`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`____________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`____________
`
`UNIFIED PATENTS, INC.,
`Petitioner,
`v.
`
`INTELLECTUAL VENTURES I, LLC,
`Patent Owner.
`____________
`
`Case IPR2016-01643
`Patent 6,775,745 B1
`____________
`
`Record of Oral Hearing
`Held: January 4, 2018
`____________
`
`
`
`
`
`
`Before JEFFREY S. SMITH, GEORGIANNA W. BRADEN, and
`DANIEL J. GALLIGAN, Administrative Patent Judges.
`
`
`
`
`
`
`
`
`
`
`
`
`
`Case IPR2016-01643
`Patent 6,775,745 B1
`
`APPEARANCES:
`
`ON BEHALF OF PETITIONER:
`
`ROSHAN MANSINGHANI, ESQUIRE
`
` JONATHAN STROUD, ESQUIRE
`Unifed Patents, Inc.
`13355 Noel Road
`Suite 1100
`Dallas, Texas 75240
`(214) 945-0200
`
`
`ON BEHALF OF PATENT OWNER:
`
`
` BRENTON R. BABCOCK, ESQUIRE
`Knobbe, Martens
`2040 Main Street.
`14th Floor.
`Irvine, California 92614
`(949) 760-0404
`
`
`
`
`
`
`
`The above-entitled matter came on for hearing on Thursday, January
`
`4, 2018, commencing at 2:05 p.m., at the U.S. Patent and Trademark Office,
`600 Dulany Street, Alexandria, Virginia.
`
`
`
`
`
`2
`
`
`
`Case IPR2016-01643
`Patent 6,775,745 B1
`
`
`P R O C E E D I N G S
`- - - - -
`JUDGE SMITH: Good afternoon. Welcome to the Patent Trial and
`Appeal Board. We're here this afternoon for a hearing in inter partes review
`matter IPR2016-01643. This is the case where Unified Patents is the
`Petitioner and Intellectual Ventures is the Patent Owner. I'd like to start by
`getting appearances of counsel. Petitioner, please step up to the podium and
`make your appearance.
`MR. MANSINGHANI: Roshan Mansinghani for Unified Patents.
`JUDGE SMITH: And who do you have with you?
`MR. MANSINGHANI: With me I have Shawn Chang from the
`Finnegan, Henderson Law Firm. James Stein, Finnegan, Henderson Law
`Firm, and Jonathan Stroud, Unified Patents.
`JUDGE SMITH: Thank you. Who do we have on behalf of Patent
`Owner?
`MR. BABCOCK: Good afternoon, Your Honors. I'm Brenton
`Babcock with Knobbe, Martens representing Intellectual Ventures. With me
`is backup counsel Ted Cannon also from Knobbe, Martens, and also Russ
`Rigby, inhouse counsel of Intellectual Ventures.
`JUDGE SMITH: Thank you. I'd like to go over a few administrative
`details quickly before we begin, about the structure of the hearing today.
`Our trial hearing order indicated that there will be 45 minutes argument for
`each side. Petitioner will go first, you will present your case in brief. Patent
`Owner, you will then be allowed to respond to Petitioner. Petitioner, if you
`wish you may reserve time for rebuttal. Do you wish to do so?
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`
`
`3
`
`
`
`Case IPR2016-01643
`Patent 6,775,745 B1
`
`
`MR. MANSINGHANI: Yes, Your Honor, about 15 minutes, Your
`Honor.
`JUDGE SMITH: 15 minutes for rebuttal, thank you. One more
`administrative detail, when you're referencing your slides please keep in
`mind to reference the particular slide number so we can keep track of which
`slide you're discussing here during the hearing. Also, when we're reading
`the transcript we can identify the slide that's being referenced. Let me set
`the clock for 30 minutes. Petitioner, when you're ready you may began.
`MR. MANSINGHANI: Good afternoon Panel. As I've mentioned
`my name is Roshan Mansinghani. I work for Unified Patents and I'm
`pleased to present to you our case today as to why we believe the challenged
`claims of US Patent Number 6,775,745 should be cancelled. Turning to
`slide 3, I'd like to start with the brief background, technological background
`and prosecution history background, to set the stage for why we believe
`these patents are invalid and should not have been granted.
`As the Panel understands this case is about caching technology. And
`as the briefing and the expert reports have made clear caches are about
`taking files that are on a slower hard drive and putting them onto a cache,
`which is a faster hard drive or other type of media, so you can get faster
`accesses to the files. The '754 patent addresses two aspects of caching
`technology. Simply the '745 patent addresses how do you decide what files
`to put into the cache, and then how do you decide when the cache is full and
`you need to make room for new files how do you decide what to take out of
`the cache.
`Turning to slide 4, we've got summarized the two problems that they
`addressed on more detail. One is the typical manner of putting files into the
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`4
`
`
`
`Case IPR2016-01643
`Patent 6,775,745 B1
`
`cache involve what the '745 patent refers to as smaller reads. That is, for
`example, if a user requested a portion of a file you would just bring that
`portion in and nothing else. But, the '745 patent thought, well, perhaps
`there's a more intelligent way to do it. Maybe you can bring in more parts of
`the file.
`The second aspect, again, is about how do you decide what to take out
`of the cache, and the '745 patent references a most recently used-least
`recently used algorithm, MRU-LRU. Most of the other literature just refers
`to it as an LRU algorithm, so I'll do so here today. In this algorithm the '745
`patent has background discussing that the way you decide what to take out
`of the cache is you just take out the data that has been accessed least
`recently.
`Or, in other words, it's been the longest time since you have referred
`to that file. You can see this in the summary of the invention section of slide
`5. There we have a quote from column 3, lines 46 through 52 of the patent
`and here we have a summation, a summary, of what is claimed in the
`independent claims. You can break this down into two areas that the '745
`patent alleges that brought novelty to the case at hand. The first is
`enhancing system performance using a hybrid caching mechanism and this
`caching mechanism is one -- this is referring to, excuse me, how you decide
`what to take out of the cache once the cache is full.
`So, as I mentioned, if you just think about recency the '745 patent
`suggests that frequency also should be taken into account. In other words,
`you shouldn't just think about when was the last time this file was accessed
`you should also think about how many times in the past has this file been
`accessed. And, in the middle of that excerpt there's a sentence that starts
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`5
`
`
`
`Case IPR2016-01643
`Patent 6,775,745 B1
`
`with "in addition" and there you see the second piece of the '745 patent's
`alleged novelty relating to how do you decide what to put into the cache.
`So, the '745 patent suggests instead of just taking a small portion of
`data let's take an extended larger read. So, in other words, taking data
`beyond what the user requested in hopes that the user will request that data
`in the future sometime and you already have it in the cache. So, on slide 6
`we map this to the actual -- one of the claims at issue, Claim 1. In yellow
`you have the reading extended segments of data limitation which refers to,
`again, what do you choose to put into the cache, and in red we have the
`hybrid caching mechanism referring to the algorithm that the '745 patent
`discloses on what do you take out of the cache taking into account both
`frequency and recency.
`On slide 7 though we believe it's important for the Board to
`understand the context of how this patent was actually allowed. What we
`have here is an excerpt and it may be off of your slides, we may have
`forgotten the citation, but it's Exhibit 1003, page 73 and 77 is where we took
`this from. This is an excerpt from the office action that was the last office
`action before there was allowance. And what you see in paragraph 10 of this
`excerpt is that the examiner had rejected the then pending independent
`claims.
`Those then pending independent claims only had the hybrid caching
`limitations, the frequency factor limitations as we've been discussing in the
`briefing. That is it only had the teachings about what do you -- how do you
`take things out of the cache. The dependent claims that appear in Claim 12,
`some of those and particularly Claim 2, had the reading extended segments
`of data limitation. So, what the examiner presented in this office action was
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`6
`
`
`
`Case IPR2016-01643
`Patent 6,775,745 B1
`
`that the hybrid caching mechanism that's considering recency and frequency
`was already known.
`But that the reading extended segments of data limitation, what do
`you put into the cache, was not before the examiner and suggested those
`would be allowable. On slide 8, in summary, we have the results of what
`occurred in the file history. The applicant did not contest the examiner's
`office action. The applicant amended the independent claim to include the
`reading extended segments of data limitation and did not contest that the
`prior art identified by the examiner taught all of the other limitations.
`What does that tell us today, that tells us today that the only thing that
`got this patent allowed was adding in the reading extended segments of data
`limitation. All of the details regarding the frequency factor, which has been
`the bulk of the briefing in this case, were things that the applicant had
`already acquiesced was in the prior art. Importantly on this slide we note
`that Karedla explicitly discloses the reading extended segments of data
`limitation. In fact, the Patent Owner doesn't even contest that this is the
`case.
`Rather, the only contest that the Patent Owner brings to ground one is
`that Karedla doesn't explicitly disclose a combination of the read-ahead
`strategy with the cache removal algorithms, they're wrong on that, we can
`show you why. But importantly for the second ground, the ground with
`Burton and Karedla, they do not contest at all that it's an appropriate
`combination of Karedla's prefetching, which is after the reading extended
`segments of data limitation along with Burton's algorithm.
`JUDGE BRADEN: So, counselor, when I read the prosecution
`history I read that in the responses to the office action that the applicant
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`7
`
`
`
`Case IPR2016-01643
`Patent 6,775,745 B1
`
`traversed the rejections and was willing to rewrite the dependent claims in
`order to put them in allowable format in order to get those claims allowed.
`How do you read that as acquiesce to the examiner's characterization?
`MR. MANSINGHANI: My understanding of the file history, Judge
`Braden, is that they did not present any arguments. They may have used the
`word "traverse" but there has been case law that says that merely just using
`that magic language doesn't turn into actually arguing that that prior art was
`not there. The bottom line is they presented no arguments that that prior art
`that the examiner cited was different than what their claims had at that time.
`JUDGE BRADEN: And in that case law that you're referencing was
`there anything as far as efficiency during prosecution or if they don't put in
`any kind of arguments that they're estopped from later arguing it?
`MR. MANSINGHANI: I don't have all the details of that case law in
`my mind right now. Judge Braden had said something that I would have to
`do further research on I'd be more than happy to submit after the hearing if
`you would like.
`JUDGE BRADEN: No, that's okay, thank you.
`MR. MANSINGHANI: Thank you. So, moving to slide 10 we have
`an overview of the grounds before you. And, again, we just want to note
`that with respect to the Burton and Karedla ground Patent Owner does not
`contest the reading extended segments of data limitation nor its combination
`with Burton. The only argument regarding combining the prefetching of
`Karedla with the cache replacement algorithms is within the anticipation
`ground itself.
`I'd like to now turn to claim construction, turning to slide 11, here, we
`have a summary of the various positions. As the Board recalls the Patent
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`8
`
`
`
`Case IPR2016-01643
`Patent 6,775,745 B1
`
`Owner attempted to construed the term "frequency factor" to essentially
`require that you had to be able to look at the frequency factor and be able to
`tell the actual number of times a file had been accessed or something close
`to it. The Board, we think, properly rejected that in its institution decision
`and instead construed the frequency factor as we have here on the slide.
`Patent Owner, then in their response after the institution decision, is
`now trying to construed the term "indicating" and "reflecting" to essentially
`require the same things that they were seeking in their preliminary response.
`We think that's inappropriate for several reasons and we've got those
`summarized on slide 12. All of these points, these six points, are laid out in
`our briefing. I'm going to focus on just a few of them for today, but I
`thought this might be helpful for you to see a summary of why the
`construction is incorrect.
`On slide 14, we have perhaps one of the most important pieces of
`evidence in the claim construction dispute. As Patent Owner see it there is a
`-- Patent Owner wants to construe "indicate," the word "indicate" that
`appears in Claim 4, different than the word "reflect" that occurs in Claim 1.
`This is inappropriate. Here, we have direct evidence from the specification.
`We have a citation to column 2, lines 46 through 48, and column 2, lines 60
`through 61 and the term frequency factor is being discussed here in the
`specification.
`You can see that the discussion is nearly identical except for the two
`words "indicate" and "reflect". And, thus, the specification uses these terms
`interchangeably, there is no difference. The Patent Owner has committed
`one of the cardinal sins of claim construction by choosing a dictionary
`definition, ignoring the specification, and then trying to conform the
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`9
`
`
`
`Case IPR2016-01643
`Patent 6,775,745 B1
`
`specification to a dictionary definition. That is inappropriate especially
`under the broadest reasonable interpretation standard.
`JUDGE SMITH: How do we know these words are being used
`interchangeably rather than it's a conscious effort on the part of the person
`who wrote this application to use the different words to provide the different
`meanings?
`MR. MANSINGHANI: Thank you, Judge Smith. Two ways we
`know that. First, when you read the complete context of the summary of the
`invention section they're talking about the very same features. These are
`two different paragraphs of the summary in the invention section, they're
`talking about this same features and that's how we know that they're being
`used interchangeably.
`There's no distinction, especially the distinction that Patent Owner is
`conjuring up, that one means -- one requires a fair degree of certainty and
`one does not. That distinction is not here in the context of it, that's the first
`way we know it. The second way we know it is because the term "fair
`degree of certainty," the term that Patent Owner contends somehow
`distinguishes these two terms, nowhere appears in the specification. Rather
`it was a cherry picked phrase from a vast amount of dictionary definitions
`that both parties have provided. On slide 16 --
`JUDGE SMITH: If the intent of the Patent Owner is to use these
`words to mean the same thing why wouldn't the same word be used
`consistently throughout the claim set? Why would one independent claim
`have one word and another independent claim have a different word? Why
`not use "indicate" or "reflect" throughout the claim? Why use indicate in
`Claim 4 and reflect in Claim 1?
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`10
`
`
`
`Case IPR2016-01643
`Patent 6,775,745 B1
`
`
`MR. MANSINGHANI: That's a good question Judge Smith. When
`we look at the specification to try to answer that question that is no apparent
`reason. It is just clear that as is common in English writing you sometimes
`use different words synonymously in the writing and that's what it appears
`that they did here. And the other point I would make, to the extent that
`there's any issue with claim differentiation that may be in the Panel's minds,
`I'll note that the doctrine of claim differentiation has very little weight in this
`context.
`And there's case law to support this because first of all, it's not -- the
`two words are not being used in the same claim. They're being used in
`different claims, and in those different claims there are different limitations
`in those claims. So, claim differentiation doesn't hold and doesn't provide
`the amount of support that Patent Owner would need to try to, again,
`manufacture a difference where none really exists. On slide 15 though, we
`have another important reason why this construction should be rejected, and
`that is it excludes embodiments, unduly excludes embodiments, from the
`specification.
`Here, we have an excerpt from column 7, lines 8 through 10. This
`section of the specification discusses tuning, a form of adjusting aspects of
`the cache, and it specifically says you can change the frequency factors.
`You can weight them, you can decrement them. You can weight them on
`their first entrance into the cache, meaning when the file first gets into the
`cache, according to the Patent Owner it should be something like the number
`1 because its being accessed for the first time.
`But the specification teaches that you can use something like
`weighting on the initial entry to even turn that number into 2 or 10 or 20,
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`11
`
`
`
`Case IPR2016-01643
`Patent 6,775,745 B1
`
`whatever the designer prefers so their construction would exclude this
`embodiment. What's interesting is that both independent claims,
`Independent Claim 1 and Claim 4, have corresponding dependent claims that
`claim this tuning embodiment. So, the independent claims must be broad
`enough to include it since the dependent claim specifically claims it, and yet
`their construction would exclude it.
`On slide 16 we have a key admission from the Patent Owner's expert
`which we think shows where Patent Owner is coming from with respect to
`this construction. We were asking the Patent Owner's expert what does it
`mean -- what does a fair degree of certainty mean, and instead of giving us
`an answer he says it depends. It's not the purpose of my analysis to draw the
`line between where we have a fair degree of certainty and where we don't.
`It's merely the purpose of my analysis to show that the prior art, which is
`Burton, does not meet that standard. This is result-oriented tactics for claim
`construction rather than looking at the intrinsic record. That was from
`Exhibit 1017, page 43, lines 3 through 9.
`JUDGE BRADEN: Counselor, if we were to accept Patent Owner's
`claim construction, would Karedla still anticipate the claims?
`MR. MANSINGHANI: Yes, Judge Braden, both of Karedla's
`algorithms would anticipate the claims. Our reply lays that out completely
`and if you'd like I could turn there and we could discuss how that's the case.
`JUDGE BRADEN: If you're going to get to it that'd be great.
`Otherwise, I'd like to talk about it at least at some point during your
`presentation.
`MR. MANSINGHANI: Thank you, thank you. I'll get to it very
`shortly then Judge Braden.
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`12
`
`
`
`Case IPR2016-01643
`Patent 6,775,745 B1
`
`
`JUDGE BRADEN: That's fine, thank you.
`MR. MANSINGHANI: So, what we see here from the Patent
`Owner's expert is that he couldn't tell us what a fair degree of certainty is so
`the result of that is that they have created a construction that is ambiguous
`and that frankly is only up to them to decide what qualifies as a fair degree
`of certainty and what does not. That's another reason why it should be
`rejected. Let me now turn to Judge Braden's question about would Karedla
`anticipate even if Patent Owner's construction was adopted and the answer
`to that is, yes. Patent Owner's construction deals with the algorithm that you
`choose to --
`JUDGE BRADEN: And what slide are you going to?
`MR. MANSINGHANI: Yeah, I'm about to tell you, Judge Braden.
`We're going to start at slide 27, I was flipping there myself. Patent Owner's
`construction relates to, again, not the reading extended segments of data
`limitation, which was the key limitation that lead to allowance, but rather the
`limitations that are directed towards removing items from the cache. And
`we've identified two algorithms disclosed in Karedla, explicitly disclosed in
`Karedla, that teach these things.
`The first one is called frequency-based LRU replacement algorithm
`and it's discussed in the Karedla reference at page 8. On the right-hand side
`of our slide you can see we've highlighted the title of the algorithm. For
`shorthand we refer to it as the FBR algorithm. Now, the context of the
`frequency-based LRU replacement algorithm was discussed on the same
`page in Karedla and we've got that on the left-hand side. Karedla introduces
`the concept of frequency-based algorithms and notes that there's a frequency
`of access count or a frequency count or a reference count.
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`13
`
`
`
`Case IPR2016-01643
`Patent 6,775,745 B1
`
`
`This is the count that the institution decision appropriately mapped to
`the frequency factors of the claims. In the right hand column -- or
`right-hand side of the slide we now talk about the frequency-based LRU
`replacement algorithm. And what this is a variation on LRU were they now
`try to take the frequency into account. And here you see we've highlighted
`that they used the reference count of lines but they do use it slightly
`differently. Here, they say unlike an LFU the reference count of lines
`repeatedly referenced in the top section is not incremented.
`So, every line has a reference count but only those in the top section
`does not. So, what are these sections, let me ask you to turn to slide 28 to
`help me explain that for you. This is an excerpt from our expert's
`supplemental declaration, Exhibit 1016 at paragraphs 97 through 107. Here,
`our expert worked out an example taken directly from the teachings of
`Karedla. In the previous slide, on Karedla page 8, it said that the cache is
`partitioned into three sections.
`So, on slide 28 you can see we've colored the three sections in red,
`yellow, and green. That doesn't necessarily mean one stops and one goes,
`it's just a way to distinguish between the three sections. And what we did
`was we used an exemplary set of accesses and filled it in. If you go to page
`29 we have the final state -- slide 29, excuse me -- we have the final state of
`the cache. So, if we look at this term, if we look at this cache what we see is
`that every single entry in this example, it's frequency count matches exactly
`the number of times it's accessed, access count, except is one, letter C.
`Letter C was actually accessed twice but here it only is accessed once,
`but what does this tell us. This tells us, one, that unlike Patent Owner's
`contentions every file is given a frequency count line just as the claim
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`14
`
`
`
`Case IPR2016-01643
`Patent 6,775,745 B1
`
`requires. It's assigned as soon as it comes in, it just may not be updated from
`time to time. But what this example illustrates is that the lines move out of
`the top section. As other files come in they move out of the top section, and
`so letter C is the only one that stayed in the top section long enough to have
`this issue of being incremented or not being incremented applied to it.
`So, what does this tell us, this shows us that even under the FBR
`algorithm explicitly disclosed by Karedla the lines were one hundred percent
`accurate except for one, and in that one case it was only off by the value one.
`Under Patent Owner's interpretation of their own construction they state that
`the frequency count does not need to be a one-to-one correspondence. Small
`variations are allowed and the reason why they have to do that is because the
`specification teaches that you can vary the frequency count numbers -- the
`frequency factor, excuse me.
`For example, in the specification if we go back to the claim
`construction section where we had that citation to the specification, I'm
`going to show you exactly an example from the specification that's close to
`matching what Karedla discloses. So slide 15 at column 7, lines 8 through
`10 this idea of tuning is discussed. And it says, other features which may
`enhance the tuning include how often to decrement the factors. You can
`actually reduce them and, for example, you can choose them to reduce them
`by one.
`So, in the patent's example you can have a frequency factor who is off
`by just one value, one integer value, from its actual access count. Well,
`that's exactly what we've shown that Karedla's algorithm would do. So, not
`only does Karedla's FBR algorithm and its frequency count disclose therein,
`turning back now to slide 27, not only does that algorithm meet the Board's
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`15
`
`
`
`Case IPR2016-01643
`Patent 6,775,745 B1
`
`appropriate construction from its institution decision because it's clearly
`based on the number of times a file was accessed, but it even meets Patent
`Owner's unduly narrow construction because Patent Owner would concede
`that there is a small amount of variation allowed for. And the key is, Your
`Honor --
`JUDGE SMITH: Can I ask you looking at slide 28, let me just make
`sure I understand this.
`MR. MANSINGHANI: Of course, Judge Smith.
`JUDGE SMITH: In slide 27 you highlighted that the top section,
`which is the most recently used section, the reference count is not
`incremented, is that right?
`MR. MANSINGHANI: Yes, Your Honor.
`JUDGE SMITH: And then in slide 28, it looks like the top section,
`the count is being incremented?
`MR. MANSINGHANI: Yes, Your Honor. So, let me explain how
`this works. So, when it says that it's not incremented not that it's not
`assigned. It is assigned, every file needs a reference count and that's taught
`explicitly in Karedla. Rather it says if you're in the top section you're not
`going to be incremented. In other words, if there's another file accessed, that
`same file, the frequency count would go up. Not that it would be there just
`that it won't go up. So, in the -- when track A first comes into the cache it's
`at the very top and it gets a frequency count of one.
`JUDGE SMITH: I see.
`MR. MANSINGHANI: It's just entered the cache.
`JUDGE SMITH: I see, okay.
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`
`
`16
`
`
`
`Case IPR2016-01643
`Patent 6,775,745 B1
`
`
`MR. MANSINGHANI: Track B comes in and track A moves down
`because this is an LRU --
`JUDGE SMITH: I see. I see, okay.
`MR. MANSINGHANI: Does that answer your question, Your
`Honor?
`JUDGE SMITH: Yes.
`MR. MANSINGHANI: Thank you. So, let's move to slide 30.
`We've seen here that Patent Owner completely mischaracterized how
`Karedla works. In our reply we properly rebutted this issue and slide 31 has
`another key legal principle that Patent Owner completely ignores in its
`briefing. Patent Owner's protestations on all of the frequency factor
`limitations regarding the prior art is simply that they can come up with
`contrived examples where maybe the frequency factor does not track well
`with the actual access count.
`But we have provided examples where it does, and our examples are
`actually in the mainstream, their examples are the exceptions. Regardless, at
`best they've shown is that the algorithms that we've identified do not always
`embody the claim method under their construction. Clearly under the
`Board's construction it does. The only dispute here is if we chose to adopt
`their construction which is wrong, but even if we do adopt their construction
`at best all they've done is said that it doesn't always embody the claim
`method, but that's not the legal standard either for anticipation or for
`obviousness.
`In the Hewlett-Packard case we've cited on slide 31, 340F.1314 at
`1326, the Federal Circuit in longstanding precedent is clear that only
`sometimes is enough and that's what we've demonstrated. Only sometimes
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`17
`
`
`
`Case IPR2016-01643
`Patent 6,775,745 B1
`
`is enough and they have no response to this. I see that I have a few minutes
`left. Before I was going to step down for my rebuttal I wanted to pause and
`ask if there were any other specific questions the Board had before I moved
`on with my presentation. If there's any other issues you wanted me to
`address.
`JUDGE GALLIGAN: This is Judge Galligan, will you get to the
`mixing of embodiments arguments that --
`MR. MANSINGHANI: Yes, let me address that now, thank you.
`JUDGE GALLIGAN: Great, thanks.
`MR. MANSINGHANI: So, Patent Owner contends that with respect
`to the reading extended segments of data limitations -- slide 23 please, to set
`the stage for this. So, here, the issue is does Karedla disclose the reading
`extended segments of data limitation as we note on this slide and as Patent
`Owner must confess they don't actually dispute whether it's disclosed. They
`just dispute whether it's actually part of or could be combined with the other
`cache replacement algorithms.
`On slide 25, we can turn there, there is clear evidence in Karedla itself
`that the algorithms it discloses are combinations of LRU, LFU, and
`read-ahead strategies with varying thresholds on the amount of data that is
`prefetched. This was in our petition, this is what the Board relied on in its
`institution decision, but they argued it doesn't say the specific algorithms
`that we identified, the frequency-based LRU and the segment, and the
`segmented LRU algorithm. Well, they're wrong on that as well.
`Slide 26 has evidence that we cited in our reply to rebut their
`erroneous contention. On page 9 of Karedla a device called the Hierarchical
`Storage Controller, HSJ, is described and prefetching is one of its features.
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`
`
`18
`
`
`
`Case IPR2016-01643
`Patent 6,775,745 B1
`
`On page 10 when it's describing that same HSJ device it says it uses
`segmented LRU, one of the algorithms that we have identified. So, Patent
`Owner's completely wrong, they've mischaracterized Karedla, it does -- not
`only discloses the general combination which, by the way, under the case
`law as our reply demonstrates is enough -- we actually have a specific
`combination here of the specific algorithms that we relied on.
`The other thing I want to note about this combination arguments
`Judges is that we have a second ground, Karedla in view of Burton. And as
`I've mentioned earlier the Patent Owner does not contest that Karedla's
`prefetching could be combined as we've explicitly stated in the petition with
`Burton in our obviousness ground. They don't contest that at all. So, even if
`you were somehow inclined to say there wasn't enough explicit disclosure in
`Karedla, which there is, our ground in both the Burton ground -- the
`Karedla-Burton ground is there and they don't contest that combination issue
`at all there. It's only with respect to the anticipation ground. Are there any
`other issues that were pressing for the Panel? I wasn't going to hit that one
`so I'm glad you mentioned it, Judge Galligan. If not I'm going to go ahead
`and sit down now and reserve the rest of my t
Accessing this document will incur an additional charge of $.
After purchase, you can access this document again without charge.
Accept $ Charge
Still Working On It
This document is taking longer than usual to download. This can happen if we need to contact the court directly to obtain the document and their servers are running slowly.
Give it another minute or two to complete, and then try the refresh button.
A few More Minutes ... Still Working
It can take up to 5 minutes for us to download a document if the court servers are running slowly.
Thank you for your continued patience.
This document could not be displayed.
We could not find this document within its docket. Please go back to the docket page and check the link. If that does not work, go back to the docket and refresh it to pull the newest information.
Your account does not support viewing this document.
You need a Paid Account to view this document. Click here to change your account type.
Your account does not support viewing this document.
Set your membership
status to view this document.
With a Docket Alarm membership, you'll
get a whole lot more, including:
- Up-to-date information for this case.
- Email alerts whenever there is an update.
- Full text search for other cases.
- Get email alerts whenever a new case matches your search.
One Moment Please
The filing “” is large (MB) and is being downloaded.
Please refresh this page in a few minutes to see if the filing has been downloaded. The filing will also be emailed to you when the download completes.
Your document is on its way!
If you do not receive the document in five minutes, contact support at support@docketalarm.com.
Sealed Document
We are unable to display this document, it may be under a court ordered seal.
If you have proper credentials to access the file, you may proceed directly to the court's system using your government issued username and password.
Access Government Site
