`Apple v. California Institute of Technology
`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`APPLE INC.,
`Petitioner,
`
`v.
`
`CALIFORNIA INSTITUTE OF TECHNOLOGY,
`Patent Owner.
`_________________________________________
`
`Case IPR2017-00219
`Patent 7,116,710
`_________________________________________
`
`DECLARATION OF BRENDAN FREY, PH.D.
`REGARDING U.S. PATENT NO. 7,116,710
`CLAIMS 1-8, 11-17, 19-22, AND 24-33
`
`Apple v. Caltech
`IPR2017-00219
`Apple 1265
`
`
`
`U.S. Patent No. 7,116,710
`Apple v. California Institute of Technology
`TABLE OF CONTENTS
`BACKGROUND ............................................................................................. 1
`
`LEGAL PRINCIPLES ..................................................................................... 6
`
`I.
`
`II.
`
`III. DIVSALAR IN VIEW OF LUBY RENDERS CLAIMS 1-8 AND 11-14
`
`OBVIOUS ........................................................................................................ 8
`
`A.
`
`A.
`
`B.
`
`C.
`
`D.
`
`E.
`
`Luby’s Table 1. ...................................................................................... 8
`
`Luby teaches irregular repetition of information bits. ........................ 11
`
`Using Luby’s irregular degrees in Divsalar results in irregular
`
`repetition of information bits even if Luby alone does not
`
`disclose irregular repetition of information bits. ................................. 15
`
`Divsalar in view of Luby discloses the “partitioning” limitation. ...... 20
`
`One of ordinary skill would have combined Divsalar and Luby. ....... 21
`
`Dr. Divsalar’s testimony confirms that it would have been
`
`obvious to combine Divsalar and Frey. ............................................... 31
`
`IV. DIVSALAR IN VIEW OF LUBY AND LUBY97 RENDERS CLAIMS
`
`15-17, 19-22, AND 24-33 OBVIOUS ........................................................... 33
`
`V.
`
`SECONDARY CONSIDERATIONS OF NON-OBVIOUSNESS .............. 34
`
`i
`
`
`
`U.S. Patent No. 7,116,710
`Apple v. California Institute of Technology
`VI. AVAILABILITY FOR CROSS-EXAMINATION ...................................... 35
`
`VII. RIGHT TO SUPPLEMENT .......................................................................... 36
`
`VIII. JURAT ........................................................................................................... 36
`
`
`
`
`
`
`
`ii
`
`
`
`U.S. Patent No. 7,116,710
`Apple v. California Institute of Technology
`
`
`
`I, Brendan Frey, Ph.D., declare as follows:
`
`1. My name is Brendan Frey.
`
`I.
`
`BACKGROUND
`
`2.
`
`I received a B.Sc. with Honors in Electrical Engineering from the
`
`University of Calgary in 1990, a M.Sc. in Electrical and Computer Engineering from
`
`the University of Manitoba in 1993, and a Ph.D. in Electrical and Computer
`
`Engineering from the University of Toronto in 1997.
`
`3.
`
`Since July 2001, I have been at the University of Toronto, where I am a
`
`Professor of Electrical and Computer Engineering and Computer Science.
`
`4.
`
`During my career I have conducted research in the areas of graphical
`
`models, error-correcting coding, machine learning, genome biology, medicine, and
`
`computer vision.
`
`5.
`
`In 2015, I co-founded Deep Genomics Inc., a startup located in Toronto
`
`that is using artificial intelligence to find new medicines. Since then I have acted as
`
`its Chief Executive Officer. Deep Genomics has received over $17M in venture
`
`capital funding, mostly from Silicon Valley investors. Deep Genomics has recruited
`
`scientists and engineers from top universities, including MIT, Stanford, the
`
`University of California, San Diego, and the University of Toronto, and from
`
`
`
`1
`
`
`
`U.S. Patent No. 7,116,710
`Apple v. California Institute of Technology
`competing biotech and software companies, including Amazon, Autodesk, Calico,
`
`and Human Longevity. In 2017, I co-founded the Vector Institute for Artificial
`
`Intelligence. The Vector Institute is internationally regarded as one of, if not the, top
`
`artificial intelligence research institutes in the world. It has over $200M in funding
`
`and its current and newly hired professors have chosen faculty positions at the
`
`Vector Institute in preference to faculty offers from leading universities, including
`
`Stanford and MIT, and to senior researcher offers from leading industrial labs,
`
`including DeepMind, Google, Facebook, Microsoft and OpenAI.
`
`6.
`
`I have received a number of honors and awards for the research I have
`
`conducted. In 2008, I was named a Fellow of the Institute for Electrical and
`
`Electronic Engineers (IEEE), an honor given to a person with an “extraordinary
`
`record or accomplishments” in the field of electrical engineering. In 2009, I was
`
`named a Fellow of the American Association for the Advancement of Science
`
`(AAAS), an honor that recognizes “efforts on behalf of the advancement of science
`
`or its applications which are scientifically or socially distinguished.” In 2009, I was
`
`awarded a Steacie Fellowship for my work on the theory and implementation of
`
`artificial and natural mechanisms for inferring patterns from data. The Steacie
`
`Fellowship is awarded by the Natural Sciences and Engineering Research Council of
`
`Canada (NSERC) to “outstanding and highly promising scientists and engineers”
`
`
`
`2
`
`
`
`U.S. Patent No. 7,116,710
`Apple v. California Institute of Technology
`who are faculty members of Canadian universities. In 2011, I received the
`
`NSERC’s John C. Polanyi Award, in recognition of my research on inferring genetic
`
`codes embedded in DNA that direct activities within cells. In 2015, I was elected as
`
`a Fellow of the Royal Society of Canada, with the following citation: “Professor
`
`Frey has contributed to the emergence of new fields of research in machine learning
`
`and genome biology. He was one of the first researchers to successfully train a deep
`
`neural network, and he was a pioneer in inventing message passing algorithms,
`
`which are now widely used. He co-developed the long-sought-after ‘splicing code’
`
`for determining how genes are expressed and introduced a new approach to
`
`understanding the genetics of disease.”
`
`7.
`
`Throughout my career I have received funding from various
`
`governmental agencies to support my research, including the Natural Sciences and
`
`Engineering Research Council of Canada, the Canadian Institutes of Health
`
`Research, and the Canadian Institute for Advanced Research.
`
`8.
`
`I have authored more than 200 publications and am named as an
`
`inventor on nine patents issued by the U.S. Patent and Trademark Office.
`
`9.
`
`A copy of my curriculum vitae is included as Exhibit 1266.
`
`10.
`
`I have reviewed the specification and claims of U.S. Patent
`
`No. 7,116,710 (the “’710 patent”; Ex. 1201). I have been informed that the ’710
`
`
`
`3
`
`
`
`U.S. Patent No. 7,116,710
`Apple v. California Institute of Technology
`patent claims priority to a provisional application filed on May 18, 2000, and to U.S.
`
`application Ser. No. 09/922,852, filed on Aug. 18, 2000.
`
`11.
`
`I have also reviewed the following references, all of which I understand
`
`to be prior art to the ’710 patent:
`
`(cid:120) Frey, B. J. and MacKay, D. J. C., “Irregular Turbocodes,” Proc. 37th
`Allerton Conf. on Comm., Control and Computing, Monticello,
`Illinois, published on or before March 20, 2000 (“Frey”; Ex. 1202.)
`
`(cid:120) Frey, B. J. and MacKay, D. J. C., “Irregular Turbo-Like Codes,”
`37th Allerton Conf. on Comm., Control and Computing, Monticello,
`Illinois, published on or before September 24, 1999 (“Frey Slides”;
`Ex. 1213)
`
`(cid:120) D. Divsalar, H. Jin, and R. J. McEliece, “Coding theorems for
`“turbo-like” codes,” Proc. 36th Allerton Conf. on Comm., Control
`and Computing, Allerton, Illinois, pp. 201-10, March, 1999
`(“Divsalar”; Ex. 1203.)
`
`(cid:120) Luby, M. et al., “Analysis of Low Density Codes and Improved
`Designs Using Irregular Graphs,” STOC ‘98, pp. 249-59, published
`in 1998 (“Luby”; Ex. 1204.)
`
`(cid:120) Luby, M. et al., “Practical Loss-Resilient Codes,” STOC ‘97, pp.
`150-159, published in 1997 (“Luby97”; Ex. 1211)
`
`12.
`
`I have also reviewed the following filings in this inter partes review:
`
`(cid:120) Petition for Inter Partes Review of U.S. Pat. No. 7,116,710 (Paper 5)
`(“Petition” or “Pet.”)
`
`(cid:120) Patent Owner’s Preliminary Response (Paper 16) (“POPR”)
`
`(cid:120) Institution Decision (Paper 17)
`
`4
`
`
`
`
`
`U.S. Patent No. 7,116,710
`Apple v. California Institute of Technology
`(cid:120) Patent Owner’s Response (Paper 34) (“POR”)
`
`(cid:120) Declaration of Professor James Davis, Ph.D. (Ex. 1206)
`
`(cid:120) Transcript of the Deposition of Dr. Michael Mitzenmacher
`(Ex. 1262) and associated exhibits (Exs. 1244-1249)
`
`(cid:120) Transcript of the Deposition of Dr. Dariush Divsalar (Ex. 1264) and
`associated exhibits (Exs. 1257-1261)
`
`(cid:120) California Institute of Technology v. Hughes Communications Inc.,
`No. 2:13-cv-07245, 2015 WL 11089495 (C.D. Cal. May 5, 2015)
`(Ex. 1267)
`
`(cid:120) Declaration of Dr. Michael Mitzenmacher (Ex. 2004)
`
`(cid:120) DVB-S2 User Guidelines (Ex. 2009)
`
`(cid:120) Declaration of Dr. Hui Jin (Ex. 2020)
`
`(cid:120) Declaration of Dr. Dariush Divsalar (Ex. 2031)
`
`(cid:120) Curriculum Vitae of Dr. Dariush Divsalar (Ex. 2032)
`
`13.
`
`I am being compensated at my normal consulting rate of $950 per hour
`
`for my work.
`
`14. My compensation is not dependent on and in no way affects the
`
`substance of my statements in this Declaration.
`
`15.
`
`I have no financial interest in Petitioner. I similarly have no financial
`
`interest in the ’710 patent.
`
`16.
`
`I have reviewed the Petition and the declaration of Dr. Davis and agree
`
`with their explanation of why the instituted claims are invalid. I have also reviewed
`5
`
`
`
`
`
`U.S. Patent No. 7,116,710
`Apple v. California Institute of Technology
`the institution decision and agree with the Board’s reasoning regarding the instituted
`
`claims. I have also read Caltech’s POPR, its POR, the declaration of Dr. Jin, and the
`
`declaration of Caltech’s experts, Drs. Mitzenmacher and Divsalar, and disagree with
`
`their challenges to the invalidity of the instituted claims.
`
`17.
`
`I understand that after submitting his declaration in this case, Dr. Davis
`
`relocated to Europe pursuant to a Fulbright Global Scholar Award. I further
`
`understand that he is unavailable to work on the Reply due to these professional
`
`obligations. As explained below, in my opinion the challenged claims are invalid.
`
`II. LEGAL PRINCIPLES
`
`18.
`
`I have been informed that a claim is invalid as anticipated under
`
`Pre-AIA 35 U.S.C. § 102(a) if “the invention was known or used by others in this
`
`country, or patented or described in a printed publication in this or a foreign country,
`
`before the invention thereof by the applicant for patent.” I have also been informed
`
`that a claim is invalid as anticipated under Pre-AIA 35 U.S.C. § 102(b) if “the
`
`invention was patented or described in a printed publication in this or a foreign
`
`country or in public use or on sale in this country, more than one year prior to the
`
`date of the application for patent in the United States.” Further I have been informed
`
`that a claim is invalid as anticipated under Pre-AIA 35 U.S.C. § 102(e) if “the
`
`invention was described in . . . an application for patent, published under section
`
`
`
`6
`
`
`
`U.S. Patent No. 7,116,710
`Apple v. California Institute of Technology
`122(b), by another filed in the United States before the invention by the applicant for
`
`patent . . . .” It is my understanding that for a claim to be anticipated, all of the
`
`limitations must be present in a single prior art reference, either expressly or
`
`inherently.
`
`19.
`
`I have been informed that a claim is invalid as obvious under Pre-AIA
`
`35 U.S.C. § 103(a):
`
`if the differences between the subject matter sought to be patented and
`
`the prior art are such that the subject matter as a whole would have been
`
`obvious at the time the invention was made to a person having ordinary
`
`skill in the art to which [the] subject matter pertains.
`
`20.
`
`I understand that a claimed invention would have been obvious, and
`
`therefore not patentable, if the subject matter claimed would have been considered
`
`obvious to a person of ordinary skill in the art at the time that the invention was made.
`
`I understand that when there are known elements that perform in known ways and
`
`produce predictable results, the combination of those elements is probably obvious.
`
`Further, I understand that when there is a predictable variation and a person would
`
`see the benefit of making that variation, implementing that predictable variation is
`
`probably not patentable. I have also been informed that obviousness does not
`
`require absolute predictability of success, but that what does matter is whether the
`
`
`
`7
`
`
`
`U.S. Patent No. 7,116,710
`Apple v. California Institute of Technology
`prior art gives direction as to what parameters are critical and which of many
`
`possible choices may be successful.
`
`III. DIVSALAR IN VIEW OF LUBY RENDERS CLAIMS 1-8 AND 11-14
`OBVIOUS
`
`A. Luby’s Table 1.
`
`21. Luby’s Table 1 is copied below.
`
`
`
`Ex. 1204 at 256.
`
`22. The lambda (λ) values in Luby’s Table 1 identify the fraction of edges
`
`of particular degrees. Id.; Ex. 1262 at 221:2-19. For example, in Code 14,
`
`approximately 49.6% of the edges in a Tanner graph for the code would have degree
`
`5. As an example, a Tanner graph for Luby’s code 14 is shown below.
`
`
`
`8
`
`
`
`U.S. Patent No. 7,116,710
`Apple v. California Institute of Technology
`
`
`
`Ex. 1247.
`
`23. A degree n edge is an edge that connects to a node of degree n. For
`
`example, in the above Tanner graph, the degree 5 edges are the edges that connect to
`
`the degree 5 message nodes. Ex. 1262 at 223:5-224:5.
`
`
`
`9
`
`
`
`U.S. Patent No. 7,116,710
`Apple v. California Institute of Technology
`24. The lambda (λ) values can be used to determine the percentage of each
`
`type of message node in the Tanner graph. Id. at 224:6-9. Using Luby’s Code 14
`
`and Luby’s disclosed example with 16,000 message bits and rate 1/2, there are
`
`112,000 edges in the Tanner graph. That is, each check node has degree 14 and there
`
`are 8,000 check nodes. So, the graph includes a total of 14 * 8,000 = 112,000 edges.
`
`25. Table 1 states that 49.6041% of the edges have degree 5. Multiplying
`
`0.496041 times 112,000 yields a fraction (55,556.592). Tanner graphs do not
`
`contain fractional edges, only an integer number of edges. Dr. Mitzenmacher
`
`explained that for this reason, some rounding is required to determine the number of
`
`edges of each degree. Id. at 221:20-223:4. Therefore, in this example, there are
`
`approximately 55,557 edges of degree 5. Using a similar procedure for the other
`
`degrees shows that for Luby’s 16,000 message node, rate 1/2 example of Code 14,
`
`there are 55,557 edges with degree 5; 19,473 edges with degree 6; 8,649 edges with
`
`degree 21; and 28,322 edges with degree 23.
`
`26. The percentage of nodes of degree n equals the number of edges with
`
`degree n divided by the degree n times the number of message nodes (which in this
`
`case is 16,000), i.e., if “fi” is the fraction of nodes with degree “i,” then
`
`fi=(number of edges with degree i) / (i) * 16,000
`
`10
`
`
`
`
`
`U.S. Patent No. 7,116,710
`Apple v. California Institute of Technology
`27. The table below shows the percentages and numbers of nodes of each
`
`degree for all the codes described in Luby’s Table 1, for Luby’s example of 16,000
`
`message bits and rate ½.
`
`Code
`14
`
`22
`
`10’
`
`14’
`
`
`
`Percentages of messages nodes of each degree
`Degree 5 Message Nodes: ~69% (~11,040 nodes)
`Degree 6 Message Nodes: ~20% (~3,200 nodes)
`Degree 21 Message Nodes: ~3% (~480 nodes)
`Degree 23 Message Nodes: ~8% (~1,280 nodes)
`Degree 5 Message Nodes: ~63% (~10,080 nodes)
`Degree 6 Message Nodes: ~23% (~3,680 nodes)
`Degree 27 Message Nodes: ~3% (~480 nodes)
`Degree 29 Message Nodes: ~4% (~640 nodes)
`Degree 30 Message Nodes: ~4% (~640 nodes)
`Degree 100 Message Nodes: ~3% (~480 nodes)
`Degree 3 Message Nodes: ~21% (~3,360 nodes)
`Degree 4 Message Nodes: ~69% (~11,040 nodes)
`Degree 16 Message Nodes: ~10% (~1,600 nodes)
`Degree 3 Message Nodes: ~22% (~3,520 nodes)
`Degree 4 Message Nodes: ~61% (~9,760 nodes)
`Degree 21 Message Nodes: ~5% (~800 nodes)
`Degree 23 Message Nodes: ~12% (1,920 nodes)
`
`A. Luby teaches irregular repetition of information bits.
`
`28. Caltech concedes that Luby teaches message nodes with irregular
`
`degree. POR at 19-20. However, it claims that, because Luby’s message nodes
`
`contain both information bits and parity bits, it is unclear whether the irregular
`
`message nodes result from irregular information bits, irregular parity bits, or both.
`
`11
`
`
`
`
`
`U.S. Patent No. 7,116,710
`Apple v. California Institute of Technology
`POR at 19-26. I disagree. A person of ordinary skill in the art (“POSA”) would
`
`have understood Luby to disclose irregular repetition of information bits.
`
`29. When each of Luby’s 16,000 message nodes are assigned to correspond
`
`to one of the 8,000 information bits or one of the 8,000 parity bits, the information
`
`bits will have different degrees. For example, in Luby’s Code 22, there would be
`
`some information bits of degree 100, some information bits of degree 30, and
`
`additional information bits of other degree. This is because, as Luby states, message
`
`nodes “with higher degree tend to their correct value quickly.” Ex. 1204 at 253.
`
`This would have led a POSA to preferentially use higher degrees for information
`
`bits. In Luby’s Code 22, there are only ~480 nodes of degree 100. Even if the POSA
`
`made all ~480 nodes of degree 100 information bits, he would still have ~7,520
`
`information bits to assign. Some of those information bits would be assigned a
`
`degree of 30, while others would be assigned different degrees. The information bits
`
`would thus be irregular. Even if a POSA were not motivated to preferentially use
`
`higher degrees for information bits, which is unlikely, there is no reason that a POSA
`
`would preferentially choose all information bits to have the same degree. A POSA
`
`would know that they are free to select information bits from a wide range of nodes
`
`with a wide range of degrees and that there is no reason that the information bit
`
`nodes should have constant degree.
`
`12
`
`
`
`
`
`U.S. Patent No. 7,116,710
`Apple v. California Institute of Technology
`30. Luby’s Code 14’ would also necessarily result in information bits of
`
`different degree. In Code 14’, there are only ~1,920 nodes with degree 23 (the
`
`highest degree). As a result, some of the information bits would have degree 23 and
`
`others would have degree 21. In other words, Luby teaches having information bits
`
`with different degrees.
`
`31. Also, even if a POSA were to randomly assign each information and
`
`parity bit to correspond to a particular node, it would have been highly unlikely for
`
`all the information bits to have the same degree. For example, in Luby’s Code 22,
`
`63% of the nodes have degree 5 and the remaining 37% of the nodes have a different
`
`degree. This means that if information and parity bits were assigned to nodes at
`
`random, for every 10 selected information bits, roughly 6 would have degree 5 and
`
`the other 4 would have a degree that is different from 5. That is, when making
`
`assignments at random, the information bits would have varying degree. What is the
`
`probability that all 8000 information bits would be assigned to nodes with the same
`
`degree? First, note that all nodes associated with information bits would need to
`
`have degree 5, because there are only 5,920 nodes with degree other than 5 (total of
`
`16,000 nodes minus 10,080 with degree 5). The probability that the first selected
`
`information bit would be associated with a node with degree 5 is 10,080 divided by
`
`16,000, that is, the number of unassociated nodes with degree 5 divided by the total
`
`13
`
`
`
`
`
`U.S. Patent No. 7,116,710
`Apple v. California Institute of Technology
`number of unassociated nodes. The probability that the second selected information
`
`bit would be associated with a node with degree 5 given that the first information bit
`
`has degree 5 is 10,079 divided by 15,999. And so on, until we compute the
`
`probability that the 8000th selected information bit would be associated with a node
`
`with degree 5 given that all preceding information bits have degree 5. That
`
`probability is 2081 divided by 8001. The probability that all information bits were
`
`assigned to nodes with the same degree is given by the product of these probabilities,
`
`which is 10-2,587.7. This probability is so vanishingly small that virtually every code
`
`generated by this process would repeat information bits with varying degrees. At a
`
`minimum, working through this analysis confirms that repeating information bits
`
`irregularly would have been obvious to a POSA from reading Luby.
`
`32. Also, when bits are assigned to nodes in one of Luby’s codes, all
`
`possibilities can be described by these two options: (a) some information bits have
`
`different degrees than others and (b) all information bits have the same degree (and
`
`as explained above, a POSA would have understood Luby to teach against this).
`
`Given that there are at most two options, option (a) would have been an obvious
`
`choice for a POSA.
`
`33. Thus, a POSA would have understood that Luby discloses information
`
`bits with different degrees.
`
`14
`
`
`
`
`
`U.S. Patent No. 7,116,710
`Apple v. California Institute of Technology
`B. Using Luby’s irregular degrees in Divsalar results in irregular
`repetition of information bits even if Luby alone does not disclose
`irregular repetition of information bits.
`
`34. Luby teaches irregular codes in which the degrees of codeword bits are
`
`irregular, i.e., some codeword bits have one degree and other codeword bits have a
`
`different degree. For example, in Luby’s “Code 14’” codeword bits have either
`
`degree 3, 4, 21 or 23. Luby teaches that such irregular codes outperform their
`
`regular counterparts.
`
`35. A POSA would have been motivated to use Luby’s irregularity in
`
`Divsalar to improve the performance of Divsalar’s code. Doing so would have
`
`resulted in repeating data “in different sub-blocks a different number of times” as
`
`required by the claims. In other words, applying Luby’s fundamental teaching – that
`
`irregular codes, with message bits that have irregular degrees, outperform their
`
`regular counterparts – to Divsalar results in a code that repeats information bits a
`
`different number of times and therefore meets the limitations of the instituted claims.
`
`36. Divsalar’s Figure 3 shows a coder with a repeater followed by an
`
`interleaver, which is then followed by an accumulator. The simplest way for a
`
`POSA to incorporate Luby’s irregularity into Divsalar would have been to modify
`
`Divsalar’s repeater to make it repeat some bits more than others. Doing so would
`
`have been simple, and would not have required modification of Divsalar’s other
`
`15
`
`
`
`
`
`U.S. Patent No. 7,116,710
`Apple v. California Institute of Technology
`components, i.e., the interleaver and accumulator. Incorporating irregularity into
`
`Divsalar’s repeater would only have required repeating some bits more than others.
`
`By contrast, a POSA would not have incorporated irregularity in Divsalar’s
`
`accumulator – doing so would have complicated the already simple accumulator.
`
`37. Divsalar’s code can be represented as a Tanner graph in which the
`
`degree of a message node equals the number of times the corresponding message bit
`
`is repeated. Exhibit 1246 shows such a Tanner graph for Divsalar’s code for the case
`
`in which q=3, i.e., the case in which the information bits are all repeated three times.
`
`Ex. 1247 shows the Tanner graph for Luby’s Code 14.
`
`16
`
`
`
`
`
`U.S. Patent No. 7,116,710
`Apple v. California Institute of Technology
`
`
`Ex. 1246
`
`
`
`
`
`
`
`Ex. 1247
`
`As shown, both Divsalar’s code and Luby’s code connect message nodes to check
`
`nodes using a random permutation. Divsalar’s coder includes the extra step shown
`
`at the right side of the Tanner graph, which corresponds to Divsalar’s accumulator.
`
`The left sides of the Tanner graphs are similar, i.e., they both include nodes and a
`
`random permutation. The difference is that Divsalar’s left nodes all have degree
`
`three (i.e., three edges intersect each node), while Luby’s left nodes have different
`
`degrees (i.e., Luby’s left nodes have degree 5, 6, 21 or 23). It would have been
`
`17
`
`
`
`
`
`U.S. Patent No. 7,116,710
`Apple v. California Institute of Technology
`obvious for a POSA to use Luby’s irregularity in Divsalar by making the degree of
`
`Divsalar’s information nodes irregular. Doing so exactly corresponds to Divsalar
`
`repeating some information bits more than others. As shown by the above Tanner
`
`graphs, Divsalar’s and Luby’s codes are similar and it would have been easy for a
`
`POSA to use Luby’s irregularity in Divsalar.
`
`38. Caltech claims that, because Luby’s irregularity is in the codeword bits,
`
`using Luby’s irregularity in Divsalar would mean changing the degree of the parity
`
`nodes at the right side of Divsalar’s Tanner graph. POR at 24-25, 28-29. I disagree.
`
`As I explain above, a POSA would not have been motivated to change Divsalar’s
`
`accumulator. A POSA likewise would not have been motivated to change the
`
`connections between Divsalar’s parity and check nodes. Changing Divsalar’s
`
`repeater to repeat bits a different number of times, which is the same as changing the
`
`degrees in Divsalar’s Tanner graph such that some information nodes are connected
`
`to more edges than others, would have been a simple and obvious way to use Luby’s
`
`irregularity in Divalar.
`
`39. That a POSA would have been motivated to use Luby’s irregularity by
`
`repeating Divsalar’s information bits irregularly is demonstrated by my own paper,
`
`Frey. Ex. 1202. Frey notes the success of Luby and goes on to describe a new code
`
`in which irregularity is incorporated into the code by repeating some information
`
`18
`
`
`
`
`
`U.S. Patent No. 7,116,710
`Apple v. California Institute of Technology
`bits more than others. Ex. 1202 at 241. My experimental results demonstrated that
`
`these irregular turbocodes perform better than the regular turbocodes that were
`
`known in the art. Id. at 246. A POSA reading Divsalar and Luby would have been
`
`motivated to make the Divsalar’s repetition irregular.
`
`40.
`
`I even suggested applying this same type of irregularity to Dr. Dariush
`
`Divsalar – the co-author of Divsalar – by e-mail on December 8, 1999. Ex. 1235,
`
`App. A. At that time, I was an active contributor and collaborator in the community
`
`that included some of the inventors of the ’710 patent. In particular, I attended talks
`
`given by Dr. Robert McEliece and Dr. McEliece attended talks that I presented.
`
`These talks included the 1998 and 1999 Allerton Conferences held by the University
`
`of Illinois Urbana-Champaign, among others. In the e-mail, I suggested “it would be
`
`interesting to extend the work that you and Bob [McEliece] have done to the case of
`
`irregular turbocodes.” Id. This is a clear suggestion to the author of Divsalar to
`
`combine the RA codes of Divsalar with the irregularity discussed in my paper, Frey,
`
`which itself was motivated by Luby. The email said only what would have been
`
`obvious to a POSA, i.e., that it would have been obvious to make Divsalar’s
`
`repetition of information bits irregular.
`
`19
`
`
`
`
`
`U.S. Patent No. 7,116,710
`Apple v. California Institute of Technology
`C. Divsalar in view of Luby discloses the “partitioning” limitation.
`
`41. By repeating some information bits a different number of times than
`
`others, the combination of Divsalar and Luby “partition[s] said data block into a
`
`plurality of sub-blocks” – e.g., a sub-block with bits repeated one number of times
`
`and another sub-block with bits repeated a different number of times.
`
`42.
`
`I do not understand Caltech to dispute that repeating bits a different
`
`number of times would disclose partitioning them into different sub-blocks, or that
`
`such partition would have been an obvious implementation of the irregular repeater
`
`of Divsalar in view of Luby. See POR at 29-30. Instead, I understand Caltech to
`
`contend that it is theoretically possible to “implement the modified Divsalar code
`
`without partitioning.” Id. Even if theoretically true, a POSA would not implement
`
`the modified Divsalar code without partitioning.
`
`43. Dr. Mitzenmacher claims that a POSA implementing Divsalar and
`
`Luby could avoid partitioning by using a random number generator to determine
`
`how many times to repeat each bit. Ex. 1262 at 381:3-382:13. But, even if such a
`
`random number generator were used, its output sequence would be known in
`
`advance. As a result, that sequence would provide the claimed partitioning because
`
`it would define in advance of repeating exactly which bits are repeated exactly
`
`which number of times.
`
`
`
`20
`
`
`
`U.S. Patent No. 7,116,710
`Apple v. California Institute of Technology
`44. Even if it were possible to alter Divsalar’s encoder to irregularly repeat
`
`without partitioning, the combination of Divsalar and Luby would still disclose
`
`partitioning because that is what a POSA would have reasonably understood from
`
`the disclosure. Dr. Mitzenmacher acknowledges his proposed implementation is
`
`merely a hypothetical that “[o]ne could envision.” Ex. 2003, ¶94. He offers no
`
`testimony that this implementation is a practical or likely implementation. Id. And,
`
`in my opinion such an implementation is not practical or likely. A POSA would
`
`instead have implemented irregular repetition of information bits in Divsalar by
`
`simply repeating some bits a certain number of times and other bits a different
`
`number of times. In doing so, the POSA would have partitioned the block of
`
`information bits into different sub-blocks as recited in the claims. This is exactly
`
`what I did in my own work where I partitioned the input bits into sub-blocks f1
`
`through fD. Ex. 1202 at 244.
`
`D. One of ordinary skill would have combined Divsalar and Luby.
`
`45.
`
`It would have been obvious to combine Divsalar and Luby for multiple
`
`reasons.
`
`46. Luby expressly states that irregularity could be used to design codes
`
`that are “far superior to any regular code.” Ex. 1204 at 257. Caltech claims that
`
`Luby’s teaching is only applicable to “Gallager” codes. POR, 30-32, 35. A POSA
`
`21
`
`
`
`
`
`U.S. Patent No. 7,116,710
`Apple v. California Institute of Technology
`would not have read Luby so narrowly, as shown by my paper, Frey, which applies
`
`Luby’s teaching to turbo-like codes. In fact, a POSA would have been encouraged
`
`to use Luby’s irregular degrees in other codes. I did exactly that and, noting the
`
`success of Luby’s irregularity, went on to incorporate irregularity in a code by
`
`irregularly repeating information bits. Ex. 1202 at 241-244. A POSA would have
`
`similarly been motivated to do what I did and obtain the benefit of Luby’s
`
`irregularity in other codes such as Divsalar’s – just as I encouraged Dr. Divsalar to
`
`do in 1998. Ex. 1235, App. A.
`
`47. Caltech states that Luby defines irregularity with respect to the lambda
`
`(λ) values in Table 1. POR at 25-26. But as I showed above, those lamba (λ) values
`
`expressly disclose irregular information bits. Thus, contrary to Caltech’s assertion,
`
`Table 1 shows that Luby’s use of irregularity is consistent with the definition offered
`
`by Dr. Davis and with the understanding of those in the art.
`
`48. Caltech does not dispute that Divsalar could be made irregular by
`
`modifying the repeater to repeat different information bits a different number of
`
`times. It likewise does not dispute that Divsalar could be made irregular by
`
`modifying its Tanner graph by redistributing a few edges. Instead, Caltech argues
`
`that such modifications were not sufficiently described and would not necessarily
`
`22
`
`
`
`
`
`U.S. Patent No. 7,116,710
`Apple v. California Institute of Technology
`result in desired performance for particular applications or have a reasonable
`
`expectation of success. POR at 37-47. I disagree.
`
`49. Rigorous mathematical analysis of codes is difficult, and, as a result,
`
`POSAs routinely developed codes by experimentation. POR at 2. Encouraged by
`
`Luby’s results, a POSA would have been motivated to tr