`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`UNIFIED PATENTS INC.
`SAP AMERICA, INC.,
`
`Petitioners,
`
`v.
`
`CLOUDING IP, LLC
`
`Patent Owner.
`
`
`Case IPR2013-00586
`Case IPR2014-00306
`Patent 6,738,799
`
`PETITIONERS’ CONSOLIDATED OPPOSITION TO PATENT OWNER’S
`CONTINGENT MOTION TO AMEND
`
`
`
`
`
`
`
`
`
`I.
`
`Introduction
`
`The Patent Owner proposes an amendment to claim 42 that the Patent
`
`Owner’s own expert admitted was disclosed by the prior art of record and that is
`
`not enabled by the originally-filed specification. For at least these reasons, the
`
`Board should deny the Patent Owner’s motion.
`
`In the Patent Owner’s Contingent Motion to Amend (hereinafter “Motion”),
`
`Patent Owner Clouding IP, LLC (hereinafter “Clouding” or “Patent Owner”)
`
`proposes a new claim 47 to replace current claim 42 in the event that claim 42 is
`
`found unpatentable. Claim 47 contains all limitations of claim 42 and adds the
`
`limitation that determining if the second computer has the latest version of the file
`
`occurs “by comparing representations of segments of the latest version of the file
`
`with representations of segments of an earlier version of the file in which ends of
`
`each of the segments of the earlier version of the file are defined by segment
`
`delimiters that are statistically determined to be optimal division points for the
`
`segments.” Motion, at 1.
`
`Clouding’s motion should be denied because the motion does not meet the
`
`basic requirement of motions to amend – that the proposed claim be patentable
`
`over the prior art. Indeed, proposed claim 47 is not even patentable over the prior
`
`1
`
`
`
`
`
`art already of record in this proceeding.1 As described below, the Williams patent
`
`(Ex. 1006) discloses the new limitations of claim 47. While Clouding attempted to
`
`distinguish Williams in its Motion, Clouding’s expert admitted at his deposition
`
`that Williams discloses every limitation that Clouding argued was absent.
`
`Accordingly, Williams anticipates proposed claim 47. And, as discussed in more
`
`detail below, other prior art also renders claim 47 unpatentable.
`
`Moreover, this new limitation is not enabled because, although the ‘799
`
`Patent contains the precise words of the new limitation (“statistically determined to
`
`be an optimal” Ex. 1001, at 8:14-18), the specification fails to describe how this
`
`functionality is achieved. For example, the ‘799 patent does not describe the
`
`1
`Clouding’s motion should also be denied because Clouding has not met its
`
`burden of establishing that it is entitled to relief it requests. Clouding relies on the
`
`declaration of its expert, which states that added limitations were not known in the
`
`prior art, but does not point to any specific prior art other than two references
`
`already at issue in this case. To be entitled to a claim amendment in an IPR, “the
`
`patent owner must show patentability over the prior art in general, and not just over
`
`the references applied by the petitioner against the original patent claims.” Toyota
`
`Motor Corporation v. American Vehicular Sciences LLC, IPR2013-00419 (PTAB)
`
`(Paper 32) at 4. See also Idle Free Sys., Inc. v. Bergstrom, Inc., IPR2012-00027
`
`(PTAB) (Paper 66) at 33.
`
`2
`
`
`
`
`
`statistical procedure, the “optimal” algorithm, or any of the factors that a person of
`
`ordinary skill in the art would consider when determining how to practice this
`
`limitation. Ex. 1019, at 53:20-54:9; 54:14-17; 55:19-56:6; Ex. 1018, at ¶ 16.
`
`Thus, one of ordinary skill in the art would be left to unduly experiment. Ex. 1018,
`
`at ¶¶ 15-19.
`
`For at least these reasons, Clouding’s Motion should be denied.
`
`II. Claim Construction
`
`Pursuant to 37 C.F.R. § 42.204(b)(3), the claims subject to inter partes
`
`review shall receive the “broadest reasonable construction in light of the
`
`specification of the patent in which [they] appear[].” See, 42 C.F.R. § 100(b); 77
`
`Fed. Reg. 48756, 48764, 48766 (Aug. 14, 2012).
`
`Clouding appears to argue that “segments” should be construed to mean
`
`“variable length portions of a file, defined by segment delimiters” and/or “a part
`
`into which a file may be divided.” Motion, at 11. Petitioners contend that the
`
`meaning of “segments” is clear to one of ordinary skill in the art, and does not
`
`require construction. Ex. 1018, at ¶ 3. If the Board disagrees, this term should be
`
`construed to mean “portions of a file.” Id. Petitioners further contend that
`
`construing “segments” to mean “variable length portions of a file” improperly
`
`narrows the claim to only “variable length” segments.
`
`3
`
`
`
`
`
`Clouding appears to argue that “ends of each of the segments of the earlier
`
`version of the file are defined by segment delimiters that are statistically
`
`determined to be optimal division points for the segments” should be interpreted to
`
`mean “optimal locations for the file segment delimiters are determined using a
`
`statistical procedure” or that “determining [the] end points . . . relies upon a
`
`statistical procedure to determine the optimum end points.” Motion, at 12.
`
`Petitioners contend that Clouding’s proposed construction is unnecessary because
`
`it simply rearranges the claim language. See Google Inc. and Yahoo! Inc.
`
`Petitioners v. Createads LLC Patent Owner, IPR2014-00200, Paper 10, at 10
`
`(PTAB) (rejecting the proposed construction as “not helpful because it simply
`
`rearranges the express claim language.”).
`
`III. Clouding’s Motion Should be Denied Because Proposed Substitute
`Claim 47 is Unpatentable
`
` “As the moving party, [Clouding] bears the burden of proof to establish that
`
`it is entitled to the relief requested.” Idle Free Sys., Inc. v. Bergstrom, Inc.,
`
`IPR2012-00027 (PTAB) (Paper 66) at 26 (citing 37 C.F.R. § 42.20(c)). A motion
`
`to amend is granted only with the patent owner “having demonstrated the
`
`patentability of those substitute claims.” Id. at 26. Clouding cannot meet this
`
`burden because, as demonstrated below, claim 47 is unpatentable over the prior art
`
`and because it is not enabled.
`
`4
`
`
`
`
`
`A. Williams renders proposed claim 47 unpatentable under 35
`U.S.C. § 102(e)
`
`Clouding incorrectly asserts that the new limitation distinguishes claim 47
`
`over Williams. Clouding’s assertion is based at least in part on an interpretation of
`
`Williams that Clouding’s own expert cannot defend and is, quite simply, wrong.
`
`The Williams patent describes methods of partitioning blocks of data into
`
`subblocks, and one such method employs hash functions and determines block
`
`partitions that are “statistically determined to be optimal division points.” Ex.
`
`1018, at ¶ 4-10. Williams describes several applications of this subblock
`
`partitioning (Ex. 1006 at 2:27-49), one being the incremental backup system shown
`
`in figure 25, which petitioners cite in the Petition. See, e.g., Petition, at 40.
`
`Williams states that its inventions use hash functions not only for
`
`partitioning subblocks or “determining subblock boundaries,” but also for
`
`confirming the identities of subblocks by comparing hash values. Ex. 1006, at
`
`12:6-10; Ex. 1018, at ¶ 6. Williams describes several types of hash functions,
`
`some “weak” and some “strong,” and describes the advantages and disadvantages
`
`of each. Ex. 1006, at 10:59-14:50; Ex. 1018, at ¶ 6. Williams teaches that
`
`“nothing less than strong one-way hash functions should be used” for confirming
`
`the identities of subblocks (Ex. 1006, at 12:19-20) due to the “security dangers”
`
`inherent in employing other hash functions. Ex. 1006, at 12:21-25; Ex. 1018, at ¶
`
`5
`
`
`
`
`
`6. Williams teaches that using weak hash functions to identify subblocks may
`
`allow intruders to modify subblocks, and that a weak hash function may produce
`
`the same hash for the modified subblock as the unmodified subblock, and the
`
`intrusion will therefore not be detected. Ex. 1006, at 12:21-33; Ex. 1018, at ¶ 6.
`
`When using hash functions to determine subblock boundaries, Williams teaches
`
`that “it would be inefficient to use hash functions from any but the weakest class”
`
`“as the determination of subblock boundaries does not require identity or
`
`cryptographic strength.” Ex. 1006 at 12:13-15 (emphasis added); Ex. 1018, at ¶ 6.
`
`Williams is clear that, despite the fact that the hash need not be strong for use in
`
`defining segment delimiters, “care should be taken to select a function that suits
`
`the application to which it is to be applied.” Ex. 1006 at 13:66-14:2; Ex. 1018, at ¶
`
`6. Williams discloses that the use of a “narrow hash function” for this purpose
`
`“provides both efficiency and a (deterministic) randomness that will enable the
`
`implementation to operate effectively over a wide-range of data.” Ex. 1006 at
`
`14:8-11 (emphasis added); Ex. 1018, at ¶ 7.
`
`Williams provides an example of a narrow hash function (reproduced below)
`
`for determining subblock boundaries that “has been implemented and tested and
`
`seems to perform very well on a variety of data types.” Ex. 1006 at 14:21-24; Ex.
`
`1018, at ¶ 7.
`
`6
`
`
`
`
`
`
`
`Williams describes that this hash function involves the constant p, which “is the
`
`inverse of the probability of placing a boundary at an arbitrary position in a
`
`randomly generated block of data, and can be set to any integer value in [0,65535]”
`
`(Ex. 1006 at 14:31-34; Ex. 1018, at ¶ 7.), and was “carefully chosen” to be 40543
`
`in the above example based a upon known hash function design guidelines. 4:36-
`
`40; Ex. 1018, at ¶ 7.
`
`1. Williams discloses “segment delimiters that are statistically
`determined to be optimal division points for the segments,”
`as recited in claim 47
`
`Clouding does not dispute that Williams discloses comparing representations
`
`of segments of the latest version of the file (i.e., hashes of subblocks from X) with
`
`representations of segments of an earlier version of the file (i.e., hashes of sublocks
`
`in shadow file S). See Response, at 37-39. Clouding only disputes whether
`
`Williams discloses “segment delimiters that are statistically determined to be
`
`optimal division points for the segments.”
`
`As described above, Williams describes the use of probabilities to determine
`
`“subblock boundaries.” Williams’ “subblock boundaries” are identical to the
`
`claimed “segment delimiters” and “division points,” and are a method of
`
`“statistically determining” the subblock boundaries. Ex. 1018, at ¶ 4-10.
`
`7
`
`
`
`
`
`Clouding’s expert Dr. Mohapatra agrees. Ex. 1019, at 58:25-59:7; 59:12-15.
`
`When asked if “the use of probability could be used to select the statistically
`
`determined optimal division points,” Dr. Mohapatra responded: “Yes.” Ex. 1019,
`
`at 59:12-15. In fact, when asked if the formula reproduced supra at 7 was “a
`
`statistical determination for division points,” he responded: “Yes.” Ex. 1019, at
`
`69:9-12.
`
`Williams further discloses that its subblock boundaries “are determined to be
`
`optimal division points,” as recited in claim 47, because, as described above,
`
`Williams discloses that its partitioning method “provides both efficiency and a
`
`(deterministic) randomness that will enable the implementation to operate
`
`effectively over a wide-range of data.” Dr. Mohapatra agrees.2 As described
`
`above, Dr. Mohapatra testified that the use of probability, as disclosed in Williams,
`
`2 Dr. Mohapatra initially tried to argue that, even though Williams describes
`
`“statistically determined” segment delimiters, it does not teach that these
`
`“statistically determined” segment delimiters are “optimal” because Williams does
`
`not “claim[] it as an efficient approach or an optimal approach.” But, as described
`
`above, Williams does exactly that – it describes its method as
`
`“efficient.” Immediately after making this statement, Dr. Mohapatra conceded that
`
`the teaching of Williams is “certainly an easy way to generate blocks.” Ex. 1019,
`
`at 70:19-71:6.
`
`8
`
`
`
`
`
`“[can] be used to select the statistically determined optimal division points.” Ex.
`
`1019, at 59:12-15. Moreover, Dr. Mohapatra testified that an “optimal” method of
`
`statistically determining division points is one that is “efficient:”
`
`Q. So statistically determined optimal decision points may not, in fact,
`be the very best decision points as compared to all other possibilities;
`correct?
`A. Yes, yes.
`Q. In fact, they may only be efficient.
`A. Yes.
`
`Ex. 1019, at 67:2-8. In fact, both the Patent Owner’s Motion as well as Dr.
`
`Mohapatra rely on “efficient” as support for “optimal.” See Motion, at 4-5; Ex.
`
`2009, at 23. Thus, the Patent Owner cannot be heard now to draw a distinction
`
`between the two.
`
`As such, Williams discloses all of the limitations of proposed claim 47, and
`
`therefore anticipates claim 47. Clouding’s motion should be denied for at least this
`
`reason.
`
`B.
`
`The New Limitations Were Disclosed in at Least One Other Prior
`Art Reference
`
`In addition to Williams, at least one other prior art reference also discloses
`
`the new limitations introduced into claim 47-- U.S. Patent No. 6,076,084 to Harlan
`
`(“Harlan”). Ex. 1020. Harlan is from the same field as Balcha and Williams, and
`
`9
`
`
`
`
`
`one of ordinary skill in the art would have been motivated to combine Harlan with
`
`Williams or Balcha because the references are all concerned with dividing blocks
`
`of data into subblocks. One of ordinary skill in the art would naturally look to
`
`efficient techniques, such as those found in Harlan, for dividing blocks into
`
`subblocks to reduce transmission and speed up the file synchronization process. In
`
`fact, reducing transmission and increasing the speed of the file synchronization
`
`process is one of the main design goals of file synchronization systems. Ex. 1018,
`
`at ¶ 14.
`
`1.
`
`U.S. Patent No. 6,076,084 to Harlan
`
`U.S. Patent No. 6,076,084 to Harlan (“Harlan”), entitled “File Transfer
`
`Method and Apparatus Utilizing Delimiters,” claims priority to January 3, 1994,
`
`issued on June 13, 2000, and is therefore prior art to the ‘799 patent under 35
`
`U.S.C. § 102(e).
`
`The method of Harlan, which transfers an “update” from computer 1 to
`
`computer 2 after determining differences between an old file and a new file, is
`
`summarized as follows:
`
`
`
`10
`
`
`
`
`
`
`Harlan, at 2:55-3:12; Ex. 1018, at ¶ 11-14. As shown above, and like the ‘799
`
`patent, Harlan compares the old file with the new file by dividing the files into
`
`segments based upon a selected delimiter, and then generating a hash (i.e.,
`
`“signature” or “representation”) for each segment. Ex. 1018, at ¶ 11.
`
`Regarding the delimiters, Harlan discloses that “[a] variety of algorithms
`
`could be used to select a delimiter,” but that “in the preferred embodiment,” the
`
`delimiter is statistically determined “by first going through the DSPT and choosing
`
`the first delimiter where the number of occurrences is greater than 50 and the
`
`average length between delimiters is greater than 1,000 and less than 30,000.” Ex.
`
`1020, at 4:31-37; Ex. 1018, at ¶ 11-13. Harlan discloses that if a delimiter is not
`
`found that meets these criteria, the criteria are eased. Ex. 1020, at 4:37-39; Ex.
`
`1018, at ¶ 13. Harlan teaches that the determined delimiters are “optimal division
`
`points,” as recited in proposed claim 47, by describing that a delimiter should not
`
`be chosen arbitrarily because “experience will show the particular criteria that
`
`11
`
`
`
`
`
`works most satisfactorily with each type of file.” Ex. 1020, at 44-46; Ex. 1018, at
`
`¶ 13.
`
`Because Harlan discloses “segment delimiters that are statistically
`
`determined to be optimal division points for the segments,” and one of ordinary
`
`skill in the art would have been motivated to combine Harlan with either Williams
`
`or Balcha, the combination of Harlan with either Williams or Balcha renders
`
`obvious proposed claim 47.
`
`C. The New Limitations Are Unpatentable Under 35 U.S.C. § 112 as
`Lacking Enablement
`
`Proposed claim 47 is unpatentable under 35 U.S.C. § 112, first paragraph
`
`because the specification of the ‘799 patent does not enable one of ordinary skill in
`
`the art to make and use the new limitations without undue experimentation. First,
`
`the specification says nothing about how delimiters are “statistically determined to
`
`be optimal division points for the segments”; Clouding’s own expert admits this.
`
`Ex. 1019, at 53:20-54:9; 54:14-17; 55:19-56:6; Ex. 1018, at ¶ 15-19. As such, one
`
`of ordinary skill in the art would need to develop from scratch a procedure for
`
`“statistically determin[ing] delimiters to be optimal division points for the
`
`segments.” As Dr. Hutchinson testifies, this would result in a level of
`
`experimentation that is undue and unreasonable. Ex. 1018, at ¶ 19.
`
`12
`
`
`
`
`
`Developing such a procedure would be difficult, and could not be done
`
`without undue experimentation. For example, the procedure must not only work,
`
`but must work “optimally,” for every file that it encounters. Ex. 1018, at ¶ 19.
`
`Ensuring this type of consistency and reliability requires a significant amount of
`
`work. Ex. 1018, at ¶ 19. Harlan and Williams each describe the difficulty of
`
`determining delimiters.3 Ex. 1018, at ¶ 18. To claim an “optimal” determination,
`
`the specification must explain how to do so. But, the specification is silent on this
`
`issue. As such, proposed claim 47 is not enabled and is unpatentable under 35
`
`U.S.C. § 112, first paragraph.
`
`IV. Conclusion
`
`The Patent Owner’s Contingent Motion to Amend attempts to add a
`
`limitation that is disclosed by the prior art of record as well as other prior art and
`
`that is not enabled. For these reasons, the Petitioners respectfully request that this
`
`motion be denied.
`
`
`
`
`3
`Dr. Mohapatra’s position that one of ordinary skill in the art could develop
`
`the necessary algorithm in “20 to 30 minutes” (Ex. 1019, at 57:20-58:4) is
`
`conclusory and unsupported. Ex. 1018, at ¶ 17.
`
`13
`
`
`
`
`
`/Michael L. Kiklis/_____________
`Michael L. Kiklis (Reg. No. 38,939)
`Scott A. McKeown (Reg. No. 42,866)
`Oblon Spivak
`1940 Duke Street
`Alexandria, Virginia 22314
`Tel: (703) 413-3000
`CPdocketKiklis@oblon.com
`CPdocketMcKeown@oblon.com
`
`Attorneys for Petitioner
`Unified Patents Inc.
`
`
`/Frank C. Cimino, Jr./__________
`Frank C. Cimino, Jr. (Reg. No. 39,945)
`Megan S. Woodworth (Reg. No. 53,655)
`S. Gregory Herrman (Reg. No. 66,271)
`DICKSTEIN SHAPIRO LLP
`1825 Eye Street, NW
`Washington, DC 20006-5403
`Tel: (202) 420-2200
`CiminoF@dicksteinshapiro.com
`WoodworthM@dicksteinshapiro.com
`HerrmanG@dicksteinshapiro.com
`
`Attorneys for Petitioner
`SAP America, Inc.
`
`14
`
`
`
`
`
`CERTIFICATE OF SERVICE
`
`The undersigned certifies that the foregoing PETITIONERS’
`
`CONSOLIDATED OPPOSITION TO PATENT OWNER’S CONTINGENT
`
`MOTION TO AMEND was electronically filed with the Patent Review Processing
`
`System and served by electronic mail on July 21, 2014, to the following:
`
`Tarek Fahmi
`Ascenda Law Group
`84 W. Santa Clara St.
`Suite 550
`San Jose CA 95113
`tarek.fahmi@ascendalaw.com
`
`
`By: /Michael L. Kiklis/
` Michael L. Kiklis
`Reg. No. 38,939
`
`
`
`
`
`
`
`Dated: July 21, 2014