UNITED STATES DISTRICT COURT FOR THE EASTERN DISTRICT OF TEXAS TYLER DIVISION MICROSOFT CORP., ET AL., v. COMMONWEALTH SCIENTIFIC AND INDUSTRIAL RESEARCH ORGANISATION COMMONWEALTH SCIENTIFIC AND INDUSTRIAL RESEARCH ORGANISATION, v. TOSHIBA AMERICA INFORMATION SYSTEMS, INC., ET AL., INTEL CORPORATION, ET AL., v. COMMONWEALTH SCIENTIFIC AND INDUSTRIAL RESEARCH ORGANISATION Case No. 6:06-CV-549 (LED Case No. 6:06-CV-550 (LED Case No. 6:06-CV-551 (LED DECLARATION OF HARRY V. BIMS, PH.D. IN SUPPORT OF MOTION FOR RECONSIDERATION -1-
I, Harry V. Bims, hereby declare as follows: 1. I have personal knowledge of the matters set forth below and, if called to testify, I could and would competently testify thereto. 2. I have been retained as an expert by all parties adverse to Commonwealth Scientific and Industrial Research Organisation ( CSIRO in the above-captioned consolidated cases. In this declaration, I set forth my opinions regarding the disclosed structure corresponding to the means to apply a data reliability enhancement in U.S. Patent No. 5,487,069 (the 069 Patent. 3. I have an extensive educational background and work experience in the field of modern telecommunications, including wireless communications. More specifically, I have studied telecommunications and systems engineering since approximately 1981, with a particular emphasis in Trellis Coded Modulation and its implementation in communication systems. Moreover, I have over 17 years of industry experience in telecommunications, including wireless communications. During this period, I designed and implemented various products that involve technologies relating to the subject matter of the 069 Patent. 4. I previously have submitted several reports and affidavits in connection with my work in the above-captioned cases, including Rebuttal Expert Reports Regarding Noninfringement and a Declaration In Support of Defendants Opposition to CSIRO s Motion for Summary Judgment of Infringement. Also, on June 5, 2008, I submitted a Declaration in Support of Defendants Response Brief Regarding Claim Construction. The opinions set forth herein are consistent with, or supported by, my previous opinions in the cases. 5. In addition to the materials that I reviewed in connection with my expert reports and declarations, I have reviewed the Court s April 3, 2009 Memorandum Opinion and Order -2-
Regarding Supplemental Claim Construction. Based on my knowledge and experience in the relevant field, and my review of the relevant materials, it is my considered opinion that no one of ordinary skill in the art would understand the rate ½ TCM encoder, described in the 069 Patent in Block 42 of Figure 7 and referenced at column 6:32-46, to include only a rate ½ finite-state encoder. 6. Rather, the rate ½ TCM encoder described and claimed in the 069 Patent necessarily integrates a finite-state encoder and a mapper together to create a single, indivisible entity. Further, as CSIRO s own expert, Dr. Monsen, agrees 1, the integrated finite-state encoder and mapper unit in the rate ½ TCM encoder operate interdependently and in unison to perform the function of enhancing data reliability. My opinions in this regard are supported by testimony from CSIRO s own expert and the inventors in this case. 2 7. On page 6 of the Order, the Court states that TCM schemes usually or typically contain both coding functions and mapping functions, and also states that the modulation scheme [in TCM] is an analog function. This finding is inconsistent with the universally accepted understanding of TCM encoders. 8. First, all TCM encoders, including the rate ½ TCM encoder depicted in block 42 of Figure 7 and referenced at column 6:32-46 in the 069 Patent, must integrate a finite-state 1 Monsen Dep. at 311:20-25, Sept. 24, 2008. 2008. 2 Monsen Dep. at 311:20-25, Sept. 24, 2008; O Sullivan Dep. at 282:11-286:12, Feb. 8, -3-
encoder and a mapper. That is a fundamental principle of TCM, which the 069 patent itself 3 and CSIRO s own experts 4, recognize. 9. Indeed, without the mapping component, there is no TCM encoder whether rate ½ or otherwise. As stated by the inventor of TCM, Dr. Ungerboeck, TCM schemes employ redundant nonbinary modulation in combination with a finite-state encoder which governs the selection of modulation signals to generate coded signal sequences. In fact, TCM encoders were developed based on Dr. Ungerboeck s realization that, contrary to conventional wisdom at the time, the finite-state encoding component should not be designed separately from the mapping component. 5 The unique and inseparable integration of encoder and mapper was indeed the essence of the breakthrough concept of TCM. Accordingly, those skilled in the art at the time of the invention would have recognized that an encoder and mapper were both necessary requirements of any TCM scheme, including the rate ½ structure depicted in the patent. 10. Further, the coding and mapping that take place within TCM encoders, including the rate ½ TCM encoder depicted in the 069 Patent, are not separate processes. Rather, they are two interdependent and intertwined operations, both digital, that jointly provide Forward Error Correction and thereby enhance data reliability. Both operations are part of the inseparable design of TCM, and both are critical to the manner in which TCM enhances data reliability by 3 069 at 9:65-66 ( combined coding and modulation schemes such as trellis-coded modulation (TCM. 4 The universally understood structure for trellis coding involves a convolutional encoder followed by a mapper. Andrews Decl. ISO CSIRO s Brief Re Claim Construction, 9; See also, Monsen Dep. at 311:20-25, Sept. 24, 2008. 5 Dr. O Sullivan, one of the named inventors of the 069 Patent, echoed this well-known principle when discussing the TCM encoder described in the patent. O Sullivan Dep. at 283:14-21, Feb. 8, 2008. -4-
maximizing the Euclidean distance of its output symbol sequences. In fact, an improper modification to the interface between the finite-state encoder and mapper such as, for example, inserting a bit-wise interleaver between the encoder and mapper would void the TCM technique and its attendant performance advantages. One of ordinary skill in the art would not consider such a coder-interleaver-mapper scheme to be a TCM design. 11. Based on the patent specification and the universally accepted understanding of TCM encoders, one of ordinary skill in the art would understand the rate ½ TCM encoder depicted in the patent to work as follows: the finite-state encoder receives one input bit and generates one coded bit, and one un-coded bit, where the un-coded bit is merely a copy of the input bit. The coded bit is used to select a coset of the QPSK signal constellation containing two signal points. In other words, the coded bit selects a subset of signal points that are separated by a larger Euclidean distance than the signal points in the original QPSK constellation. The uncoded bit from the finite-state encoder is used to select one of the two signal points in the coset that was selected based on the coded bit. This technique is known as mapping by setpartitioning, a technique that Dr. Ungerboeck himself describes as of central significance for all TCM schemes. 6 12. Using the mapping by set-partitioning technique, the mapper generates output symbols (referred to as output di-bits in the patent because each output symbol is represented by a pair of bits 7 selected from a QPSK symbol constellation composed of four candidate output 6 Ungerboeck, G. Trellis-Coded Modulation with Redundant Signal Sets Part II: State of the Art. IEEE Communications Magazine, Feb. 1987 7 By definition the di-bit symbols that are output from the rate ½ TCM encoder are still digital both the coding and mapping schemes in TCM operate in the digital domain. Neither is an analog operation. Note that all operations in Figure 7 are performed in the digital domain up to the digital-to-analog conversion performed by D-to-A blocks 49 in the figure. -5-
symbols. In other words, for each iteration of the encoding process, the mapper used in the rate ½ TCM encoder selects a partition or coset comprised of two symbols based upon the coded bit obtained from the finite state encoder. The rate ½ TCM encoder then selects one of these two symbols based upon the uncoded data bit that passed through the finite state encoder unaltered. This means that the expanded QPSK symbol constellation inside the mapper of the rate ½ TCM encoder contains redundant symbols. The mapper takes advantage of the symbol redundancy in the expanded QPSK constellation through set partitioning to enhance data reliability in the course of selecting an output symbol. 13. It is important to note that the input to the selection process of the mapper in the rate ½ TCM encoder of the 069 Patent includes both the input bit to the rate ½ TCM encoder, and the coded bit generated by the integrated finite state encoder that is joined to the mapper. This correspondence is strictly enforced by the interdependent design that integrates a finite state encoder with a mapper in the rate ½ TCM encoder. The interconnection between the integrated finite state encoder and the integrated mapper connects the bit-level coding generated by the finite state encoder to the symbol-level coding used by the selection process in the mapper, such that the integrated finite state encoder and mapper operate in unison to perform the function of applying data reliability enhancement. 14. On page 6 of the Order, the Court states the patent s context and specification provide an abundance of evidence that the corresponding structure for the purposes of determining 112, 6 equivalence is a rate ½ finite-state encoder regardless of any mapping function. 15. However, it is my opinion that requiring a mapper to be part of the rate ½ TCM encoder depicted in Figure 7 (as all TCM encoders require, by necessity is consistent with, and -6-
indeed supported by, the entirety of the patent specification, including the description of the other components identified in Figure 7. To explain this point, it is helpful to understand the interrelationship between block 42 and the blocks that follow it in Figure 7 (shown below. 16. First, based on the patent specification, one of ordinary skill in the art would understand that the block 43 di-bit interleaver immediately following the rate ½ TCM encoder in Figure 7 performs its interleaving operation on a symbol basis, rather than a bit basis. Indeed, in column 10 of the 069 patent specification, the di-bit interleaver is described as performing an interleaving operation on carrier numbers (1-12 that have been modulated by successive encoder output di-bits. 8 The use of the term modulated in this context indicates that data that the interleaver is operating on are symbols, not individual bits. 17. The inventors of the 069 patent also confirmed that the di-bit interleaver in block 43 is a symbol interleaver. In particular, beyond stating that the di-bit interleaver interleaves on a symbol-by-symbol basis, one inventor, Dr. Percival confirmed that the inventors chose the di-bit 8 069 at 10:26-31. -7-
interleaver because you would get better performance interleaving symbols rather than individual bits 9 This was echoed by another inventor, Dr. O Sullivan, during his deposition, 10 and is a basic and undisputed premise accepted by both sides experts. Nothing in the 069 patent suggests, and no one of ordinary skill in the art would understand, that the di-bit interleaver in block 43 of Figure 7 would interleave anything other than symbols. 18. Importantly, because the interleaver in block 43 is operating on symbols instead of bits, the output of the rate ½ TCM encoder in block 42 (from which the interleaver receives its input necessarily must be symbols instead of independent bits. To produce such symbols at its output, the rate ½ TCM encoder must incorporate a mapper. 19. Next, after the interleaving function is performed by block 43, the output sequence of interleaved di-bit symbols is input into block 44 on each cycle of the system s bit clock. This block is described in Figure 7 as a Differential (by Frame QPSK Encoder. The corresponding text in the specification that describes this figure goes on to further describe the function of this element as a QPSK Encoder 44 which carries out differential encoding on a frame-by-frame basis. 11 Differential encoding is a term used in the art to describe a function that computes the difference between its previous output and its current input. In particular, as Dr. Percival has explained, the function of the Differential (by Frame QPSK Encoder in block 44 was to calculate the difference in phase from one symbol to the next. 12 The function is performed by storing in internal memory the previous symbol, and subtracting its value from the 9 Percival Dep. at 350:10-351:4, Feb. 6, 2008. 10 O Sullivan Dep. at 43:4-44:12, Feb. 7 2008. 11 069 at 6:40-41. 12 O Sullivan Dep. at 294:7-20, Feb. 8, 2008-8-
current symbol. One possible realization for how differential encoding can be implemented on each symbol is as follows: 20. The 069 patent specification describes a frame as containing 12 differentially encoded di-bit symbols, one symbol for each of the 12 carriers in the transmission. The output of QPSK Encoder 44 and a synchronizing header generator 45 are combined in a frame assembly and zero pad insertion block 46 so that the frames are assembled and four zero pads inserted so that six carriers are generated to each side of, but not coincident with, the centre frequency. The assembled frames are then passed through an Inverse Fast Fourier Transform device 47 which uses a 16 point complex IFFT. 069 Patent, Col. 6:42-46. Since Block 44 performs its operation of differential encoding on a frame-by-frame basis, it must store the previous di-bit symbol for each of the 12 carriers in its internal memory. Thus, a possible realization for block 44 is as follows: -9-
21. Based on the patent s description of block 44 and Dr. Percival s testimony, the Differential QPSK Encoder depicted in block 44 necessarily does not contain a mapper or otherwise perform a mapping operation. One of ordinary skill in the art would understand that a differential encoder performs a function distinct from mapping, and nothing in the patent suggests that the differential encoder in block 44 contains a mapper. Indeed, this point is confirmed by Dr. Percival s testimony as he stated that the function that is performed by block 44 is to calculate the difference in phase from one symbol to the next. This means that the data on which the Differential QPSK encoder is operating already has been mapped into symbol form. The only structure preceding block 44 that possibly could have performed the mapping is the rate ½ TCM encoder in block 42. 22. Moreover, it is my opinion that if a mapper is not present in block 42 of Figure 7, the claimed invention would be rendered inoperable. This is because every wireless transmission system requires a mapping operation, and, besides the rate ½ TCM encoder in block 42, no other -10-
component in the transmission system depicted in Figure 7, including the differential QPSK encoder in block 44, performs a mapping operation. 23. Even if one were to incorrectly characterize block 44 to include a mapping function, then the data reliability enhancement means would have to consist of a plurality of rate ½ TCM encoders one for each sub-channel used for transmission to allow the system to operate properly. For example, the preferred QPSK embodiment described in the patent specification would have to include 12 rate ½ TCM encoders under such a scenario, since the mapper portion of each rate ½ TCM encoder would appear after interleaving of its output dibits to one of the 12 sub-channels. This does not comport with the depiction of a single rate ½ TCM encoder in Figure 7, or any other part of the patent s description. 24. The remainder of the depiction of Figure 7 and the remainder of the patent specification both support my opinion that the rate ½ TCM encoder necessarily includes an integrated finite-state encoder and a mapper, which operate in unison to provide data reliability enhancement. For example, the patent itself states that combined coding and modulation schemes, such as trellis-coded modulation (TCM, [] give improved error correction capability. 13 This statement is complimentary, not superfluous, to the description of the rate ½ TCM encoder. Indeed, one of ordinary skill in the art would have understood the statement to relate directly to the preferred embodiment depicted in Figure 7, and provide a contrast to the description in the immediately preceding three paragraphs ( 069 at 9:36-65 an alternative embodiment which can use either Reed-Solomon or convolutional coding. 25. In sum, it is my opinion that the rate ½ TCM encoder in Block 42 of Figure 7 and described in column 6 of the patent specification requires an integrated mapper that is combined 13 069 at 9:65-10:02. -11-
with an integrated finite-state encoder in order to perform the claimed function of applying data reliability enhancement. This configuration is required by every TCM encoder by definition and by necessity and design; and neither the patent specification nor anything that CSIRO s own experts said contradicts this conclusion. 26. Finally, I have not yet had the opportunity to consider the extent of the impact of the Court s April 3, 2009 Memorandum Opinion and Order as it relates to the infringement and/or invalidity issues. Nor have I seen any additional opinions proffered by CSIRO s experts on these issues, and would request the opportunity to consider and respond to any such opinions if they are forthcoming. I therefore reserve the right to supplement my previously-submitted opinions in this case. I declare under penalty of perjury, under the laws of the United States of America, that the foregoing is true and correct. Executed April 5, 2009, at Menlo Park, California. By: Harry Bims -12-