`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
` UNITED STATES PATENT AND TRADEMARK OFFICE
`
`
`
`
`
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`
`
`
`
`
`NETAPP, INC.,
`Petitioner,
`
`v.
`
`REALTIME DATA LLC,
`Patent Owner.
`____________________
`
`Case IPR2017-01660
`Patent No. 7,161,506
`____________________
`
`
`
`
`PATENT OWNER’S PRELIMINARY RESPONSE
`
`
`
`
`
`
`
`
`V.
`
`
`Case IPR2017-01660
`PATENT OWNER’S PRELIMINARY RESPONSE
`TABLE OF CONTENTS
`Introduction ..................................................................................................... 1
`I.
`Background of the ’506 Patent and challenged claim .................................... 4
`II.
`III. Petitioner’s proposed claim constructions ...................................................... 4
`IV. Given Petitioner’s delay in seeking review, the Board should decline
`to institute a third review of the ’506 Patent .................................................. 5
`The Petition improperly presents four redundant Grounds without
`distinguishing between them (Grounds 1-4) ................................................ 10
`VI. Petitioner’s Ground 1 theory—Hsu in view of Franaszek—does not
`teach “a default data compression encoder” as required by limitation E ..... 12
`A. As the Petition acknowledges, Hsu does not teach limitation E. ....... 13
`Franaszek does not teach limitation E because it always applies
`B.
`content dependent compression, if it compresses at all. .................... 18
`Regardless of whether a data type is provided, the system of
`1.
`Franaszek always analyzes data blocks to determine their
`compressibility and always designates an encoder to apply
`based on that measured compressibility. ..................................18
`Designating a compression algorithm based on the results of
`measuring a data block’s compressibility is content dependent
`data compression, not using “a default data compression
`encoder.” ...................................................................................24
`Because neither Hsu nor Franaszek teaches a “default data
`compression encoder” as required by limitation E, and there is
`no evidence that a POSA would arrive at a “default encoder”
`based on those references, no combination of the references can
`prevail. ................................................................................................ 29
`VII. Petitioner’s Ground 2 theory—Hsu in view of Sebastian—likewise
`fails ................................................................................................................ 30
`A.
`The Petition’s only allegations regarding a motivation to
`combine are conclusory, rely on motivations to combine other
`art, and improperly incorporate by reference from the
`declaration. ......................................................................................... 30
`
`
`- i -
`
`
`
`
`
`2.
`
`C.
`
`
`
`B.
`
` Case IPR2017-01660
`
`PATENT OWNER’S PRELIMINARY RESPONSE
`The only competent evidence of record, provided by two
`experts, establishes that Sebastian does not teach or suggest
`what Petitioner ascribes to it. ............................................................. 34
`VIII. Petitioner’s Ground 3 theory—Franaszek in view of Hsu—also fails
`to present a prima facie showing that the combination would meet
`limitation E ................................................................................................... 39
`IX. Petitioner’s Ground 4—Franaszek in view of Chu—is similarly
`deficient ........................................................................................................ 42
`Conclusion .................................................................................................... 44
`
`
`
`X.
`
`
`
`
`- ii -
`
`
`
`
`
`Exhibit No.
`2001
`
`2002
`
`2003
`
`2004
`
`2005
`
`2006
`
`2007
`
`2008
`
`2009
`
`2010
`
`2011
`
` Case IPR2017-01660
`
`PATENT OWNER’S PRELIMINARY RESPONSE
`EXHIBIT LIST
`
`
`Description
`Declaration of Kayvan B. Noroozi in Support of Motion for
`Admission Pro Hac Vice
`Business Confidential Settlement Agreement between
`Realtime Data and Rackspace US (filed Board Only)
`NOT USED
`
`NOT USED
`Transcript of Oral Deposition of Charles D. Creusere, August
`4, 2017, taken in IPR2017-00176 and IPR2017-00179
`NetApp’s Invalidity Contentions, served in Case No. 6:16-CV-
`961 (E.D. Tex.) on January 24, 2017
`NetApp’s Motion to Stay, filed in Case No. 6:16-CV-961
`(E.D. Tex.) on November 11, 2016
`NetApp’s Reply in Support of its Motion to Stay, filed in Case
`No. 6:16-CV-961 (E.D. Tex.) on December 8, 2016
`Declaration of Kenneth A. Zeger, as submitted in Dell Inc. et
`al v. Realtime Data LLC, IPR2017-00176, Ex. 2004
`Declaration of Kenneth A. Zeger, as submitted in Oracle
`America, Inc. v. Realtime Data LLC, IPR2016-00373, Ex.
`2022
`Declaration of Kenneth A. Zeger, as submitted in Teradata v.
`Realtime Data LLC, IPR2017-00557, Ex. 2003
`
`
`
`
`
`
`
`
`- iii -
`
`
`
`
`
`
`Case IPR2017-01660
`PATENT OWNER’S PRELIMINARY RESPONSE
`
`I.
`
`Introduction
`NetApp asks the Board to institute inter partes review of the same claims
`
`already under review in two pending petitions, and based on the same or
`
`substantially the same prior art and arguments. Section IV, infra. The Board has
`
`recently rejected NetApp’s attempts to burden the Board and Realtime with
`
`duplicative and tardy petitions that provide no potential benefit, and should do so
`
`again here. Id. There is no practical reason to institute a third, redundant petition.
`
`NetApp’s Petition is also internally redundant. The Petition presents four
`
`grounds, but does not distinguish between them or ascertain their strengths and
`
`weaknesses, as the Board has required. Section V, infra. Consistent with its
`
`precedents, the Board should deny institution in full on that basis as well. Id.
`
`Moreover, all four grounds fail to make a prima facie showing of
`
`obviousness with respect to the same limitation of claim 105: limitation E.
`
`Ground 1 relies on the combination of Hsu with Franaszek. Petitioner admits
`
`that Hsu does not teach limitation E. Pet. 21. And the testimony of two expert
`
`witnesses, one of which was adverse to Realtime, establishes that Franaszek also
`
`does not teach the limitation—refuting the Petition’s allegations. Section VI.B,
`
`infra. Based on that record, the Petition does not establish a likelihood of
`
`prevailing as to Ground 1, and should not be instituted. Section VI.C, infra.
`
`
`
`
`
`
`- 1 -
`
`
`
`
`
` Case IPR2017-01660
`
`PATENT OWNER’S PRELIMINARY RESPONSE
`Ground 2, which relies on the combination of Hsu with Sebastian, falls
`
`further short of the mark. The Petition’s Ground 2 analysis attempts to shortcut its
`
`burden of properly articulating a motivation to combine the references, and instead
`
`alleges that it would have been obvious to combine Hsu with Sebastian for the
`
`same reasons alleged with respect to Hsu and Franaszek. Section VII.A, infra. The
`
`Federal Circuit has expressly rejected such attempts at conclusory cross-
`
`referencing, finding that approach to be fatally inadequate. Id. Indeed, Petitioner’s
`
`Ground 2 theory would require a person of ordinary skill in the art to be motivated
`
`to compress data blocks that Hsu expressly teaches should not be compressed—an
`
`extremely untenable motivation the Petition does not actually support. Id.
`
`Moreover, Ground 2 also rests on the assertion that a person of ordinary skill in the
`
`art would understand Sebastian to contain a teaching that, according to two experts,
`
`Sebastian does not actually teach, and that a person of ordinary skill would not
`
`derive from Sebastian. Section VII.B, infra. Ground 2 thus fails as well.
`
`Ground 3 merely duplicates Ground 1, but flips the order of the references,
`
`placing Franaszek as the base reference. Those gymnastics are to no avail. Since
`
`neither Franaszek nor Hsu teaches limitation E, and there is no evidence that a
`
`POSA (bereft of hindsight) would arrive at limitation E in view of either
`
`reference—whether alone or in combination—Ground 3 is as meritless as Ground
`
`1. Section VIII, infra.
`
`
`
`
`
`- 2 -
`
`
`
`
`
` Case IPR2017-01660
`
`PATENT OWNER’S PRELIMINARY RESPONSE
`Ground 4 is doomed to the same fate. Whereas Ground 4 rests on the
`
`premise that Franaszek teaches limitation E, the testimony of two experts (one of
`
`them adverse to Realtime) establishes that Franaszek does not. Section IX, infra. In
`
`any case, combining Franaszek with Chu—as Ground 4 proposes—would actually
`
`ensure that the resulting system would not make use of the very aspect of
`
`Franaszek that the Petition (wrongly) relies on as teaching limitation E. Chu’s data
`
`type recognition approach will always return a data type, and so the combined
`
`system of Franaszek and Chu would never utilize Franaszek’s teachings with
`
`respect to unrecognized data types. Id.
`
`The Board should not allow NetApp to further waste the Board’s resources,
`
`or Patent Owner’s, through this third, meritless and pointless petition. It should
`
`instead deny institution in full, as to all Grounds.
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`- 3 -
`
`
`
`
`
` Case IPR2017-01660
`
`PATENT OWNER’S PRELIMINARY RESPONSE
`II. Background of the ’506 Patent and challenged claim
`Petitioner challenges claim 105 of the ’506 Patent, which recites:
`
`48 [B]
`
`Element Claim 105
`48 [PR] A computer implemented method comprising:
`48 [A]
`receiving a data block in an uncompressed form, said data block being
`included in a data stream;
`analyzing data within the data block to determine a type of said data
`block;
`compressing said data block to provide a compressed data block;
`48 [C]
`48 [D] wherein if one or more encoders are associated to said type,
`compressing said data block with at least one of said one or more
`encoders; and
`otherwise compressing said data block with a default data compression
`encoder,
`wherein the analyzing of the data within the data block to identify
`one or more data types excludes analyzing based only on a
`descriptor that is indicative of the data type of the data within the
`data block.
`
`48 [E]
`
`48 [F]
`
`III. Petitioner’s proposed claim constructions
`Petitioner has proposed that the terms “analyzing” and “analyze” should
`
`mean “directly examining” and “directly examine.” Pet. 13. Petitioner has also
`
`proposed that the term “default data compression encoder” should mean “an
`
`encoder used automatically in the absence of a designated alternative.” Id.
`
`
`
`
`
`
`- 4 -
`
`
`
`
`
` Case IPR2017-01660
`
`PATENT OWNER’S PRELIMINARY RESPONSE
`The Board does not construe claim terms unnecessary to resolve the
`
`controversy. Shenzhen Liown Electronics Co. v. Disney Enterprises, Inc.,
`
`IPR2015-01656, Paper 7 at 10 (Feb. 8, 2016) (citing Vivid Techs., Inc. v. Am. Sci.
`
`& Eng’g, Inc., 200 F.3d 795, 803 (Fed. Cir. 1999)).
`
`For purposes of this Patent Owner Preliminary Response, those terms do not
`
`require an express construction. This Response assumes arguendo that Petitioner’s
`
`proposed constructions are correct, and nonetheless demonstrates that the cited
`
`references do not teach all of the limitations of the challenged claim.
`
`IV. Given Petitioner’s delay in seeking review, the Board should decline
`to institute a third review of the ’506 Patent
`The Board recently declined to institute two of NetApp’s Petitions, citing
`
`NetApp’s delay in filing its Petitions and the prejudice to Realtime of defending
`
`serial requests for inter partes review with respect to the same patents. See
`
`IPR2017-01195, Paper 9 (“’530 Decision”); IPR2017-01196, Paper 10 (“’908
`
`Decision”). In those decisions, the Board explained that the seven factors outlined
`
`in General Plastic Indus. Co. v. Canon Kabushiki Kaisha, IPR2016-01357, Paper
`
`19 (Sept. 6, 2017), “provide a useful framework for analyzing the facts and
`
`circumstances present in this case, in which a different petitioner filed a petition
`
`challenging a patent that had been challenged already by previous petitions.” See
`
`
`
`
`
`
`- 5 -
`
`
`
`
`
` Case IPR2017-01660
`
`PATENT OWNER’S PRELIMINARY RESPONSE
`’530 Decision at 10; IPR2017-01196, ’908 Decision at 9. The General Plastic
`
`factors similarly warrant non-institution here.
`
`Factors 1 and 2 relate to successive petitions filed by the same petitioner,
`
`which is not at issue here. Factor 1 weighs in favor of institution and Factor 2 is
`
`neutral. ’530 Decision at 10; ’908 Decision at 9.
`
`Factors 3 and 4 consider Petitioner’s delay between learning of the art and
`
`filing the petition, and whether delay enabled Petitioner to “tailor its arguments to
`
`address issues identified” in prior proceedings. Id. at 11-12 & n.12.
`
`Those factors weigh strongly against institution. NetApp filed this Petition
`
`on June 22, 2017, nearly a year after it was sued. Yet it knew about the art and its
`
`specific grounds many months before. The first petition against the ’506 Patent
`
`was filed on November 14, 2016, and the Board instituted review on May 30,
`
`2017. See IPR2017-00176 (“Dell petition”), Papers 1 (Petition) and 19 (Institution
`
`Decision). And in January 2017, a second challenge was filed, which the Board
`
`later instituted and then consolidated with the Dell petition. IPR2017-00806,
`
`Papers 1 (“Teradata petition,” filed Jan. 30, 2017), 17 (Institution Decision), and
`
`19 (Order of Consolidation).
`
`The claim challenged here was challenged in both previous petitions. And
`
`the Petition presents four grounds and four references that are almost entirely
`
`duplicative of prior petitions. Ground 1 (Hsu and Franaszek) is nearly identical to
`
`
`
`
`- 6 -
`
`
`
`
` Case IPR2017-01660
`
`PATENT OWNER’S PRELIMINARY RESPONSE
`Ground 1 of the Dell and Terada petitions, which presented the same art but in the
`
`opposite order (i.e., Franaszek in view of Hsu). Ground 2 (Hsu and Sebastian) also
`
`relies exclusively on art already presented in the Dell and Teradata petitions.
`
`Ground 3 (Franaszek and Hsu) simply flips the order of the references from
`
`Ground 1 and is identical to the combination in Ground 1 of the Dell and Teradata
`
`petitions. And Ground 4 (Franaszek and Chu) uses a different secondary reference
`
`from the preceding ground but identifies no distinct relevance to that reference,
`
`instead urging that the same claim is invalid “for the same reasons as those
`
`explained above” in Ground 3. Pet. 47-48.
`
`Three of the four references also appeared repeatedly in earlier petitions
`
`against related patents. See, e.g., IPR2016-00373 (filed Dec. 22, 2015, and relying
`
`on Hsu, Franaszek, and Sebastian); IPR2017-00108 (filed Oct. 24, 2016, and
`
`relying on Hsu and Sebastian). NetApp indisputably knew of those petitions (and
`
`references) by November 11, 2016, when it cited all of them in support of its
`
`Motion to Stay in the district court. Ex. 2007 at 4. And NetApp recognized their
`
`relevance to the ’506 Patent. Ex. 2008 at 2-3 (arguing that the IPRs filed against
`
`other patents were relevant to the ’506 Patent).
`
`By January 24, 2017, NetApp had served its invalidity contentions, relying
`
`on all four references cited in the Petition—Hsu, Sebastian, Franaszek, and Chu.
`
`Ex. 2006 at 11-12, 14, 27. NetApp could have also filed its Petition then.
`
`
`
`- 7 -
`
`
`
`
`
`
` Case IPR2017-01660
`
`PATENT OWNER’S PRELIMINARY RESPONSE
`But instead, NetApp sat back and tracked key events in the filed IPRs,
`
`including upcoming institution and final written decision dates. Ex. 2008 at 1
`
`(telling the court that “the PTAB will likely complete its review on these two
`
`patents by June 2017”).
`
`With all that knowledge, NetApp deliberately waited until June 2017 to file
`
`its Petition, picking and choosing art and arguments from prior petitions.
`
`Meanwhile, Realtime filed its Patent Owner Preliminary Responses to the Dell and
`
`Teradata petitions, as well as its Patent Owner Responses in other proceedings
`
`involving the same art with respect to related data compression patents. See, e.g.,
`
`IPR2016-00373, Paper 13 (Sep. 15, 2016); IPR2016-00980, Paper 39 (Feb. 8,
`
`2017). NetApp had the opportunity to see Realtime’s arguments, other petitioners’
`
`theories, and even other experts’ testimonies, and to craft its Petition accordingly.
`
`Factors 3 and 4 thus weigh strongly against institution. See ’530 Decision at 11-12;
`
`’908 Decision at 10-11.
`
`Factor 5, which relates to Petitioner’s explanation for any delay, is to the
`
`same effect. The Petition challenges the same claim as prior petitions, yet does not
`
`explain why NetApp delayed filing, or why it did not seek to join an earlier
`
`petition, as Veritas did. See, e.g., IPR2017-01688, Papers 1 and 3 (Veritas petition
`
`for inter partes review of the ’506 Patent and accompanying motion for joinder
`
`
`
`
`
`
`- 8 -
`
`
`
`
`
` Case IPR2017-01660
`
`PATENT OWNER’S PRELIMINARY RESPONSE
`with the Dell petition). Factor 5 weighs against institution. See ’530 Decision at 12;
`
`’908 Decision at 11-12.
`
`Factors 6 and 7 relate to the resources of the Board and the statutory time
`
`limits for a Final Written Decision. Both factors weigh against institution for the
`
`same reasons outlined in the ’530 Decision, at 12-13, and ’908 Decision, at 12-14.
`
`As in those proceedings, institution here would require the Board to waste its
`
`resources on a duplicative proceeding, without any offsetting gain to the Eastern
`
`District of Texas, since the parties are set to try the ’506 patent (including its
`
`validity) in January 2017. Id. And Realtime would also incur significant prejudice.
`
`Id. Realtime has already spent more than a year defending the patentability of the
`
`’506 patent in the earlier-filed proceedings, and NetApp has offered no reason why
`
`its decision to pursue a tardy and redundant petition should force Realtime to spend
`
`another year re-arguing issues that substantially overlap with those presented in
`
`earlier petitions. Id.
`
`The Board should therefore decline to institute a third Petition as to the ’506
`
`Patent in light of General Plastic, and consistent with its decisions as to the
`
`NetApp ’530 and ’908 petitions.
`
`Moreover, for similar reasons, the Board should also exercise its discretion
`
`to deny institution under 37 C.F.R. § 325(d). Section 325(d) empowers the Board
`
`to deny institution of a petition “if another proceeding or matter involving the
`
`
`
`- 9 -
`
`
`
`
`
`
` Case IPR2017-01660
`
`PATENT OWNER’S PRELIMINARY RESPONSE
`patent is before the Office,” particularly where “the same or substantially the same
`
`prior art or arguments previously were presented to the Office.” As explained
`
`above, the NetApp Petition substantially reurges the same art and arguments that
`
`the Board and Realtime have seen in prior petitions. Instituting the petition would
`
`simply lead to a duplicative, delayed, and unproductive proceeding as to art and
`
`arguments that are already being addressed (whether identically or in substance)
`
`through the two earlier, instituted petitions. There is no reason for the Board to
`
`waste its resources, or Realtime’s, by instituting a third petition here.
`
`V. The Petition improperly presents four redundant Grounds without
`distinguishing between them (Grounds 1-4)
`The Board has the authority to deny a Petition that presents undifferentiated
`
`and redundant grounds under its discretionary power to “secure the [j]ust, speedy
`
`and inexpensive resolution of every proceeding.” 37 C.F.R. § 42.1(b). Where a
`
`Petitioner presents multiple redundant grounds and makes no meaningful
`
`distinction between them, the Board has consistently declined to institute all of
`
`those proffered grounds. See, e.g., Shopkick Inc. v. Novitaz, Inc., IPR2015-00279,
`
`Paper 7 at 27-28 (PTAB May 29, 2015) (declining to institute review on various
`
`combinations because “Petitioner does not articulate meaningful distinctions
`
`between its obviousness argument[s]”); Hyundai Motor Co. v. American Vehicular
`
`Sciences LLC, IPR2014-00658, Paper 8 at 23 (PTAB Sep. 18, 2014) (declining to
`
`
`
`
`
`
`- 10 -
`
`
`
`
`
` Case IPR2017-01660
`
`PATENT OWNER’S PRELIMINARY RESPONSE
`institute on redundant grounds because “[a] proper lack of redundancy argument
`
`requires a two-way analysis between each pair of grounds at issue; simply
`
`identifying differences in the disclosures of the prior art references is
`
`insufficient.”); Liberty Mutual Insurance Company v. Progressive Casualty
`
`Insurance Company, CBM2012-00003, Paper 7 at 2 (PTAB Oct. 25, 2012)
`
`(“[M]ultiple grounds, which are presented in a redundant manner by a petitioner
`
`who makes no meaningful distinction between them, are contrary to the regulatory
`
`and statutory mandates, and therefore are not all entitled to consideration.”).
`
`In determining whether a Petition has presented undifferentiated and
`
`redundant grounds, the Board considers “whether Petitioner articulated meaningful
`
`distinctions between its patentability arguments by, for example, explaining the
`
`relative strengths and weaknesses of various prior art combinations with respect to
`
`one or more claim limitations.” Apple Inc. v. OpenTV, Inc., IPR2015-00980, Paper
`
`11 at 25-26 (PTAB Sep. 28, 2015).
`
`The Petition here never evaluates the relative strengths and weaknesses of its
`
`duplicative Grounds; it simply alleges that the same claim is equally invalid under
`
`each of four Grounds: Hsu in view of Franaszek (Ground 1), Hsu in view of
`
`Sebastian (Ground 2), Franaszek in view of Hsu (Ground 3), and Franaszek in
`
`view of Chu (Ground 4). Moreover, Grounds 2 and 4 do not present a stand-alone
`
`invalidity analysis, but instead expressly rely on the invalidity analysis from
`
`
`
`- 11 -
`
`
`
`
`
`
` Case IPR2017-01660
`
`PATENT OWNER’S PRELIMINARY RESPONSE
`Grounds 1 and 3, respectively. See Pet. 40 (“For the same reasons as those
`
`explained above with respect to Franaszek . . . claim 105 is thus obvious in view of
`
`Hsu and Sebastian.”); id. at 47-48 (“For the same reasons as those explained above
`
`with respect to Hsu . . . claim 105 is obvious in view of Franaszek and Chu.”). In
`
`other words, the Petition not only fails to distinguish Grounds 2 and 4 from
`
`Grounds 1 and 3, it actually alleges that Grounds 2 and 4 are basically identical to
`
`Grounds 1 and 3. Id.
`
`The Petition thus presents redundant grounds without attempting to make
`
`any meaningful distinction between them, and is therefore “contrary to the
`
`regulatory and statutory mandates.” Liberty Mutual, CBM2012-00003, Paper 7 at
`
`2. Accordingly, the Board should decline to review and reject institution on
`
`Petitioner’s redundant Grounds. Id.; Teradata Ops., Inc. v. Realtime Data LLC,
`
`IPR2017-00557, Paper 14 at 35-36, 38 (PTAB Jul. 6, 2017); OpenTV, IPR2015-
`
`00980, Paper 11 at 25-26; Shopkick, IPR2015-00279, Paper 7 at 27-28; Hyundai
`
`Motor, IPR2014-00658, Paper 8 at 23.
`
`VI. Petitioner’s Ground 1 theory—Hsu in view of Franaszek—does not
`teach “a default data compression encoder” as required by limitation E
`Claim 105 requires the use of “a default data compression encoder.”
`
`Specifically, limitation E recites “otherwise compressing said data block with a
`
`default data compression encoder[.]” And, as discussed above in Section III, the
`
`
`
`
`
`
`- 12 -
`
`
`
`
`
` Case IPR2017-01660
`
`PATENT OWNER’S PRELIMINARY RESPONSE
`Petition construes a “default data compression encoder” as “an encoder used
`
`automatically in the absence of a designated alternative” (emphasis added).
`
`As the below discussion demonstrates, the prior art references relied on in
`
`Ground 1—Hsu and Franaszek—do not teach limitation E, either alone or in
`
`Petitioner’s proposed combination. Rather, both references teach compression
`
`systems that always perform testing and analysis before applying an encoder. The
`
`references thus do not teach (alone or combined) a “default data compression
`
`encoder”—i.e., “an encoder used automatically”—as required by limitation E.
`
`A. As the Petition acknowledges, Hsu does not teach limitation E.
`The Petition acknowledges that Hsu does not meet limitation E’s
`
`requirement of a default encoder. See Pet. 21 (“All of the limitations recited in
`
`claim 105 except for limitation E are expressly disclosed in Hsu.”). That is
`
`because, as explained below, Hsu’s system always recognizes the data type and
`
`three other parameters of each data block and always selects a compression
`
`algorithm based on those recognized data types and parameters. Hsu thus teaches a
`
`system that always performs content dependent compression and that never applies
`
`a “default data compression encoder.”
`
`Specifically, Hsu describes performing analysis on a block of data to
`
`identify a data block type and three redundancy metrics:
`
`
`
`
`
`
`- 13 -
`
`
`
`
`
` Case IPR2017-01660
`
`PATENT OWNER’S PRELIMINARY RESPONSE
`[E]ach data block to be compressed is analyzed. Four routines are
`used to determine a type of the data. One routine determines a “data
`classification” of a data block by analyzing 512-byte segments at the
`beginning, middle, and end of the data block. The other three routines
`each determine a “redundancy metric”
`that
`indicates
`the
`“compressibility of a block of data” by various types of compression
`encoders.
`
`Pet. 17 (citing Ex. 1002 (Hsu) at 1104).
`
`Hsu explains that it identifies the block as one of ten data types. See Pet. 17-
`
`19 (citing Hsu at 1104); Ex. 2009 at ¶ 29-31 (discussing Hsu’s teachings and
`
`explaining why “Hsu gives neither an explicit nor an implicit suggestion that it will
`
`not identify a data type”).
`
`Hsu also teaches that the three calculated redundancy metrics measure: (1)
`
`“the degree of variation in character frequency or alphabetic distribution, MAD,”
`
`(2) “the average run length of the block, MRL,” and (3) “the string repetition ratio
`
`of the block, MSR.” See Hsu at 1104 (emphasis original); Pet. 17-18 (citing Hsu at
`
`1104); Ex. 2010 at ¶ 56.
`
`Redundancy metrics calculated to have a value of less than 2.5 are
`
`disregarded, and if all three metrics are under 2.5, the block is deemed to be
`
`uncompressible. See Pet. at 17-19; Hsu at 1106; Ex. 2011 at ¶¶ 27-28.
`
`
`
`
`
`
`- 14 -
`
`
`
`
`
` Case IPR2017-01660
`
`PATENT OWNER’S PRELIMINARY RESPONSE
`On the other hand, “[i]f at least one of the [redundancy] parameters is
`
`greater than 2.5, the file is considered compressible.” Hsu at 1106. For any file that
`
`is considered compressible, Hsu will select the compression algorithm associated
`
`with the largest redundancy metric, as shown in Hsu’s Table I, reproduced below.
`
`Id. (“The maximum of the normalized metrics is then selected and used in
`
`conjunction with the file type to select the appropriate compression algorithm from
`
`the lookup table described in the following section.”); Pet. 18-19; Ex. 2010 at ¶ 57-
`
`58.
`
`See Pet. 18 (replicating table from Hsu at 1107 with annotated row delineations).
`
`
`
`
`
`
`- 15 -
`
`
`
`
`
`
`
` Case IPR2017-01660
`
`PATENT OWNER’S PRELIMINARY RESPONSE
`Hsu itself provides an example of how Table I applies: if Hsu identifies a
`
`data block as having the “high redundancy binary” data type, and the string
`
`repetition redundancy metric (MSR) has the highest value of the three redundancy
`
`metrics (and is greater than 2.5), then Hsu will compress that data block with
`
`Lempel-Ziv with the Freeze heuristic, as shown in in Table I above. See Hsu at
`
`1106.
`
`As shown in Table I, Hsu presents four combinations of identified data type
`
`and redundancy metric with no associated compression algorithm, i.e., the MRL
`
`metric with the “natural language” data type, and the MAD metric with “low
`
`redundancy binary,” “audio,” and “low resolution graphic” data types. Hsu at
`
`1106; Pet. 18-19. If the largest redundancy metric for a given data type happens to
`
`be one of the four for which Hsu’s Table I does not have a specified compression
`
`algorithm, as denoted by a double “*,” then Hsu teaches that “the next highest
`
`metric is used instead, provided it is above the threshold.” Hsu at 1106; Ex. 2010 at
`
`¶ 58. Thus if the “low redundancy binary” data type is identified, and the MAD
`
`metric is the largest, Hsu will compress the data block using the compression
`
`algorithm associated with the next largest redundancy metric (run length encoding
`
`or Lempel-Ziv), provided the metric is above 2.5. Hsu at 1106. If, on the other
`
`hand, neither of the remaining two redundancy metrics is greater than 2.5, Hsu
`
`teaches that compression should not occur, because such a scenario demonstrates
`
`
`
`
`- 16 -
`
`
`
`
` Case IPR2017-01660
`
`PATENT OWNER’S PRELIMINARY RESPONSE
`“a decision that the (poor) potential for compression is outweighed by the overhead
`
`of executing the compression algorithm.” Id. (“In some cases, the only redundancy
`
`type for which a metric is above the threshold accesses a null entry in the database
`
`of compression algorithms.”) (emphasis added); Ex. 2011 at ¶ 28 (“Hsu expressly
`
`teaches that the three redundancy metrics are used as ‘absolute indicators of
`
`compressibility’ and that they provide a determination as to whether a given fixed-
`
`sized block should be compressed or not compressed at all[.]”) (citing and quoting
`
`Hsu at 1102).
`
`Hsu thus teaches only two potential outcomes for each data block: (1)
`
`application of a content dependent encoder based on the recognized data type and
`
`redundancy metrics of that block, or (2) no compression. Thus as the Petition
`
`recognizes, Hsu does not teach a “default data compression encoder” as required
`
`by limitation E. See Pet. 19 (“Hsu explicitly discloses every limitation of the
`
`challenged claim except for the limitations related to the ‘default data compression
`
`encoder.’”); Ex. 2010 at ¶ 60-62 (“[A] POSA would have understood that Hsu
`
`discloses only a content dependent technique that always recognizes a data type for
`
`a data block and selects an appropriate compression algorithm based on the
`
`classified data type and determined redundancy metrics.”).
`
`
`
`
`
`
`- 17 -
`
`
`
`
`
`B.
`
` Case IPR2017-01660
`
`PATENT OWNER’S PRELIMINARY RESPONSE
`Franaszek does not teach limitation E because it always applies
`content dependent compression, if it compresses at all.
`The Petition urges that Franaszek teaches “using a default data compression
`
`encoder when a data type with one or more associated encoders is not identified.”
`
`Pet. 27-28. As demonstrated below, however, Franaszek does not. Instead, it
`
`expressly teaches a system that always performs analysis on the content of the data
`
`block and always designates a content dependent encoder (or does not perform
`
`compression at all), regardless of whether a data type is identified.
`
`1.
`
`Regardless of whether a data type is provided, the system of
`Franaszek always analyzes data blocks to determine their
`compressibility and always designates an encoder to apply
`based on that measured compressibility.
`Franaszek describes a block-by-block system for “compressing . . . data
`
`using a plurality of data compression mechanisms.” Ex. 1003 (Franaszek) at
`
`Abstract; Ex. 2010 at ¶ 25. As Franaszek explains, its objective is to “to
`
`dynamically compress data blocks by using the best of a given set of methods and,
`
`for dictionary-based compression, by using the best of a given set of dictionaries,”
`
`Franaszek at 3:25-28, which “increase the number of data blocks that can be stored
`
`in the second memory 20 given a fixed second memory size.” Id. at 4:14-17; Ex.
`
`2010 at ¶ 64.
`
`To accomplish that objective, Franaszek discloses a data compressor 220
`
`that compresses data blocks as they are transferred to the second memory, which is
`
`
`
`
`
`
`- 18 -
`
`
`
`
`
` Case IPR2017-01660
`
`PATENT OWNER’S PRELIMINARY RESPONSE
`illustrated in the annotated Fig. 2 and reproduced below. Franaszek at 4:25-28; Ex.
`
`2010 at ¶ 65.
`
`
`
`Franaszek explains (and shows in Fig. 2 above) that “the uncompressed data
`
`blocks 210 [] can optionally contain type information 205. The type information
`
`can be, for example[], image data encoded in a given format, source code for a
`
`given programming language, etc.” Franaszek at 4:30-35; Ex. 2010 at ¶ 66. When
`
`the optional data type information 205 is available, Franaszek uses a preselected
`
`list of compression methods based on type information availability for a data
`
`block, as shown in step 404 of annotated Fig. 4A, below. Franaszek at 5:48-54; Ex.
`
`2010 at ¶ 67. When the optional data type information 205 is not available,
`
`Franaszek sets its Compression Method List (CML) to a default list of compression
`
`methods, as shown in step 407 of annotated Fig. 4A below. Franaszek at 5:48-54;
`
`Ex. 2010 at ¶ 67.
`
`
`
`
`
`
`- 19 -
`
`
`
`
`
` Case IPR2017-01660
`
`PATENT OWNER’S PRELIMINARY RESPONSE