`Filed: February 10, 2023
`
`
`
`
`
`
`
`
`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`_____________________________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`_____________________________
`
`SAMSUNG ELECTRONICS CO., LTD.,
`Petitioner,
`
`v.
`
`CALIFORNIA INSTITUTE OF TECHNOLOGY,
`Patent Owner.
`_____________________________
`
`Case No. IPR2023-00130
`Patent No. 7,116,710
`_____________________________
`
`PATENT OWNER’S PRELIMINARY RESPONSE
`PURSUANT TO 37 C.F.R. § 42.107
`
`
`
`
`
`
`
`TABLE OF CONTENTS
`Introduction ....................................................................................................... 1
`I.
`II. Claim Construction ........................................................................................... 3
`III. The Petition is Founded on a Distortion of Kobayashi .................................... 5
`IV. Ground 1 Fails ................................................................................................... 9
`A. Kobayashi Does Not Disclose Irregular Repetition (All Claims) ............ 10
`1. Petitioner misinterprets the claims’ construction ...............................11
`2. Kobayashi does not expressly disclose irregular repetition ...............13
`3. Petitioner fails to show Kobayashi’s first encoder inherently performs
`irregular repetition ..............................................................................15
`4. Petitioner’s arguments are fatally inconsistent ...................................19
`B. Kobayashi Does Not Disclose a First Encoded Data Block Formed
`of Irregularly Repeated Bits (Claim 11) ................................................... 27
`C. Kobayashi Does Not Disclose Scrambling Irregularly Repeated
`Bits (Claims 15 & 25) ............................................................................... 29
`V. Ground 2 Fails ................................................................................................. 33
`VI. Ground 3 Fails ................................................................................................. 33
`VII. Institution Should be Denied under 35 U.S.C. §314(a) .................................. 36
`A. Fintiv Applies to This Proceeding ............................................................ 37
`B. The Fintiv Factors Weigh in Favor of Denying Institution ...................... 39
`1. Factor 1 favors denial .........................................................................39
`2. Factor 2 favors denial .........................................................................40
`3. Factor 3 favors denial .........................................................................42
`4. Factor 4 does not favor institution ......................................................45
`5. Factor 5 favors denial .........................................................................46
`6. Factor 6 favors denial .........................................................................47
`VIII. Conclusion ...................................................................................................... 50
`IX. Appendix ......................................................................................................... 52
`
`
`
`
`
`
`-i-
`
`
`
`I.
`
`INTRODUCTION
`
`This is the sixth IPR petition challenging the claims of U.S. Patent No.
`
`7,116,710 (“the ’710 patent”). Despite five prior petitions1 and two instituted trials,
`
`the Board has never found a single claim of the ’710 patent unpatentable.2 The
`
`present petition does not warrant a different outcome.
`
`Petitioner Samsung Electronics Co., Ltd. advances a challenge premised on
`
`anticipation by Kobayashi, yet various aspects of the challenged claims are simply
`
`missing from Kobayashi. Rather than meeting its burden to show that Kobayashi
`
`discloses each limitation, Petitioner seeks to avoid the claims’ requirements by
`
`misreading limitations or simply assuming that Kobayashi meets the claims despite
`
`its silence on the issue. For example, each independent claim includes a limitation
`
`directed to irregular repetition—expressed either as repeating received bits “a
`
`different number of times,” as recited in claim 11, or simply as “repeat[ing] said
`
`stream of bits irregularly,” as recited in claims 15 and 25. Kobayashi does not
`
`disclose irregular repetition as recited, and Petitioner fails to show that it is
`
`
`1 IPR2017-00210; IPR2017-00211; IPR2017-00219; IPR2015-00067; IPR2015-
`
`00068.
`
`2 The Federal Circuit summarily affirmed the final written decisions upholding
`
`all claims. Apple Inc. v. Cal. Inst. of Tech., 796 F. App’x 743 (Fed. Cir. 2020).
`
`-1-
`
`
`
`
`inherent. In fact, Petitioner actually provides examples of encoding techniques that
`
`undermine the logic on which Petitioner grounds its inherency argument.
`
`The petition is similarly deficient with respect to other limitations.
`
`Independent claim 11 recites that a first encoded data block is formed by repeating
`
`bits of a received data block different numbers of times. Yet what Petitioner
`
`identifies in Kobayashi for this element is not a block formed by irregularly
`
`repeating bits, but a block consisting of one copy of each original bit plus parity
`
`bits that are not repeated bits at all. As for independent claims 15 and 25, each
`
`requires scrambling of irregularly repeated bits, yet Kobayashi never scrambles
`
`irregularly repeated bits. Petitioner points instead to an interleaving of a single
`
`copy of each original bit with a group of parity bits, not to a scrambling of a
`
`collection of irregularly repeated bits.
`
`In addition to these deficiencies, Petitioner’s late filing of this petition
`
`warrants discretionary denial in light of the co-pending litigation in the Eastern
`
`District of Texas in which Petitioner is a defendant. Petitioner filed the present
`
`case less than a year before the scheduled trial. Under even generous estimates, the
`
`district court trial will be completed many months before a final written decision
`
`would be due in this case. Given the substantial costs that will arise due to
`
`duplication of efforts, Petitioner’s unexplained and unexcused delay, and the weak
`
`merits of the petition’s grounds, the Board should not institute trial.
`
`-2-
`
`
`
`
`Accordingly, institution of inter partes review of claims 11-17 and 19-33 of
`
`U.S. Patent No. 7,116,710 (“the ’710 patent”) should be denied.
`
`II. CLAIM CONSTRUCTION
`
`In an inter partes review, a claim is given its ordinary and customary
`
`meaning in light of the specification. 37 C.F.R. §42.100(b); Phillips v. AWH Corp.,
`
`415 F.3d 1303, 1312 (Fed. Cir. 2005) (en banc). Petitioner asserts that the term
`
`“repeat” should be construed to mean “generation of additional bits, where
`
`generation can include, for example, duplication or reuse of bits,” in accordance
`
`with a district court construction affirmed by the Federal Circuit. Pet., 8 (citing
`
`Cal. Inst. of Tech. v. Broadcom Ltd., 25 F.4th 976, 986 (Fed. Cir. 2022)
`
`(“Broadcom litigation”)).
`
`However, Petitioner extends this construction in an unreasonable way and
`
`mischaracterizes the Federal Circuit’s construction. Petitioner first asserts that (1)
`
`the Federal Circuit held that “passing an input information bit through an AND
`
`gate when the other input is a ‘1’ bit comprises ‘repeating’ the information bit”; (2)
`
`“[m]ultiplying a binary information bit by a “1” bit is equivalent to passing the
`
`information bit through an AND gate with a ‘1’ bit”, and (3) therefore,
`
`“multiplying an information bit by a “1” bit comprises ‘repeating’ the information
`
`bit.” Pet., 14-15. The Federal Circuit never said this—its claim construction never
`
`discussed either multiplication or AND gates. See Broadcom, 25 F.4th at 986.
`
`-3-
`
`
`
`
`Certainly it never said anything about multiplying by “1” being equivalent to a
`
`device that uses an AND gate, let alone whether that would constitute, by itself, a
`
`repeat under the claims. To the extent Petitioner implies otherwise, that is a
`
`mischaracterization of the Federal Circuit’s decision.
`
`Even taking as true Petitioner’s assertion that the Federal Circuit considered
`
`whether passing an input bit through an AND gate where the other bit is a “1”
`
`comprises repeating the information bit, the Federal Circuit’s discussion related to
`
`infringement, not claim construction. See id., 986-88. The Federal Circuit made it
`
`clear that it was addressing a very specific system implemented in infringing
`
`products, and that the relevant analysis considered the system’s “overall
`
`architecture,” not just an individual component. Id., 988. That system involved a
`
`device that physically connected input bits via wires to 972 separate AND gates
`
`and simultaneously transmitted duplicates of the bits via a number of selected gates
`
`ranging from 3 to 12 to be used in forming parity bits. See id., 986-88 (“[T]he
`
`physical connection of the first inputs of all 972 AND gates for simultaneous
`
`receipt of the information bit stream and the connection of the parity-bit system to
`
`the other inputs of the AND gates to selectively enable 3 to 12 of those gates at any
`
`time together implement irregular repetition.” (emphases original)). In other
`
`words, the Federal Circuit merely agreed that a device could be found to
`
`“irregularly repeat” if its overall configuration for encoding was arranged to make
`
`-4-
`
`
`
`
`multiple simultaneous duplicate copies of input bits (in irregular amounts) and to
`
`transmit the copied bits via separate wires through a selected set of a varying
`
`number of parallel AND gates. The Federal Circuit said nothing about multiplying
`
`a bit by “1”.
`
`Moreover, as discussed below in §IV.A, the petition’s invalidity case falls
`
`apart because its fundamental premise—that Kobayashi’s outer encoder
`
`necessarily multiplies bits by “1” in a pattern determined by its generator matrix—
`
`is unfounded, fatally undermining Petitioner’s inherency case. Even if Petitioner’s
`
`interpretation of the Federal Circuit’s construction were accepted, Petitioner’s case
`
`fails, as it requires reading the claims’ recited irregular repetition into Kobayashi
`
`despite the reference neither teaching nor requiring such repetition.
`
`III. THE PETITION IS FOUNDED ON A DISTORTION OF KOBAYASHI
`
`The ’710 patent is one of four Caltech patents that resulted from research
`
`performed by the inventors, Hui Jin, Aamod Khandekar, and Robert J. 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. See EX1001, 2:6-10
`
`-5-
`
`
`
`
`(linear time decoding), 6:24-60 (performance near theoretical Shannon limit);
`
`EX2005, 7 (linear time encoding).
`
`As the Board has previously found, the field of error correction coding is
`
`complex and highly unpredictable. IPR2017-00219, Paper 76 at 18. The
`
`development of IRA codes in spite of this unpredictability represented a significant
`
`advancement in encoding technology. Accordingly, the specification and claims
`
`focus strongly on precise details of encoding processes that allow the realization of
`
`IRA codes’ improved performance.
`
`By contrast, Kobayashi focuses almost exclusively on decoding. Its title
`
`describes error-correction decoding of a received data stream. Its abstract focuses
`
`on its “decoding procedure,” and discusses encoding schemes only to emphasize
`
`their unimportance, due to the broad applicability of its decoders, which it says can
`
`be “applied to many existing systems” without modifying “the transmitter side.”
`
`EX1005, Abstract. The field of invention describes “error correction of a received
`
`data stream” and never once mentions encoding. EX1005, 1:7-11. The summary
`
`likewise describes decoding, not encoding. EX1005, 4:28-59. Where the
`
`specification describes encoding steps, the specification uses simple examples with
`
`scant detail. See, e.g., EX1005, 7:46-8:2 (describing a “simple packet transmission
`
`system” performing a Hamming encoding of just 4 bits at a time).
`
`-6-
`
`
`
`
`Petitioner takes advantage of Kobayashi’s sparse encoder description by
`
`latching onto a single example encoder and reading in features that are simply not
`
`there. Petitioner describes Kobayashi’s outer encoder as performing repetitions,
`
`and in particular of multiplying input bits by “1” based on the entries in a generator
`
`matrix G. Pet., 14-15. However, as discussed below in §IV.A, Kobayashi discloses
`
`neither repetition nor multiplying for its outer encoder. Petitioner provides an
`
`illustration of the purported multiplications and sums that Kobayashi performs
`
`(Pet., 16), but this illustration is derived from Petitioner’s imagination, not
`
`Kobayashi’s disclosure. Due to Petitioner’s misinterpretation of Kobayashi,
`
`Petitioner repeatedly concludes that the outer encoder practices claim elements that
`
`are neither expressly disclosed nor inherent. See infra §§IV.A-C.
`
`Petitioner similarly mischaracterizes Kobayashi’s inner encoder. Petitioner
`
`claims that “[w]hile duobinary signaling is depicted in Figure 8 as part of the
`
`‘inner encoder,’ Kobayashi discloses that duobinary signaling is simply a
`
`transmission technique, and thus the precoder is the inner encoder component that
`
`performs the ‘second encoding’ step as claimed.” Pet., 17-18 n.7 (original
`
`emphasis). This flatly contradicts Kobayashi. Kobayashi expressly states that there
`
`are exactly two codes used on Fig. 8’s transmission side: the “outer code[ is] a
`
`(7,4) Hamming code,” and “the inner code is duobinary signaling with a precoder.”
`
`EX1005, 7:8-11. Thus, far from being simply a transmission technique, duobinary
`
`-7-
`
`
`
`
`signaling is a part of the inner code, just as much as the precoder is. Kobayashi
`
`confirms this by illustrating the “inner encoder” as being both components
`
`together, not just one taken alone. See EX1005, Fig. 8. Kobayashi also provides the
`
`encoded sequence that results from the duobinary encoding, labeling it as “I5.” See
`
`EX1005, 8:28-33. This duobinary sequence is easily distinguished from earlier
`
`binary sequences because it contains “2”s in addition to “1”s and “0”s.
`
`Kobayashi’s decoder subsequently uses duobinary sequences [interspersed with
`
`error symbols] throughout much of its decoding process, confirming that the
`
`duobinary sequence is the inner encoding, just as Kobayashi says it is. See
`
`EX1005, 8:55-10:45. This is consistent with Kobayashi’s selection of the word
`
`precoder to characterize that component of the inner encoder, suggesting it is a
`
`first component that is used prior to forming a code via the operation of a second
`
`component.
`
`Petitioner’s citations to Kobayashi do not support its position that the
`
`precoder alone functions as an encoder. The first citation (EX1005, 7:30-31)
`
`simply states that duobinary signaling is used for transmission, not that duobinary
`
`is not also an encoding step. The second citation is even worse for Petitioner’s
`
`argument: it expressly describes duobinary signaling as a type of “partial-response
`
`-8-
`
`
`
`
`channel coding.”3 EX1005, 7:43-45. Kobayashi’s description of such coding as “a
`
`bandwidth-efficient transmission technique” (EX1005, 2:21-25) does not mean that
`
`Kobayashi’s duobinary signaling module is not also performing an encoding as an
`
`essential part of the inner encoder—Kobayashi is simply describing that encoding
`
`in that manner makes for bandwidth-efficient transmission. Lastly, the cited
`
`footnote in the Valenti declaration adds nothing, merely parroting the petition
`
`while adding the phrase “in my opinion.” EX1002, ¶69 n.3.
`
`In sum, the petition’s discussion of Kobayashi is replete with
`
`mischaracterization and inaccuracies. As discussed in §§IV-VI, these defects are
`
`fatal to all of the petition’s grounds.
`
`IV. GROUND 1 FAILS
`
`Petitioner fails to show that Kobayashi anticipates claims 11-12, 14-17, 19,
`
`21-22, 24-27, 29, and 32-33 of the ’710 patent. Yet as discussed below, Petitioner
`
`fails to show that Kobayashi discloses multiple limitations in independent claims
`
`11, 15, and 25. These limitations include the irregular repetition recited in each
`
`independent claim, the first encoded data block formed of repeated bits as recited
`
`in claim 11, and the scrambling of irregularly repeated bits as recited in claims 15
`
`
`3 All emphasis added unless otherwise indicated.
`
`-9-
`
`
`
`
`and 25. Because Kobayashi has not been shown to disclose any of these
`
`limitations, Ground 1 fails.
`
`A. Kobayashi Does Not Disclose Irregular Repetition (All Claims)
`
`The petition fails to show that Kobayashi discloses irregular repetition of
`
`bits as recited in independent claims 11, 15, and 25. Claim 11 recites that “said
`
`plurality of bits are repeated a different number of times,” and claims 15 and 25
`
`each recite a “first coder operative to repeat said stream of bits irregularly.” None
`
`of these limitations has been shown to be present in Kobayashi, either expressly or
`
`inherently.
`
`Broadly speaking, Petitioner asserts that Kobayashi’s outer (Hamming)
`
`encoder discloses this limitation. See Pet., 12-17. As explained below, Kobayashi
`
`does not describe irregular repetition or disclose any details of how the outer
`
`encoder operates; it only provides the parity-check and generator matrices that
`
`characterize the resulting output. Petitioner nonetheless argues that the irregular
`
`repetition recited in the claims—i.e., repeating bits “a different number of times”
`
`or “irregularly”—is in some way equivalent to operations that might be performed
`
`by such an encoder, and that Kobayashi therefore discloses these claim limitations.
`
`As discussed in the following sections, Petitioner misapplies claim construction
`
`and anticipation principles and relies on flawed, self-contradictory reasoning to
`
`conjure up irregular repetition from a reference that never discloses it.
`
`-10-
`
`
`
`
`1. Petitioner misinterprets the claims’ construction
`As discussed above in §II, Petitioner misinterprets the Federal Circuit’s
`
`construction of “repeat,” stretching the actual construction to conclude that every
`
`act of “multiplying a bit by a ‘1’ bit comprises ‘repeating’ the information bit,” on
`
`the basis that passing a bit through an AND gate can perform an operation
`
`“equivalent” to multiplying by “1.” Pet., 14-15. Petitioner then further extends this
`
`interpretation, arguing that if an encoder, such as Kobayashi’s Hamming encoder,
`
`performs a transform that can be characterized by a non-zero generator matrix with
`
`different numbers of “1”s in different rows, then that encoder must necessarily
`
`repeat bits different numbers of times or irregularly. See Pet., 15-16. Petitioner
`
`does not justify its logical leap from a finding of infringement by a particular
`
`device that duplicated bits and sent them through a specific configuration of AND
`
`gates to a conclusion that every implementation of a transformation that could be
`
`performed using binary multiplication must repeat bits.
`
`The petition does not explain what it hopes to prove by referring to binary
`
`multiplication as being “equivalent to” an AND gate. Pet., 14-15 (“Multiplying a
`
`binary information bit by a ‘1’ bit is equivalent to passing the information bit
`
`through an AND gate with a ‘1’ bit, and thus under this construction of ‘repeat,’
`
`multiplying an information bit by a ‘1’ bit comprises ‘repeating’ the information
`
`bit.”). Certainly, the petition never asserts that any of the ’710 patent’s claims
`
`-11-
`
`
`
`
`include a means- or step-plus-function limitation, nor does it perform the
`
`corresponding analysis of whether the two types of operation are equivalents. Cf.,
`
`e.g., MPEP §2183 (discussing factors supporting finding of equivalence in means-
`
`and step-plus function claims). To the extent that Petitioner suggests that two
`
`processes cannot be patentably distinct if they can produce equivalent outputs, this
`
`reasoning would not satisfy the law of anticipation. “Anticipation requires identity
`
`of the claimed process and a process of the prior art; the claimed process, including
`
`each step thereof, must have been described or embodied, either expressly or
`
`inherently, in a single reference.” Glaverbel Societe Anonyme v. Northlake Mktg.
`
`& Supply, Inc., 45 F.3d 1550, 1554 (Fed. Cir. 1995). Even if two steps might
`
`produce equivalent results, that would not make the steps themselves identical. As
`
`to the ’710 patent claims, that a binary multiplication operation might produce an
`
`output “equivalent” in some way to a repetition operation does not mean that a
`
`process that could be accomplished with binary multiplication necessarily performs
`
`a step of repetition, much less the particular irregular repetition recited in the
`
`claims.
`
`Regardless, as discussed below, even if Petitioner’s distortion of the Federal
`
`Circuit’s opinion to equate multiplying input bits by “1” with repetition were
`
`accepted, Kobayashi’s outer encoder does not expressly or inherently disclose
`
`-12-
`
`
`
`
`performance of such operations, and Petitioner’s attempt to stretch Kobayashi’s
`
`disclosure to reach even their own interpretation of the claims is fatally flawed.
`
`2. Kobayashi does not expressly disclose irregular repetition
`The petition asserts that in the outer (Hamming) encoder depicted in Fig. 8
`
`of Kobayashi, “the information bits are repeated irregularly such that information
`
`bits are repeated a different number of times.” Pet., 14-16. However, the petition
`
`does not (as it could not) claim that Kobayashi expressly describes its outer
`
`encoder as performing repetition, much less irregular repetition. See id. The
`
`petition’s argument is instead premised on a theory that Kobayashi’s outer encoder
`
`performs repetition by multiplying input bits by “1.” Id. However, even assuming
`
`that multiplying bits by “1” constitutes repetition (which the petition fails to
`
`substantiate, see supra §II), Kobayashi never discloses any such multiplication for
`
`its outer encoder.
`
`While Kobayashi describes the inputs to its encoder and how the encoder’s
`
`output is mathematically related to its inputs, it provides almost no details as to
`
`what process its outer encoder uses to generate its outputs. Kobayashi describes
`
`Fig. 8 as having two codes: for “the outer code, a (7, 4) Hamming code is used and
`
`the inner code is duobinary signaling with a precoder.” EX1005, 7:9-11.
`
`Kobayashi never describes its outer (Hamming) encoder as performing repetition.
`
`Kobayashi also never describes its outer encoder as multiplying input bits by
`
`-13-
`
`
`
`
`anything. Kobayashi says that a 28-bit packet is “segmented into blocks of k=4
`
`bits, and each block is then encoded to a codeword of length n=7, by using a (7,4)
`
`Hamming code.” EX1005, 7:49-65. Kobayashi specifies that the Hamming code’s
`
`“parity-check and generator matrices are given in systematic form by” the
`
`following matrices:
`
`
`
`Id. The remainder of Kobayashi’s discussion of the outer encoder merely specifies
`
`what the resulting 49 bit values are for its example input; thereafter, Kobayashi
`
`discusses what is done subsequently with the output, never returning to elaborate
`
`on how the outer encoder operates. See EX1005, 7:66-8:5. While Kobayashi
`
`specifies the result of its encoding (namely, that the output of its encoder is related
`
`to the input by the specified generator and parity-check matrices), it never provides
`
`any details of how the output is actually computed. In particular, Kobayashi never
`
`states that the bits input to the Hamming encoder are repeated in any way, much
`
`less irregularly. Nor does Kobayashi say that the input bits are multiplied by “1”s
`
`at all, and certainly not different numbers of times. Accordingly, even under
`
`-14-
`
`
`
`
`Petitioner’s interpretation of “repeat,” there is no express teaching in Kobayashi
`
`that its outer encoder repeats bits at all, much less “irregularly” or “a different
`
`number of times” as required by the independent claims.
`
`3. Petitioner fails to show Kobayashi’s first encoder inherently
`performs irregular repetition
`In an attempt to avoid Kobayashi’s lack of express disclosure, Petitioner
`
`turns to an inherency argument, but this argument is grounded in unsupported and
`
`false assumptions. Petitioner’s case rests on a mistaken assumption that an encoder
`
`characterized by a particular generator matrix necessarily performs a pattern of
`
`repetitions corresponding to the “1”s in that generator matrix.
`
`Inherency sets a high bar to show that a limitation is necessarily present in
`
`the cited prior art. “[A]nticipation by inherent disclosure is appropriate only when
`
`the reference discloses prior art that must necessarily include the unstated
`
`limitation.” Therasense, Inc. v. Becton, Dickinson and Co., 593 F.3d 1325, 1332
`
`(Fed. Cir. 2010) (quoting Transclean Corp. v. Bridgewood Servs., Inc., 290 F.3d
`
`1364, 1373 (Fed. Cir. 2002)) (emphasis and alteration in original). “Inherent
`
`anticipation requires that the missing descriptive material is ‘necessarily present,’
`
`not merely probably or possibly present, in the prior art.” Trintec Indus., Inc. v.
`
`Top-U.S.A. Corp., 295 F.3d 1292, 1295 (Fed. Cir. 2002) (quoting In re Robertson,
`
`169 F.3d 743, 745 (Fed. Cir. 1999)).
`
`-15-
`
`
`
`
`In addition to the erroneous application of claim construction discussed in
`
`§IV.A.1, Petitioner’s inherency argument relies on further mistaken reasoning.
`
`Petitioner asserts that “any type of linear code using a non-zero generator matrix
`
`will ‘repeat’ input bits because the process of multiplying a vector of information
`
`bits by the generator matrix will necessarily involve multiplying input bits by ‘1’
`
`bits.” Pet., 15. More specifically, Petitioner asserts that because Kobayashi’s
`
`generator matrix G has different numbers of “1”s in different rows, the encoder
`
`characterized by that generator matrix multiplies different input bits by “1”
`
`different numbers of times, with this multiplication producing irregular repetition
`
`of the input bits. See Pet., 15-16. However, Petitioner is wrong.
`
`Even presuming Petitioner’s interpretation of the “irregular repetition”
`
`limitations is correct (which, as discussed in §IV.A.1, it is not), Petitioner’s
`
`reasoning still fails because Petitioner assumes the existence of operations by
`
`Kobayashi’s encoder that are neither disclosed nor inherent. Petitioner’s assertion
`
`(p. 15) that “any type of linear code using a non-zero generator matrix will ‘repeat’
`
`input bits because the process of multiplying a vector of information bits by the
`
`generator matrix will necessarily involve multiplying input bits by ‘1’ bits” is
`
`unsubstantiated and simply not true. Petitioner points to nothing in Kobayashi
`
`stating that any input bit of its Hamming encoder is multiplied by any bit at all.
`
`This is unsurprising, since Kobayashi provides no details of how its Hamming
`
`-16-
`
`
`
`
`encoder transforms the input bits into output bits. Indeed, Kobayashi never states
`
`that any of the elements depicted in the transmitter of Fig. 8 multiplies an input bit
`
`by another bit. See EX1005, 7:5-8:33, Fig. 8. The petition also never explains why
`
`it should be true that if an encoder is characterized by a nonzero generator matrix,
`
`it necessarily performs its encoding using binary multiplication. See Pet., 15.
`
`Petitioner’s further assertion that because Kobayashi’s generator matrix G
`
`has different numbers of “1”s in different rows, the encoder multiplies different
`
`input bits by “1” different numbers of times, thereby producing irregular repetition
`
`of the input bits, fails for similar reasons. Since Kobayashi never describes any of
`
`the input bits to the Hamming encoder being multiplied by anything, it certainly
`
`never discloses them being multiplied by “1” different numbers of times. Nor is
`
`there any explanation of why it would be necessary to multiply input bits by “1”
`
`different numbers of times to produce the encoding described by Kobayashi. See
`
`Pet., 15.
`
`The testimony the petition cites to also fails to substantiate Petitioner’s
`
`assertion that every encoder with a non-zero generator matrix must multiply input
`
`bits by “1,” much less that Kobayashi’s encoder multiplies input bits by different
`
`numbers of “1”s according to the rows of the generator matrix G. Dr. Valenti
`
`repeats the petition’s statement “under this construction, any type of linear code
`
`using a non-zero generator matrix will ‘repeat’ input bits because the process of
`
`-17-
`
`
`
`
`multiplying a vector of information bits by the generator matrix will necessarily
`
`involve multiplying input bits by ‘1’ bits,” but this circular argument puts the cart
`
`before the horse. EX1002, ¶64. That is, Dr. Valenti assumes that Kobayashi
`
`involves a “process of multiplying a vector of information bits by the generator
`
`matrix” to reach the conclusion that such an operation necessarily multiplies input
`
`bits by “1.” But the initial presumption that Kobayashi’s encoder multiplies
`
`information bits by the generator matrix is unsupported. In the next sentence, he
`
`states, “For example, as I described above, Kobayashi’s Hamming encoder
`
`multiplies each 4-bit sub-block of I1 by the 4x7 generator matrix G, resulting in
`
`seven 7-bit codewords.” EX1002, ¶65. However, this is the first time Dr. Valenti
`
`actually asserted that Kobayashi performs such multiplications, and Kobayashi
`
`itself says nothing about performing multiplications, so this assertion is
`
`unsupported. See EX1002, ¶¶61-65. Dr. Valenti presents a figure (reproduced in
`
`the petition at p. 16) purporting to show the multiplication, but this is Petitioner’s
`
`creation, not Kobayashi’s disclosure. EX1002, ¶¶66-67.
`
`Dr. Valenti also cites to his own discussion of “matrix multiplication and
`
`modulo-2 arithmetic” (id. (citing EX1002, Section V.B (¶¶28-35)), but this
`
`undermines his opinion rather than supporting it. There, he says that matrix
`
`multiplication “is typically performed by taking the inner product of the vector
`
`with each of the columns of the matrix” involving certain multiplication and
`
`-18-
`
`
`
`
`addition operations, not that it can only be performed by such an inner product.
`
`EX1002, ¶28. Far from confirming that Kobayashi necessarily performs matrix
`
`multiplication in its first encoding or that such multiplication would necessarily
`
`involve multiplication of bits, Dr. Valenti posits that this is merely possible, falling
`
`short of establishing inherency.
`
`4. Petitioner’s arguments are fatally inconsistent
`Petitioner’s own arguments undermine the assumptions of its inherency
`
`case: those arguments imply that, contrary to the petition’s assertion, (1) a
`
`component of an encoder can perform an operation that has a non-zero generator
`
`matrix with different numbers of “1”s in different rows without performing any
`
`irregular repetition of input bits whatsoever; and (2) an encoder could perform
`
`Kobayashi’s Hamming encoding in particular without ever using an AND gate or
`
`binary multiplication operation on the input bits. These conclusions are apparent
`
`from Petitioner’s own statements in this and other petitions, and they are
`
`substantiated by at least one further example of an encoder that encodes according
`
`to Kobayashi’s Hamming code without any repetition of bits. In light of these fatal
`
`deficiencies, Petitioner’s arguments cannot stand.
`
`-19-
`
`
`
`
`The arguments elsewhere in the petition provide an example of why a
`
`component of an encoder4 that can be characterized by a non-zero generator matrix
`
`need not multiply input bits to perform its linear transform, and also need not
`
`repeat input bits different numbers of times even if the generator matrix has
`
`different numbers of “1”s in different rows. When discussing Kobayashi’s
`
`precoder, Petitioner describes it as “an accumulator” that performs a “linear
`
`transformation.” Pet., 19-20. Petitioner evokes the same misleading “equivalence”
`
`reasoning as it uses for its irregular repetition arguments, asserting that “the
`
`accumulation operation is equivalent to multiplying the 1x49 vector I3 by a 49x49
`
`generator matrix GA,” which it says is constructed as follows:
`
`
`
`
`4 Petitioner misidentifies Kobayashi’s precoder as being an encoder that
`
`performs the inner encoding operation. In fact, it is