`
`571-272-7822
`
`
`
`
`
`
`IPR2015-02001, Paper No. 40
`IPR2016-00157, Paper No. 23
`February 1, 2017
`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`____________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`____________
`
`PALO ALTO NETWORKS, INC.,
`Petitioner,
`
`v.
`
`FINJAN, INC.,
`Patent Owner.
`____________
`
`Case IPR2015-02001
`Case IPR2016-001571
`Patent 8,225,408 B2
`____________
`
`Held: January 5, 2017
`____________
`
`
`
`
`
`BEFORE: THOMAS L. GIANNETTI, MIRIAM L. QUINN, and
`PATRICK M. BOUCHER, Administrative Patent Judges.
`
`The above-entitled matter came on for hearing on Thursday,
`January 5, 2017, commencing at 1:30 p.m., at the U.S. Patent and
`Trademark Office, 600 Dulany Street, Alexandria, Virginia.
`
`
`
`
`1 Case IPR2016-00157 has been consolidated with IPR2015-
`02001 (“the consolidated proceeding”). Cases IPR2016-00955
`and IPR2016-00956 have been consolidated and joined with the
`consolidated proceeding.
`
`
`
`Case IPR2015-02001
`Patent 8,225,408 B2
`
`APPEARANCES:
`
`
`
`ON BEHALF OF THE PETITIONER:
`
`
`
`
`
`
`
`
`
`
`
`BRIAN EUTERMOSER, ESQUIRE
`ORION ARMON, ESQUIRE
`
`Cooley LLP
`380 Interlocken Crescent
`Suite 900
`Broomfield, Colorado 80021-8023
`
`JAMES HANNAH, ESQUIRE
`AAKASH JARIWALA, ESQUIRE
`SHANNON H. HEDVAT, ESQUIRE
`Kramer, Levin, Naftalis & Frankel, LLP
`990 Marsh Road
`Menlo Park, California 94025-1949
`
`
`
`
`
` 2
`
`ON BEHALF OF PATENT OWNER:
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Case IPR2015-02001
`Patent 8,225,408 B2
`
`
`P R O C E E D I N G S
`- - - - -
`JUDGE GIANNETTI: So this is the second hearing
`today. This case is IPR2015-2001, which has been joined with
`case IPR2016-00157, Palo Alto Networks Inc., versus Finjan, Inc.
`This is the final hearing. I am Judge Giannetti. On the screen to
`my left is Judge Boucher. Judge Quinn is on the screen to my
`right. They are participating in this hearing remotely. And Judge
`Boucher will be presiding. So Judge Boucher.
`JUDGE BOUCHER: Thank you. Just to clarify the
`record, there are actually two other IPRs involved here as well,
`IPR2015-2001 and 2016-157 have been consolidated with Palo
`Alto Networks as the petitioner, but then there is also 2016-955
`and 956, which have been joined to these proceedings, filed by
`Blue Coat Systems. And those two have also been consolidated.
`Before we begin, I just want to remind the parties that
`the hearing is open to the public and a full transcript of it will
`become part of the record. Also please keep in mind that
`anything that is projected on the screen will not be viewable by
`me or by Judge Quinn. So when you refer to an exhibit on the
`screen, please state the slide, exhibit or page number that you are
`referring to for the record. Also I want to remind the parties that
`demonstrative exhibits are not evidence and have not been filed
`as part of the record. In this case, I haven't seen any objections
`from either side to the demonstrative exhibits.
`
`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 IPR2015-02001
`Patent 8,225,408 B2
`
`
`And I also want to remind the parties that arguments are
`limited to those that are supported by the written briefing. To the
`extent that an argument is made or evidence presented that lacks
`support in the written briefing, we will disregard it when we
`render our final written decision.
`So now if we could get appearances from the parties,
`please, beginning with the petitioner.
`MR. EUTERMOSER: Brian Eutermoser on behalf of
`petitioner, Palo Alto Networks.
`JUDGE BOUCHER: Thank you, Mr. Eutermoser. And
`for the patent owner.
`MR. HANNAH: Good afternoon, Your Honors. James
`Hannah on behalf of Finjan. And with me is Shannon Hedvat.
`JUDGE BOUCHER: And we allotted one hour to each
`side for argument. So if you want to begin, petitioner, when you
`are ready, and let us know how much time you would like to keep
`for rebuttal, and I will keep track of the time here.
`MR. EUTERMOSER: Thank you, Your Honor. I think
`we just need a minute to get the slides set up. But I would like to
`retain 20 minutes for rebuttal.
`JUDGE BOUCHER: So whenever you are ready,
`please go ahead and begin.
`MR. EUTERMOSER: Good afternoon. May it please
`the Board. In these proceedings, petitioner, Palo Alto Networks,
`respectfully requests that the Board find all independent claims
`
`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
`
` 4
`
`
`
`
`
`
`
`Case IPR2015-02001
`Patent 8,225,408 B2
`
`and certain dependent claims of the '408 patent invalid as
`obvious.
`JUDGE GIANNETTI: Mr. Eutermoser, I think you
`have to get closer to the microphone.
`MR. EUTERMOSER: Is that better?
`JUDGE GIANNETTI: That's better.
`MR. EUTERMOSER: Unless the Board directs
`otherwise, I plan to proceed through the topics shown on slide 2
`of petitioner's demonstratives in the order shown here. So I'll
`start with a quick overview of the '408 patent and then talk briefly
`about claim construction, state of the art and knowledge of a
`person of skill in the art, and then deal with the disputed claim
`limitations.
`So this case and the '408 patent is about virus detection
`using parse trees. As we'll see, Finjan didn't invent this field, and
`prior art discloses and teaches all the limitations in the claims.
`This is Figure 2 of the '408 patent shown on slide 4 of our
`demonstratives. And this outlines the three separate stages
`described in the claims of the '408 patent. It takes an incoming
`stream of code. That code is passed through a tokenizer and then
`a parser and then an analyzer, one after the other.
`And using lexical analysis to convert a data stream to a
`stream of tokens, that's what happens in the tokenizer. That
`wasn't new. Nor was using a parser to identify patterns of tokens
`that correspond to potentially malicious code. Nor was using an
`
`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
`
` 5
`
`
`
`
`
`
`
`Case IPR2015-02001
`Patent 8,225,408 B2
`
`analyzer to then detect those patterns that indicate potentially
`malicious code. And as we'll see, using parse trees in connection
`with the parser to store and analyze the tokens and patterns of
`tokens was also well known.
`Now, I just want to touch briefly on the claim
`construction issues or potential issues in this case. Petitioner
`proposed BRIs for four different claim terms, first being parse
`tree. The Board adopted patent owner's proposed construction of
`parse tree. However, the instituted claims are invalid for the
`same reasons under both the Board's and Palo Alto Network's
`constructions, which was confirmed by expert testimony from
`Dr. Rubin. And Finjan does not challenge obviousness based on
`any alleged differences between the constructions adopted by the
`Board and proposed by Palo Alto Networks.
`JUDGE BOUCHER: Are there any claim construction
`disputes at all?
`MR. EUTERMOSER: I don't believe so, in terms of the
`actual construction of the terms. The next term is instantiating a
`scanner for the specific programming language. And just as with
`parse tree, I don't believe that Finjan disputes obviousness based
`on any distinctions between the Board's construction and the
`construction proposed by Palo Alto Networks.
`JUDGE BOUCHER: What about an incoming stream
`of program code, are the parties consistent in the way they
`construe that phrase in the claims?
`
`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
`
` 6
`
`
`
`
`
`
`
`Case IPR2015-02001
`Patent 8,225,408 B2
`
`
`MR. EUTERMOSER: I believe so. I'm not aware of
`any disputes there. In our view, incoming stream of program
`code is exactly what it sounds like, a stream of characters or bits
`or packets that is received over a network by a computer.
`The remaining two terms, "dynamically building" and
`"dynamically detecting," aren't directly relevant to what appear to
`be the key issues in dispute in these proceedings. However, there
`is no dispute as to -- at least for purposes of this proceeding -- as
`to how those terms should be construed. The Board concluded
`consistent with Palo Alto Networks, petitioner's proposed
`construction that "dynamically building" should require that the
`building step in the claims of the '408 patent must overlap with
`the time period for the receiving step. In other words, the parse
`tree must be built, the tree must be parsed during a time period
`that overlaps with the time period during which the incoming
`stream is received.
`And there's a similar requirement, again agreed by the
`parties and adopted by the Board, for "dynamically detecting,"
`which requires that the time period during which malicious code
`is or potential exploits are detected must overlap with that period
`during which the stream is parsed and the parse tree is built.
`Now, so I want to turn next to the state of the art in
`2004 and particularly the knowledge of a person of skill in the art
`in that time period. You may be asking why is that so important
`in this case. It's important here because Finjan has made 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
`
` 7
`
`
`
`
`
`
`
`Case IPR2015-02001
`Patent 8,225,408 B2
`
`dynamically building and detecting limitations the key issues in
`this case despite the fact that there is little, in fact, almost no
`disclosure concerning those terms in the specification of the '408
`patent. It matters because Finjan's main argument is that a person
`of ordinary skill reading Chandnani would assume that each stage
`of Chandnani's detection engine processes an entire data stream
`before passing any portion of that stream on to the next stage in
`Chandnani's detection process or system.
`But this argument by the -- that patent owner makes
`assumes away basic and foundational knowledge of a person
`skilled in the art concerning how data is transmitted and received
`over a network. And as we'll see, admissions from Finjan's own
`experts, patent owner's own experts, demonstrate that this
`argument is inconsistent with how computer networks like the
`Internet actually work and how a person of ordinary skill would
`have interpreted Chandnani's teachings regarding virus detection
`as it applies to an incoming stream of code or a stream of packets
`received over the Internet.
`JUDGE BOUCHER: So as I understand it, the dispute
`between the parties seems to hinge primarily on whether or not
`Chandnani discloses these dynamic limitations that appear in the
`claims. So is it fair to say that you are relying on this knowledge
`of skill in the art to draw inferences from Chandnani as to how
`exactly that works? I mean, to what extent are you relying on 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
`
` 8
`
`
`
`
`
`
`
`Case IPR2015-02001
`Patent 8,225,408 B2
`
`general knowledge of one of skill in the art as opposed to express
`disclosures in Chandnani?
`MR. EUTERMOSER: Well, I don't think we are
`relying exclusively on the knowledge of one of skill in the art and
`I don't think you can separate those two. Under KSR you have to
`interpret the reference's teachings in light of the knowledge of a
`person of ordinary skill in the art. In this case there are express
`disclosures in Chandnani that make clear that the entire detection
`process has to be completed before the incoming data stream has
`been completely received at the computer. So in our view, that
`disclosure alone is enough to teach the dynamic limitations
`because if the detecting and the parsing step have to be completed
`before the code is received, that means all these time periods
`overlap, which is what the Board's constructions require.
`Now, there are other disclosures in Chandnani about
`how these individual functions work and process token by token
`or character by character that confirm that that's how Chandnani
`works and that's how one of skill in the art would have interpreted
`Chandnani. And those disclosures in Chandnani are completely
`consistent with the knowledge of a person of skill in the art
`concerning packetized networks and in general how a streaming
`flow of data is processed by a computer, specifically with respect
`to virus detection.
`So as I just mentioned, to set the table for my state of
`the art discussion here, I would like to point out that most of this
`
`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
`
` 9
`
`
`
`
`
`
`
`Case IPR2015-02001
`Patent 8,225,408 B2
`
`evidence, if not all of it, comes from admissions made by patent
`owner's own experts during depositions in this case. For
`example, Dr. Medvidovic confirmed -- I'm sorry?
`JUDGE QUINN: What slide are you on?
`MR. EUTERMOSER: I'm sorry. I'm on slide 18. I'll
`try to remember to call that out each time.
`And this slide shows that as Dr. Medvidovic admitted,
`modern computer networks both in 2004 and today are
`packet-based meaning that anything transmitted over a network is
`not transmitted as a single block of data. It's first broken into a
`series of data packets.
`Moving on to slide 19, as an example, consider a typical
`web page. The content of that web page gets fragmented into
`components that fit inside individual packets which are then
`transmitted one after the other in serial fashion between the web
`server and a client computer. The testimony here comes from
`another of patent owner's expert's, Dr. Bims.
`It's also clear --
`JUDGE BOUCHER: I know that the patent owner
`didn't object to the demonstratives, but is this testimony that you
`are referring to on these slides some of the testimony that they did
`move to exclude?
`MR. EUTERMOSER: I don't think what we've seen so
`far. I can think of one slide that we will get to that is the subject
`of their motion to exclude, I believe.
`
`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
`
` 10
`
`
`
`
`
`
`
`Case IPR2015-02001
`Patent 8,225,408 B2
`
`
`JUDGE BOUCHER: Okay.
`MR. EUTERMOSER: And actually, thinking more
`about that, I do recall that at least as it relates to Dr. Medvidovic,
`they have moved to exclude a number of portions of his
`testimony concerning data packets and packetized networks in
`general. So there's a decent chance they have moved to exclude
`portions that I'm presenting now.
`JUDGE GIANNETTI: They have moved to exclude
`their own expert's testimony?
`MR. EUTERMOSER: That's correct. And again, in
`our view, these are admissions from a party opponent. And I'm
`happy to address our response to their motion now. I planned to
`do that on rebuttal.
`JUDGE GIANNETTI: Let's hear from them first. I'm
`interested in that.
`MR. EUTERMOSER: It sounds like you are asking to
`leave that aside and deal with it on rebuttal?
`JUDGE GIANNETTI: Yes, the exclusion issue. You
`can continue. I just wanted to clarify that.
`MR. EUTERMOSER: Sure. So as I mentioned, a
`person of skill in the art knew how packetized data networks
`work, knew how a web page gets fragmented into multiple
`packets before being transferred over a network and knew how to
`process those packets as they pass through a network and get
`received by a computer. A fundamental piece of network-based
`
`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
`
` 11
`
`
`
`
`
`
`
`Case IPR2015-02001
`Patent 8,225,408 B2
`
`communications is -- I think marshalling is one of the terms
`Dr. Medvidovic used -- but relates to the process of packing and
`unpacking a larger piece of content in a series of packets.
`Now, in the context of the '408 patent, I'm now looking
`at slide 21, which shows Figure 2 of the '408 patent. What this
`means for purposes of the claimed invention is that each packet in
`an incoming data stream or each byte or character within each
`packet has to pass sequentially through the various stages or
`components that are claimed in the invention. So again, you see
`the byte source here on the left side of Figure 2. That is the
`incoming stream at least for purposes of the independent claims.
`That feeds directly into the tokenizer. The output of the tokenizer
`is fed into the parser. The output of the parser is then fed into the
`analyzer which detects the presence of potential exploits using the
`claim language. Again, the fact that this must happen
`sequentially, basic understanding of a person of skill in the art
`and confirmed by Dr. Medvidovic on cross-examination.
`JUDGE BOUCHER: So there was an argument in the
`briefing, I think, that the patent owner made that at least in some
`instances it might not be possible to do a full analysis on a
`packet-by-packet basis. I think one of the examples they used
`was detecting the language. Do you agree that there are instances
`in which it may not be possible to detect the language without
`receiving the entire stream first?
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`
` 12
`
`
`
`
`
`
`
`Case IPR2015-02001
`Patent 8,225,408 B2
`
`
`MR. EUTERMOSER: I don't. At least not that I'm
`aware of. And that's the piece that I'm about to get into now.
`And this, again, is confirmed by testimony from patent owner's
`own experts. The script languages, which is what both patents
`deal with at least on some level. JavaScript, I believe, is
`disclosed by Chandnani, and the '408 patent deals with script
`language viruses as well -- or excuse me, script language analysis
`or exploits.
`JavaScript in particular is specifically used to embed
`active content in web pages. That's as true today as it was in
`2004. The data packets that make up a web page often contain
`both include HTML code which essentially defines the page as
`well as embedded scripts which might perform some function,
`including JavaScript. And you may have JavaScript, for
`example, embedded with HTML within a packet or a group of
`packets.
`Now, those scripts include tags that identify the script
`language. And those tags are sufficient to identify the script
`language used. So it wouldn't be necessary to receive and process
`an entire data stream. Far from it. A tag that identifies HTML or
`JavaScript, example, would exist within a single packet or a
`group of packets that could be at the beginning of the stream and
`then would tell the computer receiving the remainder of the
`stream what language followed in the subsequent packets.
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`
` 13
`
`
`
`
`
`
`
`Case IPR2015-02001
`Patent 8,225,408 B2
`
`
`I'm looking at slide 24 which sets forth some of the
`testimony of Dr. Medvidovic essentially confirming this fact. It
`was well known in the art that all that was necessary to identify a
`particular programming language was a particular tag, which
`again did not require processing of an entire incoming data
`stream.
`
`JUDGE BOUCHER: Didn't he also testify that you
`wouldn't necessarily have the tags in some of these packets?
`MR. EUTERMOSER: I don't recall that testimony
`specifically. I will agree, I think, if I understand your question,
`that every packet doesn't necessarily contain a tag that indicates
`the language, which again, the tag may specify the language of
`other packets that follow, at least potentially.
`JUDGE BOUCHER: Okay. But there's always going
`to be a tag?
`MR. EUTERMOSER: In every packet?
`JUDGE BOUCHER: No, but at least at the beginning
`of every stream?
`MR. EUTERMOSER: That's my understanding at least
`according to things like HTML and JavaScript and other script
`languages that I'm familiar with. As of 2004, there was going to
`be a tag that identified the language used in the data stream.
`JUDGE BOUCHER: Okay. Thanks.
`MR. EUTERMOSER: I don't want to overstate, but I
`am reasonably certain that's true with respect to web-based
`
`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
`
` 14
`
`
`
`
`
`
`
`Case IPR2015-02001
`Patent 8,225,408 B2
`
`content, for example, the content of a web page that's gets
`communicated as a series of packets.
`And again, you have here on slide 25, Dr. Medvidovic
`confirming that a packet payload might stay HTML 4.0 as the doc
`type, and that would indicated to the receiving computer that the
`incoming code was HTML.
`JUDGE BOUCHER: But none of this is in Chandnani,
`
`right?
`
`MR. EUTERMOSER: Well, Chandnani refers to
`determining the script language used in the incoming data stream.
`It doesn't talk about exactly how or at least it doesn't talk on the
`level of packets about examining a single packet to determine the
`language used in that stream. It takes, I think, a higher level view
`of how to process the incoming code.
`JUDGE BOUCHER: But this argument that tags would
`be sufficient to identify the language, that's all based in the
`knowledge of one of skill in the art. There's no express disclosure
`in Chandnani to support this; is that correct?
`MR. EUTERMOSER: Well, what Chandnani describes
`is identifying a program language by performing lexical analysis
`of the incoming stream or performing an analysis step on the data
`stream. And our argument is that what that would mean to one of
`skill in the art and the only way to perform that analysis is on a
`packet-by-packet basis and looking to the script tag to identify the
`language.
`
`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
`
` 15
`
`
`
`
`
`
`
`Case IPR2015-02001
`Patent 8,225,408 B2
`
`
`Well, let me back up for a second. Our position is that
`Chandnani discloses performing that step, that lexical analysis
`step to identify the language on a stream of data. Patent owner
`argues that what that means, and this is an assumption they make,
`not based on a disclosure in Chandnani, they say that the only
`way you could possibly identify the script language used in an
`incoming data stream is to process an entire stream of data. So
`patent owner is treating the stream as this discrete object that
`must be processed in its entirety. Our point here is that that
`interpretation again is unsupported by anything in Chandnani and
`also inconsistent with the knowledge of one of skill in the art with
`respect to how network-based communications work and how
`script languages are identified in particular.
`Now, one other point that's relevant to the dynamically
`building and detecting limitations, a person of skill in the art in
`2004 recognized that it was always going to be a goal of
`increasing processing speed and reducing delay. That's always
`been true at least as long as there's been network-based
`communications like the Internet. In 2004 that was probably
`more true than it is today because you are going to see more delay
`as data gets processed across a lower bandwidth than we have
`now.
`
`Dr. Medvidovic, I'm looking at slide 28, he just
`confirmed that speed is important and always has been important
`when you are processing an incoming stream and trying to
`
`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 IPR2015-02001
`Patent 8,225,408 B2
`
`identify potential computer viruses, because that activity often
`happens in realtime, meaning as the data is being received and
`while a user is waiting to view a web page, for example.
`Now, Dr. Bims --
`JUDGE BOUCHER: But he also testified, didn't he,
`that increasing speed can come with trade-offs in other
`objectives. I don't know that it's fair to characterize his testimony
`as saying that, I don't know, speed is some kind of overarching
`goal, is it?
`MR. EUTERMOSER: I think it is fair to characterize
`that as his testimony. I agree that he testified that there are other
`goals and there are trade-offs. Specifically I think he referred to
`accuracy as being a goal. And sometimes accuracy can be at
`odds with speed. But it was always a goal to speed up processing
`or reduce delay. Now, how you do that without sacrificing
`accuracy, I think, is a challenge. But in terms of the overall or
`overarching goal of data processing and computer programming
`in general to reduce latency, I don't think there's much dispute as
`to that fact.
`Now, patent owner's other expert, Dr. Bims, he, in fact,
`confirmed that essentially it would have been obvious to process
`an incoming data stream dynamically as that term has been
`construed by the Board. Now, he testified, as you'll see here at
`slide 29, and I'm referring to Exhibit 1065 at page 30, lines 40
`through 15. This is Dr. Bims' testimony on cross-examination.
`
`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
`
` 17
`
`
`
`
`
`
`
`Case IPR2015-02001
`Patent 8,225,408 B2
`
`He testified that a person of ordinary skill would understand that
`techniques such as detecting as well as analyzing the parse tree
`nodes would take place as the program code is being downloaded
`and that potentially exploits can be identified before the entire
`web page content has been streamed to the computer.
`So what he's saying is that a person of ordinary skill
`would recognize based on nothing more than the need to process
`an incoming data stream in real time and the goal of reducing
`user-perceptible delay, that was enough for a person of ordinary
`skill to recognize that you would perform an entire detection
`process, which in the case of Chandnani and the '408 patent
`would include the tokenizing, parsing, detecting or analysis steps.
`You would do all of that before the file had even arrived at the
`computer.
`JUDGE GIANNETTI: Counsel, what is petitioner's
`contention as to who a person of ordinary skill is?
`MR. EUTERMOSER: I'm sorry?
`JUDGE GIANNETTI: Who is a person of ordinary
`skill for this case?
`MR. EUTERMOSER: I think our position is that it's
`someone with at least a bachelor's degree in computer science or
`related field and two or three years experience. I believe both
`sides are close and may disagree as to two to four years of
`experience. But I think both sides would agree it's somebody
`with some experience.
`
`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
`
` 18
`
`
`
`
`
`
`
`Case IPR2015-02001
`Patent 8,225,408 B2
`
`
`JUDGE GIANNETTI: A general background in
`computer science would be enough?
`MR. EUTERMOSER: I've got it here. A person of
`ordinary skill held a bachelor's degree or the equivalent in
`computer science and three to four years of additional experience
`in the field of computer security. So I think it's fair to say that at
`least a degree and some experience in the fields relating to
`computer science. And again, in our view, processing an
`incoming data stream over a network in 2004 was fundamental to
`that field. I don't think there's much dispute as to that fact.
`JUDGE BOUCHER: So why is the speed important
`here? Is the argument that because speed is important, one of
`skill in the art would have known from Chandnani to do it on a
`packet-by-packet basis rather than wait for the entire stream to
`arrive and then do the analysis?
`MR. EUTERMOSER: There's no dispute that you have
`to process the incoming stream on a packet-by-packet basis. The
`question and the dispute between the parties is whether
`Chandnani teaches processing every packet in an incoming data
`stream by one stage of a particular detection method before you
`pass any of those packets on to the next stage. And the point is
`that given the need to speed up processing or reduce delay, there
`is no reason a person of skill in the art would even consider
`waiting for every packet in a data stream to arrive, particularly
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`
` 19
`
`
`
`
`
`
`
`Case IPR2015-02001
`Patent 8,225,408 B2
`
`where the computer wouldn't necessarily know when that stream
`was going to end, before proceeding with the processing.
`JUDGE GIANNETTI: Unless you had to do that in
`order to accomplish the task. Unless you had to -- if parsing
`required that you look at more than one packet, then you would
`have to wait until you had enough packets to make the parsing
`meaningful to performance; is that correct?
`MR. EUTERMOSER: That may be true. But that
`doesn't imply that even if you do need more than one packet to
`perform the parsing step, it doesn't imply that you wait until you
`receive the entire stream. Again, there's nothing in Chandnani
`that teaches that would necessarily be the case. This argument
`comes from patent owner saying that, well, it would be necessary
`to process an entire stream, the entire data stream as a whole to
`determine the language before moving on to the next step.
`Our point is that's just not the case. It wasn't the case in
`2004. And given the motivation of one of skill in the art to speed
`up processing and virus detection specifically so you could
`identify threats before the file is saved at the computer, that there
`is no way -- there is no reason that one of skill in the art would
`have interpreted Chandnani as teaching that you perform that at
`any stage on an entire stream before proceeding to the next stage.
`JUDGE BOUCHER: My impression, and I'm curious
`whether or not you think it's a fair characterization, is that the
`petitioner came up with an argument based on Chandnani and
`
`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
`
` 20
`
`
`
`
`
`
`
`Case IPR2015-02001
`Patent 8,225,408 B2
`
`then Kolawa and Walls, but primarily based on Chandnani. And
`then patent owner came back and pointed out some potential
`ambiguities in the disclosure there. And to try to patch those up
`in reply, the petitioner has now relied on the knowledge of one of
`skill in the art. Is that a fair characterization of what this
`argument is?
`MR. EUTERMOSER: I don't believe so. I think the
`patent owner raised an argument in its patent owner response that,
`again, is inconsistent with what is basic and fundamental about
`how network and stream-based communications work. Again,
`Chandnani discloses expressly that its entire detection method
`should be performed before the data stream is even resident on
`the computer, which teaches the dynamically limitations by itself.
`And I can pull that up now.
`JUDGE BOUCHER: So I guess fundamentally I'm
`wondering why are we talking about this stuff in such detail if we
`are relying on the express disclosure of Chandnani?
`MR. EUTERMOSER: Well, I think Chandnani doesn't
`go into this level of detail because it was assumed by one of skill
`in the art that this is how you would process an incoming data
`stream. Again, this is all rebutting patent owner's argument that
`Chandnani says something it doesn't say. Again, Chandnani just
`says that you perform multiple operations in succession on the
`data stream. Patent owner argues that that means you have you
`have to process the entire data stream before passing the stream
`
`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
`
` 21
`
`
`
`
`
`
`
`Case IPR2015-02001
`Patent 8,225,408 B2
`
`from one stage to the next. And this argument is just to
`demonstrate that that is fundamentally at odds with the
`knowledge of one of skill in the art.
`JUDGE BOUCHER: Okay. So all of this is essentially
`rebuttal. So let's look at what Chandnani actually discloses, then.
`MR. EUTERMOSER: Certainly. So what we have up
`on the screen is from Chandnani, Exhibit 1003. This is on
`page 14 at column 9, lines 12 through 16. And what this is
`saying is that if you look at the highlighted portion, the script
`language virus detection methodologies which are disclosed
`throughout Chandnani may be performed on a file or a data
`stream received by the computer through a network before the
`file is stored, copied, executed, open on the computer. So what
`this is saying is that Chandnani's detection method, which again,
`includes tokenizing, parsing and detecting malicious code, that
`entire process can be performed either on a file, which is one
`embodiment, or on a data stream received over a network. And
`that entire detection process can be completed, entirely performed
`before the file is stored