`Filed: November 7, 2017
`
`
`
`
`
`
`
`
`
`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`_____________________________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`_____________________________
`
`
`APPLE INC.,
`Petitioner,
`
`v.
`
`CALIFORNIA INSTITUTE OF TECHNOLOGY,
`Patent Owner.
`_____________________________
`
`Case IPR2017-00219
`Patent No. 7,116,710
`
`
`
`_____________________________
`
`
`
`PATENT OWNER’S RESPONSE
`PURSUANT TO 37 C.F.R. § 42.120
`
`
`
`
`
`
`
`TABLE OF CONTENTS
`
`Dr. Davis’s evasiveness during his deposition undermines his
`
`Divsalar in view of Luby does not disclose or lead to irregular
`
`STATEMENT OF PRECISE RELIEF REQUESTED .................................. 1
`I.
`INTRODUCTION AND OVERVIEW OF ARGUMENT ............................ 1
`II.
`III. OVERVIEW OF THE ART .......................................................................... 5
`A. Divsalar .............................................................................................. 8
`B.
`Luby ................................................................................................... 9
`C.
`Frey .................................................................................................. 10
`IV. DR. DAVIS’S TESTIMONY SHOULD BE GIVEN LITTLE
`WEIGHT .................................................................................................... 11
`A. Dr. Davis’s testimony includes basic errors demonstrating a
`lack of credibility .............................................................................. 12
`B.
`Dr. Davis’s testimony is not independent .......................................... 14
`C.
`credibility.......................................................................................... 15
`V. GROUND 1: DIVSALAR IN VIEW OF LUBY DOES NOT
`RENDER CLAIMS 1-8 AND 11-14 OBVIOUS ........................................ 17
`A.
`Legal Principles ................................................................................ 18
`B.
`repetition of information bits............................................................. 19
`1. The cited references do not disclose irregular repetition of
`information bits ............................................................................. 19
`2. The petition relies on improper hindsight to combine Luby
`and Divsalar .................................................................................. 26
`required by claim 1 ........................................................................... 29
`and Luby as proposed ....................................................................... 30
`1. The Board should not reconsider arguments it has already
`rejected ......................................................................................... 30
`2. Luby does not teach that irregular repetition of information
`bits results in any improvement ..................................................... 30
`
`The petition fails to establish a motivation to combine Divsalar
`
`C.
`D.
`
`The cited references do not disclose the “partitioning” step as
`
`
`
`-i-
`
`
`
`
`The petition’s proposed modifications to Divsalar and
`
`The petition does not and cannot show that either its proposed
`modification or the Divsalar-Luby combination in general
`
`3. Luby does not provide a POSA any motivation to modify
`Divsalar’s repeater ........................................................................ 33
`4. Luby does not describe irregularity in the context of turbo
`codes ............................................................................................. 35
`5. The petition’s citation to Frey undermines its motivation to
`combine argument ......................................................................... 35
`6. The petition’s proposed modification was not a “simple” or
`“routine” change ........................................................................... 37
`E.
`Khandekar’s thesis are not supported by any teaching in Luby ......... 39
`F.
`would have a reasonable expectation of success ............................... 44
`VI. GROUND 2: THE COMBINATION OF DIVSALAR, LUBY, AND
`33 OBVIOUS ............................................................................................. 48
`A.
`There is no motivation to combine Divsalar, Luby and Luby97 ........ 48
`B.
`challenged claims .............................................................................. 49
`VII. OBJECTIVE INDICIA OF NON-OBVIOUSNESS.................................... 50
`A. Nexus between the Objective Evidence and the Claims .................... 51
`B.
`Long-felt need and failure of others .................................................. 54
`C.
`Industry Praise .................................................................................. 56
`D. Unexpected Results........................................................................... 58
`E.
`Commercial Success ......................................................................... 60
`VIII. CONCLUSION .......................................................................................... 62
`IX. APPENDIX ................................................................................................ 64
`
`LUBY97 DOES NOT RENDER CLAIMS 15–17, 19–22, AND 24–
`
`The petition fails to explain how the combination discloses the
`
`
`
`-ii-
`
`
`
`
`I.
`
`STATEMENT OF PRECISE RELIEF REQUESTED
`
`Apple, Inc. (“Petitioner”) filed a petition for inter partes review of claims 1-
`
`8, 10-17, and 19-33 of U.S. Patent No. 7,116,710 (the “’710 patent”, EX1201).
`
`The Board issued its decision instituting trial (“Decision,” Paper 17) on two of the
`
`four petitioned grounds and with respect to all but two of the challenged claims,
`
`claims 10 and 23. The patent owner (“PO” or “Caltech”) hereby requests that the
`
`Board now issue a final written decision rejecting all grounds of challenge still
`
`remaining, and confirming that claims 1-8, 11-17, 19-22, and 24-33 are not
`
`unpatentable.
`
`II.
`
`INTRODUCTION AND OVERVIEW OF ARGUMENT
`
`The ’710 patent is one of four Caltech patents that resulted from research
`
`performed by the inventors, Drs. Jin, Khandekar, and McEliece, in 1999-2000.
`
`The patents claim inventions directed to a revolutionary class of error-correction
`
`codes, dubbed “irregular repeat and accumulate codes,” or “IRA codes,” which
`
`rivaled and surpassed the performance of the best known codes at that time. No
`
`other code known at the time could boast linear encoding, linear decoding, and
`
`performance near the theoretical Shannon limit.
`
`The IRA codes described in the ’710 patent were the culmination of more
`
`than a year of research and analysis by the inventors into different code structures.
`
`As even Petitioner’s expert acknowledges, the field of error correction coding is a
`
`
`
`-1-
`
`
`
`
`complex and highly unpredictable one. Design of new error correction codes
`
`typically requires extensive experimentation by experts in the field in order to
`
`identify a viable code structure, create useable encoders and decoders, and
`
`demonstrate the capabilities of the code’s performance. New code structures
`
`require rigorous simulation and analysis to determine whether they can be reliably
`
`encoded and decoded. Features that may improve performance in one code may
`
`have detrimental effects in others, and results were unpredictible.
`
`In arguing that the instituted claims are unpatentable, Petitioner relies on a
`
`combination of two prior art references: the Divsalar reference, which describes a
`
`method of encoding using repeat accumulate (RA) codes, and the Luby reference,
`
`which describes a set of codewords that are based on application of irregular
`
`bipartite graphs to Gallager’s LDPC codes. Neither reference discloses the
`
`limitation of irregularly repeating information bits, which is required by all of the
`
`’710 claims, and a person of ordinary skill in the art would not have been
`
`motivated by Luby to incorporate irregular repetition into Divsalar.
`
`The petition fails to describe how or why a person of ordinary skill in the art
`
`would have been motivated by Luby, which describes graphs in which the degree
`
`of the codeword is irregular, to make the repetition of the information bits in the
`
`encoding described in Divsalar irregular. Luby does not even describe an encoding
`
`process, and thus does not describe information bits. Petitioner does not point to
`
`
`
`-2-
`
`
`
`
`any teaching anywhere in the art that would suggest making information bits in the
`
`Divsalar code irregular and wholly lacks discussion of whether there would be any
`
`reasonable expectation of success—a critical requirement of an obviousness
`
`inquiry under Graham v. John Deere. Rather than apply the description in Luby of
`
`irregular codewords to Divsalar’s RA codes, the petition instead proposes a
`
`modification (which is not described by Luby) of a Tanner graph illustrated in a
`
`non-prior art Ph.D. thesis of Dr. Khandakar—an argument the Board already
`
`rejected in its institution decision. To the contrary, objective indicia of
`
`nonobviousness confirm that IRA encoding and decoding methods and systems
`
`were groundbreaking developments that overcame long recognized problems in
`
`previously known error correction codes, and were widely hailed as a
`
`revolutionary invention.
`
`PO submits with this Response declarations from Dr. Dariush Divsalar
`
`(EX2031) and Dr. Michael Mitzenmacher (EX2004). Drs. Divsalar and
`
`Mitzenmacher are preeminent experts in the fields of information theory and error
`
`correction coding. Dr. Mitzenmacher is a Professor of Computer Science at
`
`Harvard University who was a co-author on the Luby reference, and Dr. Divsalar is
`
`a principal scientist at Caltech’s Jet Propulsion Laboratory and was a co-author on
`
`the Divslar reference.
`
`
`
`-3-
`
`
`
`
`A petitioner in an inter partes review bears the unshifting burden of proving
`
`unpatentability, and the petition here should be rejected as failing to demonstrate
`
`that a person of ordinary skill in the art would have been motivated to combine the
`
`descriptions of the Luby and Divsalar reference, that the proposed combination
`
`would have a reasonable expectation of being successful, or would result in IRA
`
`codes as claimed. Dr. Divsalar and Dr. Mitzenmacher further explain that a POSA
`
`would not have considered combining their respective works as proposed, and
`
`would certainly not have expected the success let alone the superior performace of
`
`IRA codes.
`
`As Dr. Mitzenmacher explains, Luby did not show that irregularity in
`
`general improves an error correction code; rather, Luby concluded that some
`
`irregular bipartite graphs are better than regular bipartite graphs when used to
`
`describe the parity check matrix of a Gallager code. But nothing in Luby suggests
`
`how irregularity might be applied to other code structures, particularly code
`
`structures like Berrou’s turbo codes or Divsalar’s RA code that are not derived
`
`from bipartite graphs or parity check matrices, or to the repetition of information
`
`bits in an encoding process.
`
`Even if a person of ordinary skill attempted to combine the descriptions in
`
`Divsalar and Luby—notwithstanding the lack of any motivation to do so—that
`
`combination would not result in IRA codes as claimed. As Dr. Mitzenmacher
`
`
`
`-4-
`
`
`
`
`explains, the irregular code structure described in the Luby reference is not the
`
`same as the irregular repetition performed in IRA codes, and any attempt to
`
`combine Luby’s irregularity with Divsalar’s code would not result in IRA codes
`
`claimed. The Divsalar code is already “irregular” under Luby’s definition of
`
`irregularity, and modifying Divsalar’s repeater to be irregular would not add
`
`further “irregularity” to the bipartite graph that Luby teaches.
`
`For these reasons, all of the remaining grounds of challenge must be denied.
`
`III. OVERVIEW OF THE ART
`
`The field of error correction coding has historically been characterized by
`
`significant experimentation and unpredictable results.1 Since it is mathematically
`
`impossible to prove the performance of most codes, researchers typically engaged
`
`in extensive trial-and-error and experimentation with various code structures to
`
`determine whether new codes were capable of improved performance. Even when
`
`well-performing codes are identified, the reasons for the improved performance
`
`were often not understood. EX2004 ¶ 46.
`
`
`
`
`
`
`1 Dr. Mitzenmacher provides an overview of certain error correction coding
`
`principles and terminology at ¶¶ 29-45 of his declaration (EX2004).
`
`
`
`-5-
`
`
`
`
`As Petitioner’s expert conceded during cross-examination:
`
`What you would really like to be able to do is a formal
`
`mathematical analysis of the strength of the codes that
`
`you are working with, but that’s often really hard. So
`
`often what the engineers in particular would do is … take
`
`a variety of different [codes], run simulations and … then
`
`I will get a general sense of what the [mathematical]
`
`analysis would have shown me. … [I]t might even be
`
`impossible to do the mathematical analysis.
`
`EX2033 256:21-257:12 (emphasis added). Caltech’s expert, Dr. Mitzenmacher,
`
`likewise explains that discoveries had to be made via extensive experimentation.
`
`EX2004 ¶ 47. As a result, it was rarely the case that a researcher could reasonably
`
`predict that a particular modification would result in an improvement in the
`
`performance of a code. Id.
`
`The unpredictability of the field is demonstrated by the history of two
`
`particular classes of codes that were considered to be the best known codes in the
`
`mid- to late-1990s: turbo codes and Gallager codes.2 EX2004 ¶¶ 47-56. Turbo
`
`codes were created by Claude Berrou in 1993 by concatenating in parallel two
`
`previously known convolutional codes. Id., ¶¶ 50-51. Berrou’s turbo codes were
`
`
`2 The cited references themselves also corroborate the unpredictability in the
`
`field at the time.
`
`
`
`-6-
`
`
`
`
`met with significant skepticism, and Berrou himself could not explain why turbo
`
`codes perfomed as well as they did; many who attended his presentation believed
`
`he had made an error in his initial experiments. After his results were
`
`independently confirmed, turbo codes became widely regarded as one of the best
`
`known code structures of that time. Id., ¶ 51.
`
`Gallager codes, often referred to as LDPC codes, were first proposed in a
`
`paper by Dr. Robert Gallager in 1963.3 EX2004 ¶ 52. In 1963, however, Dr.
`
`Gallager did not have the tools to show the performance of his codes and they
`
`remained largely forgotten for decades. Id., ¶ 53. It was not until the late 1990s
`
`that researchers in the field, including the authors of the Luby reference,
`
`rediscovered Gallager’s LDPC codes and demonstrated that they can be as
`
`effective as Berrou’s turbo codes. Id.
`
`New developments in the field of error correction coding have frequently
`
`been the result of surprising breakthoughts during experimentation with new
`
`
`3 Modern coding theory sometimes uses the term “LDPC code” more broadly to
`
`refer to codes that can be represented using a parity-check matrix with low density.
`
`In 2000, however, the term denoted Gallager’s original codes, which a narrowly
`
`defined structured.
`
`
`
`-7-
`
`
`
`
`classes of codes. The IRA codes claimed in the ’710 patent were no exception to
`
`this phenomenon.
`
`A. Divsalar (EX1203)
`
`The Divsalar reference describes the work of Dr. Dariush Divsalar, along
`
`with two of the inventors of the ’710 patent (Drs. McEliece and Jin), in developing
`
`a repeat accumulate (RA) code. EX2031 ¶ 16-32.
`
`RA codes as taught in Divsalar are nonsystematic codes, meaning that only
`
`the encoded codeword bits (parity bits) are transmitted. RA codes always perform
`
`regular repetition of information bits; and every repeated bit in an RA code is
`
`separately accumulated to generate a new parity bit.4 At a rate of 1/q (e.g., a repeat
`
`degree of 4 results in a coding rate of ¼), RA codes are impractically slow.
`
`Indeed, Dr. Divslar explains that the codes were never intended to be competitive
`
`error correction codes, nor would they have been mistaken as such—they were
`
`designed as a research tool for learning about certain characteristics of turbo codes
`
`and similar codes. EX2031 ¶ 27, 28, 32; EX2004 ¶ 58.
`
`
`4 In contrast, subsets of information bits in IRA codes are combined using an
`
`XOR operation or modulo-2 addition, and the sums are then accumulated. As a
`
`result of these differences, IRA codes exhibit significantly better performance than
`
`RA codes.
`
`
`
`-8-
`
`
`
`
`Additionally, Divsalar did not analyze RA codes using parity-check matrices
`
`or Tanner graphs, and at the time of the invention of the ’710 patent, such an
`
`analysis would not have been common: turbo codes and LDPC codes were viewed
`
`as two distinct types of codes using different approaches to code design and
`
`analysis. EX2031 ¶26; EX2004 ¶¶ 54-56.
`
`B.
`
`Luby (EX1204)
`
`The Luby reference describes certain sets of codewords that are based on
`
`application of irregular bipartite graphs to Gallager’s LDPC codes. EX2004 ¶¶ 59-
`
`64. A bipartite graph is divided into two distinct sets of nodes: (1) message nodes
`
`or variable nodes, representing the codeword that is transmitted; and (2) check
`
`nodes, representing constraints on the message nodes. EX2004 ¶¶ 40, 60. In the
`
`graph, message nodes may only be connected to check nodes, and vice versa.
`
`Bipartite graphs and parity check matrices are different ways of representing the
`
`same thing: the relationship between message bits and check equations. Bipartite
`
`graphs are often used as a visualization of the code’s parity check matrix. Because
`
`Gallager’s original LDPC code used only regular parity check matrices (i.e., parity
`
`check matrices represented by regular graphs), Luby proposed modifying
`
`Gallager’s code to use irregular graphs and parity check matrices. Luby thereby
`
`teaches certain sets of irregular codewords, but does not define an encoding
`
`process for achieving those codewords. Id., ¶ 60.
`
`
`
`-9-
`
`
`
`
`Luby did not, as Petitioner contends, conclude that all “irregular codes” are
`
`better than all “regular codes.” Id., ¶ 61. Rather, Luby found that certain irregular
`
`graphs, applied to Gallager’s LDPC codes, perform better than regular graphs in
`
`the same context. Many irregular Gallager codes, however, do not perform better
`
`than regular Gallager codes. Thus, Luby was largely devoted to explaining how to
`
`carefully design an irregular Gallager code that would be an improvement over
`
`corresponding regular Gallager codes. Id., ¶¶ 61-62.
`
`Luby also narrowly defines what is meant by “irregularity” in the context of
`
`that reference. Id., ¶ 63. Petitioner’s assertion that Luby broadly discovered
`
`characteristics of irregular codes is incorrect: Luby’s irregularity specifically
`
`referred to irregular graphs, in which the degree of the codeword is irregular. Id.,
`
`EX1204, p. 249. Luby did not consider regular or irregular repetition of
`
`information bits; indeed, Luby does not refer to information bits at all, since the
`
`analysis focuses on the encoded codeword and there is no disclosure of an encoder.
`
`EX2004 ¶¶ 63-64.
`
`C.
`
`Frey (EX1202)
`
`The Frey reference is not included in any ground of challenge asserted in
`
`this proceeding and, as such, any discussion in the petition regarding Frey should
`
`be given little weight. The petition does not make the requisite Graham inquiry or
`
`
`
`-10-
`
`
`
`
`demonstrate that Frey is prior art, nor does it rely on Frey as a ground of
`
`challenge.
`
`If Frey is considered, it only undermines any suggestion that irregular
`
`repetition would be introduced into a code of Divsalar. Id., ¶¶ 65-69. Contrary to
`
`Petitioner’s assertions, Frey did not find “that irregularity improved turbocodes.”
`
`Pet. 36. Rather, Frey found that the performance of his irregular turbocodes was
`
`highly sensitive to the permutation and puncturing scheme used. Id., ¶ 69;
`
`EX1202, p. 6. Most of the irregular turbocodes tested by Frey performed worse
`
`than Berrou’s original turbo code, and the one degree profile (out of nine) that was
`
`not markedly worse exhibited an error floor too high for any practical use.
`
`EX2004 ¶¶ 66-67. Indeed, Frey concludes that additional research and
`
`experimentation is necessary to reduce the error floor. Id., ¶ 68; EX1202, p. 6. If
`
`anything, Frey confirms that finding a workable irregular code structure is difficult
`
`and unpredictable (and beyond the grasp of Frey), not that a vague notion of
`
`irregularity can universally improve any error-correction code.
`
`IV. DR. DAVIS’S TESTIMONY SHOULD BE GIVEN LITTLE WEIGHT
`
`Petitioner relies on a declaration from Dr. Davis that should should be given
`
`little weight. As discussed below, Dr. Davis showed a general lack of knowledge
`
`about relevant work in this field, his declaration does not appear to reflect his own
`
`
`
`-11-
`
`
`
`
`independent work and opinions, and he was evasive and unresponsive to
`
`straightforward questions asked during his deposition.
`
`A. Dr. Davis’s testimony includes basic errors demonstrating a lack
`of credibility
`
`The Federal Circuit has found that basic technical errors are an important
`
`clue to witness credibility. See, e.g., Merck & Co. v. Teva Pharm. USA, Inc.,
`
`347 F.3d 1367, 1371 (Fed. Cir. 2003) (holding pharmaceutical testimony of a
`
`chemist to be less credible compared to the testimony of pharmacologists and
`
`noting chemist made errors that those in the art would have considered basic).
`
`Here, Dr. Davis could not answer basic questions about Berrou, the seminal paper
`
`on turbo codes, without rereading the entire article. EX2033, 54:17-60:3. He
`
`could not give an opinion on what “irregular” meant in the field, and implied such
`
`information was unhelpful or extraneous to the Board. Id., 87:7-89:16.
`
`During cross-examination, he candidly admitted he failed to consider the full
`
`scope and content of many of the references. He acknowledged that, in his direct
`
`testimony, he did not consider Frey's puncturing of parity bits despite Frey’s
`
`teaching puncturing as a necessity to keep the overall rate of the code constant.
`
`Id., 148:5-14; 150:3-10. Dr. Davis also displayed a curious lack of knowledge
`
`regarding Frey’s discussion of its error floor, mistakenly thinking Frey used the
`
`term “flattening effect” to refer to the top flat portion of the regular code's graph.
`
`Id., 165:3-166:18; see also EX2034 at 1 (cited at Frey p.6 and equating “flattening
`
`
`
`-12-
`
`
`
`
`of the error-curve” and “the so-called ‘error floor’”). The possibility of an error
`
`floor did not even cross his mind when asked why Frey’s irregular code did not
`
`show a lower BER than Berrou for high SNR. EX2033, 162:20-163:8. Despite
`
`acknowledging that error floors are undesirable, he confirmed that he did not at all
`
`consider Frey’s error floor in his obviousness analysis. Compare id. 166:8-10 with
`
`261:20-262:18. Although the petition and Dr. Davis mention Frey, Dr. Davis
`
`displayed a striking unfamiliarity with the reference.
`
`As to Luby, Dr. Davis repeatedly contended that the reference does not
`
`define what it means by “irregular” despite Luby’s express statement to the
`
`contrary. EX2033, 181:5-183:9, 194:4-18; EX1204, p. 249. A similar incomplete
`
`and cursory analysis of Divsalar is reflected in Dr. Davis’ direct testimony. For
`
`example, Dr. Davis proposes modifying an RA code discussed in the Kandakar
`
`thesis such that half the information bits are modified from degree 3 repeat to
`
`degree 2 repeat. A key conclusion of Divsalar, however, was that any value for
`
`repeat less than three would result in very poor performance. See EX1203, p. 6 (“It
`
`follows from (5.6) that an RA code can have word error probability interleaving
`
`gain only if q >= 3.”). See also EX2031 ¶ 27, 37.
`
`His unfamiliarity with the actual teachings of cited references, as well as
`
`with the actual knowledge in the relevant art, reflects a flawed and incomplete
`
`
`
`-13-
`
`
`
`
`Graham inquiry, and compels Dr. Davis to use improper hindsight. The testimony
`
`of Dr. Davis should be discounted accordingly.5
`
`B. Dr. Davis’s testimony is not independent
`
`While the petition and expert declaration are expected to be consistent,
`
`expert testimony that simply tracks and repeats the petition is entitled to little
`
`weight. Wowza Media Sys., LLC v. Adobe Sys., Inc., IPR2013-00054, Paper 16, 4
`
`(2013). Here, the petition and the Davis declaration show striking similarity,
`
`including the same language. For example, the sections discussing Ground 1 are
`
`nearly identical. Compare Pet., 34-47, with EX1206, ¶¶ 398-424. In addition,
`
`significant portions of Dr. Davis’s declaration were copied wholesale from Dr.
`
`Frey’s unsworn expert report in a completely different litigation proceeding and
`
`produced nearly two years prior to Dr. Davis’s declaration in this case. Compare
`
`
`5 Dr. Davis’s errors and unfamiliarity with the key references should also be
`
`considered in view of his admission that none of his publications related to repeat-
`
`accumulate codes, low-density parity-check codes, turbo codes, or irregular codes
`
`in general. EX2033, 27:4-28:9. Perhaps unsurprisingly, he also testified that he
`
`never attended the Allerton Conference on Communication, Control and
`
`Computing because “the work that I do in coding theory wasn’t being presented at
`
`that conference.” Id., 32:14-22.
`
`
`
`-14-
`
`
`
`
`EX1206, ¶¶ 22-46, with EX1217, ¶¶ 35-53, 55, 57-60, 63. This significantly
`
`undercuts the independence and objectivity of Dr. Davis’s testimony.
`
`C. Dr. Davis’s evasiveness during his deposition undermines his
`credibility
`
`In 10X Genomics, Inc. v. Univ. of Chicago, the Board explained that expert
`
`evasiveness or unresponsiveness during cross examination would reduce the
`
`weight of the expert’s direct testimony. IPR2015-01157, Paper 30, 2 (2016).
`
`During his deposition, Dr. Davis repeatedly refused to provide meaningful answers
`
`to straightforward questions about the field of error coding that should have easily
`
`been answerable by one of ordinary skill at the time. For example, he evaded
`
`questions on whether Berrou’s Figure 5 showed a relationship between bit error
`
`rate and signal-to-noise ratio despite the axes being clearly labeled as such.
`
`EX2033, 56:19-57:6, 58:19-59:3.
`
`Regarding “irregular,” a key term in this trial, he avoided answering whether
`
`his definition of irregular was the conventional meaning of “irregular” as generally
`
`used in the field of error correction codes. Id., 66:10-68:4. He avoided answering
`
`where the prior art provided a definition of “irregular” that was the same as his
`
`definition. Id., 72:17-75:18. He avoided answering what definition of “irregular”
`
`he would use in the field of error correction codes generally. Id., 78:18-81:12. He
`
`avoided answering whether his definition of “irregular” was consistent with the
`
`definition used with Tanner graphs. Id., 83:21-87:6. He avoided answering
`
`
`
`-15-
`
`
`
`
`whether Frey teaches reducing repetition to achieve irregular repetition of
`
`information bits (it does not). Id., 136:3-137:1. His unresponsiveness during cross-
`
`examination on this pivotal term is striking and warrants discounting his testimony.
`
`Other striking instances of evasiveness include his avoidance of answering
`
`whether Frey’s Figure 3(b) graph shows data points for the range of 8-12 degrees.
`
`Id., 138:21-140:9. He was unresponsive for seven pages of transcript regarding the
`
`simple question of whether puncturing would lead to a lower rate relative to a code
`
`without puncturing (Frey states puncturing is critical for rate reduction). Id.,
`
`152:12-159:14. He similarly avoided answering questions about prior art at issue
`
`in related IPRs. Id., 249:2-251:21, 269:21-272:12.
`
`The contrast between cross-examination and redirect further confirms that
`
`Dr. Davis did not act as an independent expert. Redirect occurred after a break
`
`during which Dr. Davis had a “discussion about the substance of the testimony and
`
`the general nature of the redirect” with Apple’s counsel. Id., 275:9-13. This
`
`discussion enabled Dr. Davis to be far more responsive and direct for Apple’s
`
`counsel. This witness behavior is precisely the sort of behavior the Board has
`
`condemned in decisions like 10X Genomics. These shenanigans are inimical to the
`
`efficiency of Board proceedings and the integrity of the patent system. The
`
`appropriate response is to accord little or no weight to the direct and redirect
`
`testimony of Dr. Davis.
`
`
`
`-16-
`
`
`
`
`V. GROUND 1: DIVSALAR IN VIEW OF LUBY DOES NOT RENDER
`CLAIMS 1-8 AND 11-14 OBVIOUS
`
`The petition fails to demonstrate that claims 1-8 and 11-14 would have been
`
`obvious in view of the combination of Divsalar and Luby for at least the following
`
`reasons. First, neither Luby nor Divsalar disclose irregular repetition of
`
`information bits. Second, for claims 1-8, the petition fails to identify a
`
`“partitioning” step in Luby or Divsalar. Third, a person of ordinary skill in the art
`
`would not have been motivated by Luby to incorporate irregular repetition into
`
`Divsalar. Luby teaches that modifying classical Gallager codes using carefully
`
`constructed irregular bipartite graphs may improve performance of certain
`
`codewords, but does not teach irregular repetition of information bits, much less
`
`suggest that irregularity may be applied in other ways to other classes of codes in
`
`order to achieve similar improvements. Fourth, the petition’s proposed
`
`modification of Divsalar is not based on any teaching in Luby. Finally, Petitioner
`
`offers no evidence to support its burden of demonstrating that the proposed
`
`combination of Divsalar and Luby would have a reasonable expectation of
`
`succeeding – a defect that in itself is sufficient to deny Ground 1. Indeed, the
`
`evidence indicates that the opposite is true. For these reasons, Ground 1 should be
`
`rejected.
`
`
`
`-17-
`
`
`
`
`A. Legal Principles
`
`In order to establish that a patent claim is obvious under 35 U.S.C. § 103,
`
`one must first determine (1) the scope of the prior art, (2) differences between the
`
`prior art and the claims at issue, and (3) the level of ordinary skill in the art—
`
`“Against this background, the obviousness or nonobviousness of the subject matter
`
`is determined,” with additional “secondary considerations” given to certain indicia
`
`of nonobviousness. KSR Intern. Co. v. Teleflex Inc., 550 U.S. 398, 404 (2007)
`
`(citing Graham v. John Deere Co., 383 U.S. 1, 17-18 (1950)). Those challenging a
`
`claim must provide some articulated reasoning that includes identifying “a reason
`
`that would have prompted a person of ordinary skill in the relevant field to
`
`combine the elements in the way the claimed new invention does.” Id.
`
`Importantly, it is also a petitioner’s burden to show that at the time of the
`
`invention there was a “reasonable expectation of success” for the proposed
`
`combination. Intelligent Bio-Sys. v. Illumina Cambridge, 821 F.3d 1359, 1367-68
`
`(Fed. Cir. 2016); see also MPEP § 2143.2.I (“Obviousness requires a reasonable
`
`expectation of success”). Thus, merely identifying elements in the prior art is not
`
`sufficient to establish obviousness—a person of ordinary skill must have
`
`reasonably expected that the combination would have succeeded for its intended
`
`purpose. DePuy Spine, Inc. v. Medtronic Sofamor Danek, Inc., 567 F.3d 1314,
`
`1326 (Fed. Cir. 2009) (“Although predictability is a touchstone of obviousness, the
`
`
`
`-18-
`
`
`
`
`‘predictable result’ discussed in KSR refers not only to the expectation that prior
`
`art elements are capable of being physically combined, but also that the
`
`combination would have worked for its intended purpose.”).
`
`Accordingly, where, as here, a petitioner fails to explain or provide evidence
`
`as to how the proposed combination would predictably result in the improvement
`
`that allegedly motivated the combination, the Board must decline to find the claims
`
`unpatentable for obviousness. See, e.g., JTEKT Corp. v. GKN Automotive, Ltd.,
`
`IPR2016-00046, Paper No. 27 at 28-29 (Jan. 23, 2017).
`
`B. Divsalar in view of Luby does no