`
`_________________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`_________________
`
`SAMSUNG ELECTRONICS AMERICA, INC. and
`SAMSUNG ELECTRONICS CO. LTD.
`
`Petitioners,
`
`v.
`
`FASTVDO LLC
`Patent Owner.
`
`_________________
`
`U.S. Patent No. 5,850,482
`Case IPR2016-01179
`
`_________________
`
`SAMSUNG ELECTRONICS AMERICA, INC.
`AND SAMSUNG ELECTRONICS CO. LTD.’S
`REPLY TO PATENT OWNER FASTVDO’S RESPONSE
`
`
`
`TABLE OF CONTENTS
`
`IPR2016-01179
`U.S. Patent No. 5,850,482
`
`
`Page(s)
`
`B.
`
`I.
`II.
`
`INTRODUCTION ........................................................................................... 1
`FASTVDO’S SOLE CHALLENGE TO THE
`OBVIOUSNESS OF MOST CLAIMS IS TO CHALLENGE
`THE COMBINATION OF KATO’S COMPLEMENTARY
`EMBODIMENTS ............................................................................................ 1
`A. Kato Provides A Reason To
`Combine Embodiments – Adding Unequal
`Error Protection To Prevent Error Propagation .................................... 1
`In Combining Kato’s Embodiments,
`There Is A Reasonable Expectation Of Success ................................. 10
`III. FASTVDO ALLEGES ONLY ONE
`CLAIM ELEMENT IS “NOT DISCLOSED
`OR SUGGESTED,” AN ELEMENT THAT IS OBVIOUS
`IN VIEW OF KATO’S ERROR-RESILIENT DATA STORAGE .............. 16
`A.
`Storage In Kato’s Fourth Embodiment ............................................... 20
`B.
`Storage In Kato’s First Embodiment ................................................... 20
`IV. KATO TEACHES AN “ASSOCIATED
`SECOND PORTION OF EACH CODE WORD
`INCLUDES INFORMATION REPRESENTATIVE OF
`A RESPECTIVE PORTION OF THE ORIGINAL DATA” ........................ 22
`THE “DATA LINK” CLAIMS ARE SIMILARLY OBVIOUS .................. 24
`V.
`VI. CONSTITUTIONALITY .............................................................................. 25
`VII. CONCLUSION .............................................................................................. 26
`CERTIFICATE OF WORD COUNT ........................................................................ 1
`CERTIFICATE OF SERVICE .................................................................................. 2
`
`
`
`i
`
`
`
`IPR2016-01179
`U.S. Patent No. 5,850,482
`
`
`TABLE OF AUTHORITIES
`
`
`
`Page(s)
`
`Cases
`
`Apple, Inc., Requester v. ClassCo Inc., Patent Owner,
`Appeal No. 2015-000186, 2015 WL 1871500 (PTAB Apr. 21, 2015) ......... 15, 25
`ClassCo, Inc. v. Apple, Inc.,
`838 F.3d 1214 (Fed. Cir. 2016) ..................................................................... 16, 25
`Pers. Web Techs., LLC v. Apple, Inc.,
`848 F.3d 987 (Fed. Cir. 2017) .............................................................................. 14
`
`
`
`ii
`
`
`
`LIST OF PREVIOUSLY FILED EXHIBITS
`
`IPR2016-01179
`U.S. Patent No. 5,850,482
`
`
`Exhibits 1001-1013: Filed and served June 16, 2016 with Petitioner’s Petition for
`
`Inter Parties Review of U.S. Patent No. 5,850,482 (Dkt. 3).
`
`Exhibit 1014: Filed and served November 16, 2016 with Petitioner’s Updated
`
`Mandatory Notice Under 37 C.F.R. § 42.8(a)(3) (Dkt. 14).
`
`Exhibit 1015: Filed and served February 1, 2017 with the Joint Motion To
`
`Terminate Proceedings As To Microsoft Mobile Inc., And Microsoft Corporation
`
`(Dkt. 17).
`
`OTHER CITED EXHIBITS
`
`Exhibit 2006: Filed and served March 23, 2017 with FastVDO, LLC’s Response
`
`(Dkt. 22).
`
`LIST OF NEWLY-FILED EXHIBITS
`
`Exhibits 1016-1017: Concurrently filed with this Petitioners’ Reply:
`
`No.
`
`1016
`
`1017
`
`Description
`
`Declaration of Kenneth A. Zeger, Ph.D. in FastVDO LLC v.
`AT&T Mobility LLC et al., Case No. 3:16-cv-00385 (Dkt 157-1).
`Transcript of June 2, 2017 Deposition of Kenneth A. Zeger, Ph.D.
`
`iii
`
`
`
`I.
`
`INTRODUCTION
`
`IPR2016-01179
`U.S. Patent No. 5,850,482
`
`
`FastVDO’s Response is more notable for what it does not say than for what
`
`it does say. FastVDO does not dispute that the primary Kato reference teaches or
`
`suggests the vast majority of claim elements. Rather, FastVDO focuses on
`
`purported shortcomings in Petitioners’ presentation of the rationale for combining
`
`Kato’s embodiments with one another and with the Wei reference. As even
`
`FastVDO’s expert admitted at his deposition, however, the facts do not necessarily
`
`bear out FastVDO’s assertions. The challenged claims are obvious in view of
`
`Petitioners’ proposed combinations.
`
`II.
`
`FASTVDO’S SOLE CHALLENGE TO THE
`OBVIOUSNESS OF MOST CLAIMS IS TO CHALLENGE THE
`COMBINATION OF KATO’S COMPLEMENTARY EMBODIMENTS
`A. Kato Provides A Reason To
`Combine Embodiments – Adding Unequal
`Error Protection To Prevent Error Propagation
`
`After spending the bulk of its Response discussing data storage elements
`
`found in only three claims (see Section III) FastVDO on page 34 for the first time
`
`presents its sole global non-obviousness argument. The central thrust of
`
`FastVDO’s argument is encapsulated in its statement that “the Petition and
`
`Dr. Stevenson propose to incorporate Kato’s fourth embodiment into the first
`
`embodiment in order to provide a benefit that Kato’s first embodiment already
`
`possesses.” Response at 35. As FastVDO’s expert Dr. Zeger later admitted during
`
`1
`
`
`
`IPR2016-01179
`U.S. Patent No. 5,850,482
`
`his deposition, however, and as the Petition demonstrated, Kato’s first embodiment
`
`alone does not already provide the benefit added by the fourth embodiment.
`
`Kato’s first embodiment has code words comprising two parts, CJi’s and
`
`CEi’s. Petition at 24. The first portion CJi is encoded from Ji, called the “category
`
`index.” Among other things, category index Ji determines the word length for the
`
`code word’s second portion, CEi, encoded from the “remainder data” Ei. Id. at 26-
`
`27.
`
`One benefit of this split field coding approach is that it exhibits resilience to
`
`certain minor channel errors. In fact, similar to the ’482 system, errors in decoding
`
`Kato’s second code word portions – the CEi’s – do not propagate. Id. at 40. Not so
`
`for the first code word portions. Errors in decoding the CJi’s can propagate. And, if
`
`those errors lead to miscalculating the word length of the corresponding second
`
`portion of the code word, they can affect all subsequent code words. This is likely
`
`to lead to what the ’482 refers to as, and what FastVDO’s own expert admitted
`
`may be “catastrophic” synchronization errors. Id.; see also Ex. 1017, 111:1-
`
`112:25; Ex. 1001, 6:17-21. This is precisely why Kato later teaches the importance
`
`of providing protection against such “word length” errors. Petition at 40.
`
`So, FastVDO’s assertion – that Petitioners incorporate Kato’s fourth
`
`embodiment “in order to provide a benefit that Kato’s first embodiment already
`
`possesses” – is simply false. Kato’s fourth embodiment teaches providing higher
`
`2
`
`
`
`error protection to first code word portions Pi that determine the word length of
`
`IPR2016-01179
`U.S. Patent No. 5,850,482
`
`
`second code word portions Ri, and providing less or no protection to the Ri’s,
`
`where errors cannot propagate. Petition at 28-31. Kato’s first embodiment alone
`
`provides no such added protection to its first code word portions CJi, in which (like
`
`the fourth embodiment’s Pi’s) decoding errors can result in catastrophic
`
`synchronization errors. Id. at 30-31, 40, 47. Even FastVDO’s expert Dr. Zeger
`
`was forced to admit this during deposition.
`
`Dr. Zeger described Kato’s discussion in columns 15 and 16 of Kato’s first
`
`embodiment as offering “robust correction of propagation of errors,” where there
`
`are small errors in decoded values for immediately preceding input data (Di–1).
`
`See Ex. 1012, 15:54-16:12. But, he admitted that this embodiment does not protect
`
`against misdecoded Ji “word length” errors that could result in code word
`
`synchronization loss:
`
`Q. So you would agree that there’s a difference between the
`potential effects in this [first] embodiment of Kato in a Ji error and an
`Ei error on error propagation; correct?
`A. I think the way I would say it is that different types of errors
`in Kato can result in either propagation of errors or naturally robust
`correction of propagation of errors.
`Q. And is there a naturally robust correction of propagation
`of errors for the Ji example we’ve been discussing, where a
`subsequent Ji leads to an improper code word length?
`
`3
`
`
`
`IPR2016-01179
`U.S. Patent No. 5,850,482
`
`
`A. That’s an example of an error that could propagate in Kato
`that’s not addressed by the Column 15 and Column 16 description
`of how Kato naturally prevents certain error propagations.
`Q. And a similar error in Ei’s, this is not – there’s not an
`example provided here of where a subsequent Ei error would
`propagate; is that right?
`A. There’s no example in Kato of Ei errors that would
`propagate.
`Q. Correct. Said another way, Ei’s are resilient to errors; they
`won’t propagate?
`A. Well, as I said before, I wouldn’t ascribe error resilience to a
`particular set of bits; it’s to the system as a whole.
`Q. Okay. But the system as a whole provides that Ei errors
`don’t propagate?
`A. That’s true.
`Q. And the system, as we’ve discussed it, there is some
`propagation of errors with Ji’s?
`A. As I said before, some of the Ji error propagation is not
`naturally corrected as described in Column 15 and 16.
`Q. And one of ordinary skill in the art reading Kato would
`have understood that this error propagation that results from Ji
`errors could lead to catastrophic effects; correct?
`MR. HELGE: Object to form.
`A. Well, a person of skill in the art would understand that loss
`of synchronization or propagation of errors in general could lead to
`catastrophic errors, as we discussed before.
`
`4
`
`
`
`IPR2016-01179
`U.S. Patent No. 5,850,482
`
`
`Q. And that errors that are limited in the way that the Ei errors
`are limited would not lead to such effects; correct?
`A. Well, any type of errors that don’t propagate would probably
`not cause catastrophic errors, and among those would be the -- what
`you call the Ei errors.
`
`Ex. 1017, 111:1-112:25 (emphases added).
`
`Thus, as the Petition stated (e.g., at 30-31, 40, 47), it would have been
`
`obvious to add Kato’s unequal error protection scheme from its fourth embodiment
`
`to provide exactly the same benefit to the first embodiment: namely, providing
`
`protection against errors in the Ji’s that are likely to propagate, a benefit not
`
`provided in Kato’s first embodiment. Dr. Zeger acknowledged this:
`
`Q. Okay. The error protection that’s provided that we
`discussed in 15 and 16, does it do any good if the Ji’s are wrong?
`A. It’s not addressed to the Ji; so those would be examples
`where you would get propagation.
`Q. And those are exactly the examples that the fourth
`embodiment provides additional error protection for; correct?
`A. That -- that is basically correct.
`
`Ex. 1017, 134:17-25 (emphasis added).
`
`Dr. Zeger later sought to walk back his admissions about error propagation
`
`in Kato’s first embodiment. He testified that in earlier admitting that Ji errors
`
`would propagate, he was assuming that Kato’s first embodiment code words were
`
`5
`
`
`
`concatenated, and that without such concatenation, even an error decoding Ji that
`
`IPR2016-01179
`U.S. Patent No. 5,850,482
`
`
`resulted in an improper word length Mi would not necessarily propagate due to
`
`loss of synchronization, or what Dr. Zeger called “boundary errors.” See Ex. 1017,
`
`134:21-150:25. He concluded this portion of testimony stating: “[Kato] certainly
`
`doesn’t give a scenario where there’s going to be error propagation due to
`
`boundaries having problems.” Id., 150:23-25.
`
`Dr. Zeger’s new assumptions about Kato – first, that Kato’s first
`
`embodiment does not teach or suggest concatenated code words, and second, that
`
`Ji errors that affect word length Mi of the variable-length CEi’s would not lead to
`
`loss of synchronization or “boundary errors” for subsequent code words – are
`
`incorrect. Kato’s discussion of its first embodiment decoder makes this clear:
`
`The buffer memory 119 receives the word length Mi from the
`sub decoding circuit 121, and calculates the position of the starting
`edge of the coded data CJi following the coded data CEi on the basis
`of the received word length Mi.
`
`Ex. 1002, 12:58-62 (emphasis added).
`
`First, Kato’s first embodiment code words encoded in the manner discussed
`
`above do follow one another. The next code word’s first portion CJi follows the
`
`previous code word’s second portion CEi. Id.
`
`6
`
`
`
`Second, regardless of whether these code words are “concatenated
`
`IPR2016-01179
`U.S. Patent No. 5,850,482
`
`
`together,” because the word length Mi of a first code word’s remainder data is
`
`used to determine the starting edge of the next code word to be decoded, errors in
`
`calculating that word length Mi – such as by incorrectly decoding Ji – would
`
`necessarily create the very boundary problems Dr. Zeger admitted can result in
`
`synchronization errors and “catastrophic” error propagation:
`
`The decoding apparatus of FIG. 1(b) operates as follows. The
`coded data Ci fed via the input terminal 112 is temporarily stored in
`the buffer memory 119 in the main decoding circuit 113. The sub
`decoding circuit 120 in the main decoding circuit 113 reads out the
`coded data CJi from the buffer memory 119 and decodes the coded
`data CJi into the category index Ji …
`The sub decoding circuit 121 in the main decoding circuit 113
`receives the category index Ji from the sub decoding circuit 120, and
`calculates the code length Mi of the remainder data Ei from the
`category index Ji by referring to Table 1. The sub decoding circuit
`121 reads out the coded data CEi from the buffer memory 119 which
`has a code length corresponding to Mi bits …
`The buffer memory 119 receives the word length Mi from the
`sub decoding circuit 121, and calculates the position of the starting
`edge of the coded data CJi following the coded data CEi on the basis
`of the received word length Mi. The buffer memory 119 updates the
`reading pointer in accordance with the calculated position of the
`
`7
`
`
`
`starting edge of the coded data CJi to make preparations for the
`decoding of subsequent data.
`
`IPR2016-01179
`U.S. Patent No. 5,850,482
`
`
`Ex. 1002, 12:35-65 (emphases added).
`
`Simply put, if a word length Mi is incorrectly calculated due to a
`
`misdecoded Ji, the starting position for decoding the next code word will be
`
`incorrect. This is, by definition, a synchronization error, and one that will
`
`propagate, likely to catastrophic effect. In Kato’s decoder, described above, with a
`
`bad Mi, the subsequent code word’s CJi will be read from the wrong memory
`
`location. Thus, Ji will likely be incorrect, as will the word length Mi determined
`
`therefrom. As a result, the remainder data Ei and the starting point for decoding the
`
`next word’s CJi will also likely be wrong, and so on, and so on.
`
`As the Petition states, this is precisely the type of propagation of errors that
`
`the addition of error correction codes to first code word portions, such as CJi, that
`
`“enable the determination of the word length of a variable-length code word” is
`
`meant to remedy. Petition at 30-31, quoting Ex. 1002 at 32:14-20.
`
`Dr. Zeger acknowledged that it was known prior to the ’482 that this
`
`“catastrophic” propagation of errors could potentially cause “big problems”:
`
`Q. But this was a known problem prior to the ’482; namely, the
`propagation of errors can lead to catastrophic effects; correct?
`MR. HELGE: Object to form.
`
`8
`
`
`
`A. It was certainly known before the ’482 that loss of
`synchronization can cause big problems.
`
`IPR2016-01179
`U.S. Patent No. 5,850,482
`
`
`Ex. 1017, 41:8-14. He also acknowledged that if a POSA’s goal was to avoid such
`
`catastrophic effects due to such an error, a POSA would have understood that a
`
`potential solution was to error protect that field: “I do agree with you that, if one
`
`goal is to prevent catastrophic errors, then one should – one possibility is to protect
`
`the prefix field.” Id., 58:19-22.
`
`So FastVDO’s Response argument – that “the Petitioners’ alleged reason for
`
`incorporating Kato’s fourth embodiment into its first embodiment is to provide a
`
`benefit that the first embodiment already possesses” – is clearly wrong. Depending
`
`as it does on this false premise, the balance of FastVDO’s argument against
`
`combining Kato’s embodiments falls apart, too.1
`
`
`
`
`
`
`1 Page 43 of FastVDO’s Response purports to quote a position from page 37 of the
`
`Petition that is nowhere argued in the Petition, so FastVDO’s responsive argument
`
`against “relying on common sense in lieu of reasoned analysis” is a red herring.
`
`9
`
`
`
`In Combining Kato’s Embodiments,
`There Is A Reasonable Expectation Of Success
`
`IPR2016-01179
`U.S. Patent No. 5,850,482
`
`
`B.
`
`FastVDO next faults the Petition’s assertion (based on Dr. Stevenson’s
`
`testimony) that Kato’s third and fourth embodiments – teaching transforming and
`
`quantizing, and unequal error protection, respectively – could each be added to the
`
`first embodiment’s source coding apparatus to arrive at the claimed invention
`
`“without the need for undue experimentation, yielding predictable results.”
`
`Response at 44-45. However, this challenge, too, offers less than meets the eye.
`
`First, FastVDO offers no actual substantive challenge to combining the third
`
`embodiment with the first. There is no suggestion from its expert that such a
`
`combination would be beyond the skill of a POSA, and indeed no argument from
`
`its expert on this combination of embodiments whatsoever. Therefore, this
`
`challenge is not a challenge at all.
`
`Second, as to the combination of the first and fourth embodiments, the
`
`Petition described adding to the first embodiment two closely related aspects of the
`
`fourth embodiment, namely: a) storing data in a data storage format that separates
`
`first code word portions that are not resilient to errors from second code word
`
`portions that are, and b) providing higher error protection to the first code word
`
`portions than to the error-resilient second code word portions. Petition at 28-31,
`
`47-48, 50-52, 55, 57-59. It is this obvious addition that Petitioners contend could
`
`10
`
`
`
`be performed without undue experimentation. Id. at 40. In describing the
`
`IPR2016-01179
`U.S. Patent No. 5,850,482
`
`
`respective functions of Kato’s first and fourth embodiments, and in describing the
`
`expertise that a POSA would have, Dr. Stevenson’s supporting declaration
`
`provides ample evidence as to why a combination of these embodiments could be
`
`performed without undue experimentation, yielding predictable results. Ex. 1005,
`
`¶¶ 50-68, 80-88.
`
`Significantly, the ’482 itself describes and claims the application of “unequal
`
`error protection means as known to those skilled in the art.” And, such means
`
`“suitable for use with the … present invention” had been “known” since at least
`
`1968, as Dr. Zeger admitted:
`
`Q. And so you would agree that unequal error protection, which
`the patent refers to as suitable for use with the method and apparatus
`of the present invention, was known at least as early as 1968; correct?
`A. I think that’s accurate.
`
`Ex. 1017, 41:8-14.
`
`In arguing against the predictable nature of adding unequal error protection
`
`to Kato’s split field coded data, FastVDO’s response somewhat confusingly points
`
`to Dr. Stevenson’s deposition discussion of his and his colleagues’ work on “this
`
`source coding technique that allows you to do unequal error protection in the
`
`channel coding.” Ex. 2006, 101:9-11 (emphasis added). FastVDO argues that this
`
`work suggests that adding unequal error protection to Kato’s source coder would
`
`11
`
`
`
`have required “undue experimentation.” However, nothing in that discussion, or
`
`IPR2016-01179
`U.S. Patent No. 5,850,482
`
`
`in the referenced research, is cited as relating to actually adding “unequal error
`
`protection” (the very element Petitioners and Dr. Stevenson propose adding from
`
`the fourth embodiment), let alone adding it to Kato’s source codes.
`
`In fact, when asked, Dr. Stevenson stated that “I remember at some point we
`
`– you know, my kind of vague recollection, I remember discussing with this group
`
`unequal error protection, I’m not sure we ever did it though.” Id., 107:3-7. So,
`
`FastVDO’s Response relies on research that didn’t focus on adding unequal error
`
`protection as providing insight into supposed (but unexplained) difficulties with
`
`adding unequal error protection, which the ’482 itself acknowledges was nearly 30
`
`years old. Such reliance is misplaced.
`
`The remainder of FastVDO’s argument on this point appears to suggest that
`
`because there was ongoing research in this field at that time – including by
`
`Dr. Stevenson and colleagues – no incorporation of the admittedly old idea of
`
`applying unequal error protection could have been performed without undue
`
`experimentation. FastVDO offers no legal support for this proposition, and
`
`Petitioners are aware of none.
`
`And, indeed, if applying unequal error protection to split field code word
`
`portions such as those in the ’482 and in Kato was not within the skill of a POSA
`
`12
`
`
`
`outside of the ’482 disclosure, the ’482 provides nothing more that would enable
`
`IPR2016-01179
`U.S. Patent No. 5,850,482
`
`
`one to perform it. Rather, as the Response (at 7) describes, the ’482 merely offers:
`
`[T]he method and apparatus of the present invention is suitable
`for use with unequal error protection means as known to those
`skilled in the art… Specifically, the prefix fields of the encoded data
`are preferably channel encoded with an appropriately high level of
`error protection in order to provide a high probability that the prefix
`fields will be decoded correctly. Because the associated suffix fields
`are error resilient, however, the suffix fields may be channel
`encoded with a lower level of error protection or may not be channel
`encoded, thereby providing no error protection.
`
`Ex. 1001, 16:10-22 (emphasis added); see also id., 17:1-4.
`
`Thus, the ’482 itself acknowledges that “unequal error protection means as
`
`known to those skilled in the art” are suitable for use with the claimed invention.
`
`And, as it also states, those unequal error protection methods are suitable for use
`
`because while higher protection is needed for its prefix fields, “the associated
`
`suffix fields are error resilient.”
`
`As described in more detail above, the same is true of Kato. Kato’s second
`
`code word portions – CEi’s – are similarly error-resilient, and a POSA would have
`
`had the same reasonable expectation of success in applying similar well-known
`
`unequal error protection methods such as “add[ing] error correction codes of
`
`higher ability to error-propagatable data portions then the ability of error correction
`
`13
`
`
`
`codes added to other data portions” described in Kato’s fourth embodiment
`
`IPR2016-01179
`U.S. Patent No. 5,850,482
`
`
`(Ex. 1002, 31:60-62), as he would employing the “known unequal error protection
`
`means” cited in the ’482. Again, Kato describes this combination of features in at
`
`least as much detail as the challenged claims.
`
`The remainder of FastVDO’s challenge essentially boils down to an
`
`assertion that Petitioners failed to lay out a detailed “bodily incorporation” of
`
`Kato’s fourth embodiment into its first embodiment. Even FastVDO’s cited case
`
`makes clear that this is not necessary, though, particularly when adding simple
`
`well-understood prior art solutions (here, error correction codes): “The amount of
`
`explanation needed to meet the governing legal standards…necessarily depends on
`
`context. A brief explanation may do all that is needed if, for example, the
`
`technology is simple and familiar and the prior art is clear in its language and
`
`easily understood.” Pers. Web Techs., LLC v. Apple, Inc., 848 F.3d 987, 994 (Fed.
`
`Cir. 2017) (emphasis added).
`
`The structures of two prior art references (let alone embodiments of the
`
`same reference) need not be “physically combinable” to render a claim obvious.
`
`FastVDO’s non-obviousness challenge based on failure to “physically combine”
`
`Kato’s embodiments is similar to patent owner’s failed PTAB non-obviousness
`
`14
`
`
`
`challenge in Apple, Inc., Requester v. ClassCo Inc., Patent Owner.2 As the Board
`
`IPR2016-01179
`U.S. Patent No. 5,850,482
`
`
`there noted:
`
`Patent Owner appears to argue that it would not have been obvious to
`one of ordinary skill in the art to have bodily incorporated Gulick into
`Fujioka. We are not persuaded by Patent Owner’s argument at least
`because “[t]he test for obviousness is not whether the features of a
`secondary reference may be bodily incorporated into the structure of
`the primary reference.... Rather, the test is what the combined
`teachings of those references would have suggested to those of
`ordinary skill in the art.” (“[I]t is not necessary that the inventions
`of the references be physically combinable to render obvious the
`invention under review.”); and In re Nievelt, 482 F.2d 965, 968
`(CCPA 1973) (“Combining the teachings of references does not
`involve an ability to combine their specific structures.”).
`
`Id.
`
`The Federal Circuit agreed, stating:
`
`KSR does not require that a combination only unite old elements
`without changing their respective functions. Instead, KSR teaches that
`“[a] person of ordinary skill is also a person of ordinary creativity, not
`an automaton.” And it explains that the ordinary artisan recognizes
`“that familiar items may have obvious uses beyond their primary
`purposes, and in many cases a person of ordinary skill will be able to
`
`
`2 Appeal No. 2015-000186, 2015 WL 1871500, at *4 (PTAB Apr. 21, 2015).
`
`15
`
`
`
`IPR2016-01179
`U.S. Patent No. 5,850,482
`
`
`fit the teachings of multiple patents together like pieces of a puzzle.”
`The rationale of KSR does not support ClassCo’s theory that a person
`of ordinary skill can only perform combinations of a puzzle element A
`with a perfectly fitting puzzle element B. To the contrary, KSR
`instructs that the obviousness inquiry requires a flexible approach.
`Here, the Board faithfully applied this flexible approach to find that
`the [prior art] combination “would have resulted in no more than [a]
`predictable result.”
`
`ClassCo, Inc. v. Apple, Inc., 838 F.3d 1214, 1219 (Fed. Cir. 2016) (citations
`
`omitted).
`
`Thus – as did patent owner’s challenge in ClassCo – FastVDO’s challenge
`
`to the obvious combination of Kato’s first and fourth embodiments must fail.
`
`III. FASTVDO ALLEGES ONLY ONE CLAIM ELEMENT IS “NOT
`DISCLOSED OR SUGGESTED,” AN ELEMENT THAT IS OBVIOUS
`IN VIEW OF KATO’S ERROR-RESILIENT DATA STORAGE
`FastVDO does not expressly set forth any supposed deficiencies with
`
`Petitioners’ proposed claim constructions, but instead “disagrees with [them] to the
`
`extent they differ from the claim constructions reached by the Board in the
`
`Institution Decision (Paper 15) and from the claim constructions reached by the
`
`District Court in co-pending litigation. See generally Ex. 1014.” Response at 12.
`
`While otherwise deferring to these decision constructions, though, FastVDO’s
`
`Response for the first time introduces a new construction for “data storage
`
`16
`
`
`
`medium.” FastVDO never included this construction in its Preliminary Response.
`
`IPR2016-01179
`U.S. Patent No. 5,850,482
`
`
`Nor did FastVDO or its expert Dr. Zeger offer a construction for this phrase in the
`
`underlying litigation in which the Claim Construction Order cited by FastVDO
`
`issued. See Ex. 1016. Thus, the Board should resist FastVDO’s invitation to
`
`narrow a claim term for which FastVDO and its expert have never previously
`
`argued a construction.
`
`Regardless of if or how this term is construed, as discussed above, Kato
`
`unquestionably teaches storing data that is resilient to errors separate from data that
`
`is not resilient to errors. And, as in the ’482, the error protection provided to the
`
`error-resilient data portions is lower than the error protection provided to the
`
`separately stored portions where an error can propagate through. The only
`
`difference FastVDO’s Response cites between the data storage described in Kato
`
`and the data storage described in the ’482 is that in Kato, a higher level of error
`
`protection is provided to data portions that were previously stored together, while
`
`in the ’482 these portions are not stored together until after error protection has
`
`been applied. This is nothing but an obvious variation, as Dr. Stevenson noted in
`
`his original declaration. E.g., Ex. 1005, ¶ 114.
`
`One of ordinary skill reading Kato would clearly understand which blocks of
`
`data in Kato’s data storage region receive the higher level of error protection based
`
`17
`
`
`
`on Kato’s extensive disclosure of its unequal error protection scheme. Dr. Zeger
`
`IPR2016-01179
`U.S. Patent No. 5,850,482
`
`
`acknowledged this:
`
`Q. If we’re calling the first portions Pi’s and the second
`portions Ri’s, do I know in advance that the Pi’s are going to get the
`higher level?
`A I think that’s the point of Claim 9, is that the Pi’s, after
`they’re outputted and then they go through the error control circuit, so
`after they’re out of memory, they would get the higher protection.
`
`Ex. 1017, 162:12-18; see also Petition at 36-37, citing Ex. 1002, Fig. 7, 24:66-
`
`25:8; 32:35-40; 35:18-21.
`
`Additionally, to the extent the claims require it, one of ordinary skill would
`
`understand that such data could again be stored after the unequal error protection is
`
`applied. In fact, it would be odd if after applying this unequal error protection the
`
`Kato data was never again stored. Nowhere does FastVDO even suggest this.
`
`Some (non-limiting) examples of such potential storage locations in Kato
`
`identified during Dr. Zeger’s deposition would be, e.g., a buffer storage in the
`
`modulator (prior to transmission) in Kato’s fourth embodiment, or, after applying
`
`unequal error protection to Kato’s first embodiment, the first embodiment’s buffer
`
`situated in the decoder from which the code word portions are read out prior to
`
`decoding.
`
`18
`
`
`
`FastVDO contends that pointing to examples of storage other than the RAM
`
`IPR2016-01179
`U.S. Patent No. 5,850,482
`
`
`is introducing a new theory. To the contrary, these examples are merely further
`
`evidence in support of Dr. Stevenson’s initial assertion, cited in the Petition, that
`
`“in view of Kato’s disclosure of providing unequal error protection to data Pi and
`
`Ri, it would have been obvious to provide unequal error protection to the separate
`
`data store regions (or data blocks) in which they are stored.” Id. Dr. Stevenson
`
`previously noted that RAM was only an “example” of storage, and discussed
`
`additional exemplary possibilities for data storage during redirect at his deposition
`
`in response to FastVDO’s counsel’s cross-examination questioning about the data
`
`storage limitation. Ex. 2006, 137:7-139:1.
`
`Additionally, the ’482’s disclosure itself suggests that unequally error
`
`protected data is stored similarly to Kato’s data storage. In both cases, the data is
`
`separated into a block of first code word portions, which are to receive higher error
`
`protection, and a separate block of second code word portions which are to receive
`
`less or no error protection. So, if one of ordinary skill desired to provide protection
`
`for channel errors in the storage medium (as opposed to the transmission medium)
`
`– a need the ’482 acknowledges was known in the prior art (Ex. 1001, 5:1-5) – the
`
`’482’s claimed error-protected data storage is no more than an obvious variation on
`
`similar data storage described and shown in Kato. See Ex. 1005, ¶ 114.
`
`19
`
`
`
`A.
`
`Storage In Kato’s Fourth Embodiment
`
`IPR2016-01179
`U.S. Patent No. 5,850,482
`
`
`FastVDO admits that after ECCs are added to the Pi’s that are output from
`
`the encoder in Kato’s fourth embodiment, the “ECC-encoded data is modulated
`
`and transmitted out.” Response at 24.
`
`At his deposition, Dr. Zeger admitted that a modem is an example of a
`
`modulator, and that in modems “there often are buffers all over the place.”
`
`Ex. 1017, 159:17-24, 160:13-14; see generally id., 160:8-25. So, storing these
`
`unequally error-protected data blocks of Pi’s and Ri’s in modem storage prior to
`
`transmission is at least implicitly suggested in Kato, which already makes clear
`
`that its data storage regions are not limited to storage on a RAM or a magnetic d