267 |
Introduction
The Issues
a) In the 495 patent, claims 1, 6, 8 and 15;
b) In the 790 patent, claims 1, 10 and 13.
1. A method in a controller communicably attached with a Digital Subscriber Line modem pair, the method comprising:
collecting operational data from the DSL modem pair, wherein the operational data comprises current operational data and historical operational data;
analyzing at least a portion of the collected operational data;
generating a margin-related parameter set based on the operational data analyzed; and
instructing the DSL modem pair to operate in accordance with the generated margin-related parameter set.
6. The method of Claim 1, wherein analyzing the operational data comprises determining what margin-related parameter value will cause the DSL modem pair to meet a performance target or target threshold.
8. The method of claim 1, wherein analysing at least the portion of the collected operational data comprises one or more operations selected from the group consisting of:
comparing a current margin related parameter value of the DSL modem pair represented within the current operational data against a corresponding threshold to determine whether a target value is met; and
comparing a historical margin-related parameter value of DSL modem pair represented within the historical operational data against a corresponding threshold to determine whether a target value is met.
15. A controller to monitor a plurality of Digital Subscriber Line Modem pairs, wherein the controller comprises:
a collection module to collect operational data from at least one DSL modem pair, wherein the operational data comprises current operational data and historical operational data;
an analysis module coupled with a connection module to analyse at least a portion of the collected operational data;
an instruction signal generation module coupled with the analysis module,
wherein the instruction signal generation module is to generate a margin-related parameter set based on the analysis and wherein the instruction signal or generation module to instruct one or more of the DSL modem pairs to operate in accordance with the margin-related parameter set.
1. A method in a DSL controller coupled to a DSL line, the method comprising:
operating the DSL line in the current line profile;
collecting operational data relating to operation of the DSL line in the current line profile;
evaluating whether to transition operation of a DSL line from a current line profile to one of a plurality of target line profiles by implementing in hardware or software or a combination thereof:
a profile state transition matrix indicating: a plurality of possible transitions from the current line profile to one of the plurality of target line profiles, and a priority value specifying the priority of transitions for each profile,
a plurality of sub-rules, each to evaluate a feasibility or infeasibility of the plurality of target line profiles within the profile state transition matrix,
a plurality of threshold tables, wherein each of the plurality of threshold tables emphasize or de-emphasize the individual sub-rules, and
overall rule that utilizes the one or more sub-rules to determine the feasibility or infeasibility for each of the plurality of target line profiles, wherein the plurality of sub-rules and the overall rule are based on reported and estimated data from the collected operational data;
determining the feasibility or infeasibility of the current line profile and each of the plurality of target line profiles based on the collected operational data and dependent on the outputs from some or all of the sub-rules and the overall rule; and
selecting a one of the plurality of target line profiles in which to operate the DSL line, wherein the selected line profile comprises whichever of the current line profile and the plurality of target line profiles has a highest priority in the profile state transition matrix, and is not evaluated to be infeasible; and
operating the DSL line in the selected profile.
10. The method claim 1 further comprising adaptively updating at least one of the following:
the profile state transition matrix;
at least one of the plurality of threshold tables; and
a weighting applied to old data.
13. A DSL controller coupled with a DSL system to evaluate whether to instruct a DSL line operating in a current line profile to transition to operation in one of a plurality of target line profiles, a DSL controller comprising:
a state transition control unit implementing in software or hardware or a combination thereof;
a profile sate transition matrix indicating: a plurality of possible transitions from the current line profile to one of the plurality of target line profiles, and a priority value specifying the priority of transitions for each profile,
a plurality of sub-rules, each to evaluate a feasibility or infeasibility of the plurality of target line profiles with the profile state transition matrix,
a plurality of threshold tables, wherein each of the plurality of the threshold tables emphasise or de-emphasise individual sub-rules, and
an overall rule that utilised the one or more sub-rules to determine the feasibility or infeasibility for each of the plurality of target line profiles, wherein the plurality of sub-rules and the overall rule are based on reported and estimated data from the collected operational data;
a data collection module to collect operational data relating to operation of the DSL line in the current profile;
an analysis module to determine the feasibility or infeasibility of the current line profile and each for the plurality of target line profiles based on the collected operational data and dependent on the outputs from some or all of the sub-rules and the overall rule; and
a line profile selector module to select one of the plurality or target line profiles in which to operate a DSL line, wherein the selected line profile comprises whichever of the current line profile and the plurality of the target line profiles has a highest priority in the state transition matrix and is not evaluated to be infeasible.
Background
ADSL
The typical ADSL system
.png)
.png)
The operation of the modems
The witnesses
Person skilled in the art
Common general knowledge
DSM level 0 – Simply static spectrum management with no real DSM.
DSM level 1 – Sometimes known as Dynamic Line Management (DLM). This seeks to control a single line without exchanging information between lines. The data rate and transmit power of the line are controlled to avoid unnecessary crosstalk.
DSM level 2 – This involves coordination of power allocation among multiple lines and takes account of line conditions and service requirements of multiple lines in addition to its own line conditions.
DSM level 3 – This requires the most coordination between lines and involves "vectoring" to coordinate simultaneous adjustments to multiple lines.
The 495 patent
"Systems, devices, methods and techniques that allow users to adjust and adapt transmission power margin(s), spectral densities, and the like dynamically to changing DSL environmental and operational solutions would represent a significant advancement in the field of DSL operation. Moreover, monitoring and evaluation of the power margins etc. used in the DSL environment and operation by an independent entity can assist, guide and (in some cases) control users' activities and equipment, and would represent a significant advancement in the field of DSL operation."
495 Claim construction
i) Controller
ii) Operational data
iii) Collecting current operational data and historical operational data
iv) A portion of operational data
v) Generating
vi) Margin-related parameter set
Controller
Operational data
Collecting current operational data and historical operational data
.png)
A portion of operational data
Generating
Margin-related parameter set
Claim 1 – conclusion on construction
Other claims (6, 8 and 15)
495 Infringement
A controller communicably attached with a Digital Subscriber Line
collecting operational data from the DSL modem pair, wherein the operational data comprises current operational data and historical operational data
analysing at least a portion of the collected operational data
generating a margin-related parameter set based on the operational data analyzed
and
instructing the DSL modem pair to operate in accordance with the generated margin-related parameter set
Claim 1 infringement – conclusion
Infringement – other claims
495 Novelty
"The present invention is directed to a system and method that efficiently, accurately and quickly detects a suitable stored fast retrain profile to permit the resumption of ADSL communications in the presence of changing line conditions in a dual POTS/ADSL communication system."
495 Obviousness
(1) (a) Identify the notional person skilled in the art;
(b) Identify the relevant common general knowledge of that person;
(2) Identify the inventive concept of the claim in question or if that cannot readily be done, construe it;
(3) Identify what, if any, differences exist between the matter cited as forming part of the "state of the art" and the inventive concept of the claim or the claim as construed;
(4) Viewed without any knowledge of the alleged invention as claimed, do those differences constitute steps which would have been obvious to the person skilled in the art or do they require any degree of invention?
Hendrichs vs claim 1
i) Implement it with a central controller (from either common general knowledge as exemplified by the DSL Advances textbook, Kerpez or the DSL Management standard itself); and
ii) Implement it in a system which also stored data for "errored seconds" in the MIB.
Rahamim vs claim 1
"… the CO modem then evaluates the performance of the link by evaluating several parameters, including for example, signal to noise ratio (SNR), automatic gain control (AGC) levels, bit error rates and input power."
i) First it seems to me that the first thing a person reading Rahamim would seek to do would be to contemplate implementing the document as stated and use the parameters mentioned (such as bit error rate). By choosing to use the errored seconds data instead, the skilled person is not taking a step adding to the disclosure but is doing something different. That does not preclude a finding of obviousness but it is a factor to take into account.
ii) Second Dr Adams accepted that one option was to measure Cyclic Redundancy Check (CRC) errors directly without recourse to the MIB data. This would be no worse than using the data in the MIB but would not fall within the claim on the defendant's approach.
iii) Third, there is nothing in Rahamim or in the common general knowledge to give the skilled person the idea of a modem collecting data stored in the MIB.
iv) Fourth, I thought Dr Adams' opinion in cross-examination was not a strong one. I will refer to one example. When it was put to him that at the priority date modems did not access the data in the MIB and Rahamim had not stated where the BER data comes from, his view was that a person implementing Rahamim would realise that "it might be possible" for the ATU-C to access the MIB data.
Kerpez vs claim 1
"Data can be collected about individual telephone lines and cables from loop databases, by automated test equipment, or from DSL modems. This data may then be fed into an advanced DSL management system with a database of DSL loop and noise characteristics, and an analysis engine that tailors DSL deployments to the actual individual line characteristics, to increase provisioned DSL bit-rates while simultaneously increasing reliability and lowering maintenance costs."
"A coordinated implementation of advanced measurement, database storage, analysis, and control of DSL loop and transmission parameters enables a "force multiplier" effect, leveraging existing copper by adding intelligence, control, and communications.
[…]
Adding DSM and intelligent management will enable higher-speed services, with a level of reliability and cost effectiveness that will profitably leverage the copper loop plant to the upper curve for years to come."
(i) Putting Kerpez into practice
i) First, Table 1 of Kerpez illustrates the point that different kinds of noise may be addressed in different ways. Impulse noise may be addressed by increasing interleaving, since as the skilled person would know, it has the character of a short term burst. Crosstalk may be addressed by DSM (i.e. by addressing power and margin). The idea that different remedies may suit different problems would make sense to the skilled reader.
ii) Second, Dr Adams' view was that there were fairly simple techniques to get to a choice between what might be the probable cause of the problem. He said for example that all you need to do is to know that impulse noise is the cause of the problem and therefore increase interleaver depth. This makes sense since the skilled person would know that increasing interleaving is designed to mitigate burst errors. Similarly Prof McLaughlin accepted that you may surmise a burst of errors was impulse noise.
iii) Third, although it is the case that Dr Adams' own PhD thesis was in this area and was undertaken years after the priority date, that did not contradict his opinion because it was concerned with more detailed analysis, to identify down to the level of what type of impulse noise was present and what the possible sources of that impulse noise might be.
iv) Fourth, Prof McLaughlin in paragraph 149 said that:
"Identification of individual impairments [i.e. the causes of the errors] is problematic and likely to have required significant computational capacity, and so would have been better dealt with by carrying out the analysis offline as suggested by Kerpez figure 2".
The Professor is not saying identification could not be done, only that significant computational capacity would be required. Moreover the idea that the processing would be "offline" needs to be treated carefully. The claim is not limited to a system operating online in real time. The BT DLM systems alleged to infringe operate overnight. By the time the RAMBo selects a new profile, the operational data being considered is many hours out of date. It does not represent the conditions actually on the line at the time the processing occurs. But that does not take the method outside claim 1.
v) Fifth, although Kerpez refers to G.selt modems as modems which in the future will produce useful information, Dr Adams did not accept that Kerpez's proposal needed G.selt modems to be available. He referred to the discussion in Kerpez of getting information from ADSL and ADSL2 modems. His view was that the skilled person would know there was information available in the MIBs.
vi) Sixth and importantly, I think Dr Adams was better placed than Prof McLaughlin to describe the approach of a skilled person working in a company at the relevant time. Although Prof McLaughlin clearly has significant industrial experience, he has less direct industrial experience than Dr Adams. Dr Adams' view was that distinguishing between crosstalk and impulse noise was within the capabilities of a skilled person working in industry. He recognised that there was no off the shelf system available to do it but his view was that such a system was not difficult to produce. The timescales would not be short, but they were not undue.
(ii) Collecting historical and current data
(iii) "generating" a parameter set
Claim 1 on the claimant's wide construction
Kerpez vs the other claims
Conclusion on 495
The 790 patent
"Systems, methods and techniques that permit implementation of a wide variety of line profiles and transitions between such profiles automatically and with ease in communication systems such as DSL systems would represent a significant advancement in the art. In particular, prioritization and implementation of transition options in the communication system would represent a considerable advancement in the field of DSL service rates and associated ranges."
.png)
790 Claim construction
i) Line profile
ii) DSL controller
iii) Profile state transition matrix
iv) Plurality of target profiles
v) Possible transitions, priority and feasibility and the sub-rules, threshold tables and overall rule.
vi) Reported and estimated data
vii) Determining the feasibility of … each of the plurality of target line profiles
Line profile
DSL controller
Profile state transition matrix
"A method in a DSL controller…comprising…evaluating whether to transition … from a current line profile to one of a plurality of target line profiles by implementing in software or hardware or a combination thereof: a profile state transition matrix indicating … [possible transitions and a priority value for those transitions for each profile]".
Plurality of target line profiles
Possible transitions, priority and feasibility, and the sub-rules, threshold tables and overall rule
reported and estimated data
Determining the feasibility of … each of the plurality of target line profiles
790 Infringement
(a) A method in a DSL controller coupled to a DSL line
Clearly the way the RAMBo works satisfies this feature.
(b) operating the DSL line in the current line profile
The relevant logic starts with the DSL line in a current profile. Any current profile will do.
(c) collecting operational data relating to operation of the DSL line in the current line profile
There is no dispute about this.
(d) evaluating whether to transition operation of a DSL line from a current line profile to one of a plurality of target line profiles
The evaluation process carried out in the RAMBo is represented by the logic I have referred to above. It decides whether to move from the current line profile to one of a number of other target line profiles.
(e) by implementing in hardware or software or a combination thereof:
There is no dispute about this. The method is implemented in software running on the RAMBo.
(f) a profile state transition matrix indicating: […]
This raises a major dispute between the parties which I will address below in the Profile State Transition Matrix section.
(g) a plurality of sub-rules, each to evaluate a feasibility or infeasibility of the plurality of target line profiles within the profile state transition matrix,
(h) a plurality of threshold tables, wherein each of the plurality of threshold tables emphasize or de-emphasize the individual sub-rules,
(i) overall rule that utilizes the one or more sub-rules to determine the feasibility or infeasibility for each of the plurality of target line profiles,
(j) wherein the plurality of sub-rules and the overall rule are based on reported and estimated data from the collected operational data;
(k) determining the feasibility or infeasibility of the current line profile and each of the plurality of target line profiles based on the collected operational data and dependent on the outputs from some or all of the sub-rules and the overall rule; and
These parts of the claim are addressed in the Rules and Thresholds section below.
(l) selecting a one of the plurality of target line profiles in which to operate the DSL line, wherein the selected line profile comprises whichever of the current line profile and the plurality of target line profiles has a highest priority in the profile state transition matrix, and is not evaluated to be infeasible; and
This is disputed and can be dealt with in the Profile State Transition Matrix section.
(m) operating the DSL line in the selected profile
There is no dispute about this.
Profile State Transition Matrix
The case based on reverse engineering a profile state transition matrix
Reverse engineering
Priority between interleaving and cap level
The case based on the profile cap level table
Rules and Thresholds (features (g), (h), (i), (j) and (k))
feature (g)
Feature (h)
Feature (i)
Feature (j)
Feature (k)
Infringement of the 790 patent - conclusion
790 Obviousness
i) Both Hendrichs and Gross are systems in which intelligence resides in the modems and sets transmission parameters. The defendant argued it was obvious over either reference to produce a system with a central controller remote from the modems which set configuration parameters. I very much doubt that. Neither document is concerned with such an arrangement at all.
ii) The claimant pointed out that neither Hendrichs nor Gross disclosed the use of estimated data. That is true but using an estimated value of the kind I have found satisfies the claim (the safe rate used in the NGA system) does not strike me as inventive.
Overall conclusion
i) None of the defendant's systems (20CN, 21CN and NGA) infringe the 495 patent. The patent is valid. However if the claims were wide enough to cover the defendant's systems, claims 1, 8 and 15 would be obvious over Kerpez while claim 6 would not be. Thus on the wide construction of the 495 patent, only the NGA system would infringe a valid claim.
ii) The 790 patent is valid and infringed by the defendant's NGA system.