UNITED STATES PATENT AND TRADEMARK OFFICE
`____________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`____________
`
`APPLE, INC.,
`Petitioner,
`
`v.
`
`ANDREA ELECTRONICS CORP.,
`Patent Owner.
`_____________
`
`Cases IPR2017-00626 and IPR2017-00627
`Patent 6,363,345 B2
`_____________
`
`Record of Oral Hearing
`Held: April 25, 2018
`____________
`
`
`
`Before STEPHEN C. SIU, MICHAEL R. ZECHER, and JEREMY M.
`PLENZLER, Administrative Patent Judges.
`
`
`
`
`
`
`
`
`
`
`
`
`____________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`____________
`
`APPLE, INC.,
`Petitioner,
`
`v.
`
`ANDREA ELECTRONICS CORP.,
`Patent Owner.
`_____________
`
`Cases IPR2017-00626 and IPR2017-00627
`Patent 6,363,345 B2
`_____________
`
`Record of Oral Hearing
`Held: April 25, 2018
`____________
`
`
`
`Before STEPHEN C. SIU, MICHAEL R. ZECHER, and JEREMY M.
`PLENZLER, Administrative Patent Judges.
`
`
`
`
`
`
`
`
`
`
`
`
`
`Cases IPR2017-00626 and IPR2017-00627
`Patent 6,363,345 B2
`
`APPEARANCES:
`
`ON BEHALF OF THE PETITIONER, APPLE, INC.:
`
` JEFFREY P. KUSHAN, ESQUIRE
` THOMAS A. BROUGHAN, III, ESQUIRE
` SIDLEY AUSTIN, LLP
` 1501 K Street, N.W.
` Washington, D.C. 20005
` (202) 736-8914
`
`
`ON BEHALF OF THE PATENT OWNER, ANDREA ELECTRONICS
`CORP.:
`
` WILLIAM D. BELANGER, ESQUIRE
` BRADLEY T. LENNIE, ESQUIRE
` PEPPER HAMILTON, LLP
` 125 High Street
` 19th Floor, High Street Tower
` Boston, MA 02110
` (617) 204-5100
`
`
`The above-entitled matter came on for hearing on Wednesday,
`April 25, 2018, commencing at 1:00 p.m., at the U.S. Patent and
`Trademark Office, 1961 Stout Street, Denver, Colorado.
`
`
`
`
`
`
`
`
`
`
`
`
`2
`
`
`
`Cases IPR2017-00626 and IPR2017-00627
`Patent 6,363,345 B2
`
`
`P R O C E E D I N G S
` JUDGE ZECHER: Please be seated. All right. Good
`afternoon. I'm Judge Zecher. I'm here with two of my
`colleagues that are participating remotely as you can tell
`(inaudible). We have Judge Siu who is on my left and Judge
`Plenzler who is on my right.
` This is an oral hearing for two proceedings. Cases
`IPR2017-00626 and IPR2017-00627, both of which address U.S.
`Patent No. 6,363,345. In each case we instituted an IPR as
`to claims, 1 through 25 and 38 through 47 based on various
`grounds.
` As we outlined in the trial hearing order, we gave
`each party a total of 45 minutes of arguments. We're going
`to have Petitioner present first given that that they carry
`the burden of persuasion here. They can reserve a certain
`amount of rebuttal time at which point Patent Owner will
`present their case and then Petitioner will use the remainder
`of their rebuttal time.
` Just to begin the proceeding and so the record is
`clear, I'd like each person to step up -- each party starting
`with Petitioner to step up to the microphone, state your name
`and then followed by Patent Owner. And then just as a reminder,
`because we have two judges participating remotely,
`any reference to slides need to be clear and explicit and
`obviously when talking please be at the microphone.
` Petitioner?
`
`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
`
`
`
`3
`
`
`
`Cases IPR2017-00626 and IPR2017-00627
`Patent 6,363,345 B2
`
` MR. KUSHAN: Good afternoon. My name is Jeff
`Kushan. I'm with Sidley Austin (inaudible).
` JUDGE ZECHER: Thank you.
` MR. BELANGER: Good afternoon. William Belanger
`with Pepper Hamilton on behalf of Andrea. With me is Brad
`Lennie, also with Pepper Hamilton.
` JUDGE ZECHER: Thank you very much. All right.
`I'll turn the floor over to Mr. Kushan and how many -- how
`much time would you like to reserve for rebuttal?
` MR. KUSHAN: We would like to reserve approximately
`20 minutes for rebuttal.
` JUDGE ZECHER: Okay. You may begin.
` MR. KUSHAN: Thank you, Your Honors. Today we are
`going to focus, as you've asked us to, on the issue
`(inaudible) proceedings. I'm going to be covering the 626
`proceedings based on Hirsch and my colleague, Mr. Broughan,
`will be covering the proceeding based on Helf which is the
`627 proceeding.
` These proceedings -- oh, I'm sorry. Would you like
`(inaudible)?
` JUDGE ZECHER: Oh, yes. Certainly approach. Thank
`you.
` MR. KUSHAN: Thank you. So at a high level there
`are essentially three fundamental problems we see in
`responses (inaudible) patent owner.
` First, they've identified a number of supposed
`
`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
`
`
`
`Cases IPR2017-00626 and IPR2017-00627
`Patent 6,363,345 B2
`
`distinctions relative to the claims but the claims don't have
`language carrying those distinctions into relevance in the
`case.
` The second issue is that they've been using a
`(inaudible) KSR.
` And finally, we think there are a number of issues
`in which they have mischaracterized what the references are
`teaching and what actually the testimony has been.
` If you could go to Slide 3. Now importantly, the
`independent claims that are being contested have not been
`disputed as being anticipated by Hirsch or by Helf. In this
`case Hirsch is the one I'm going to focus on. There's really
`three disputes in the 626 proceeding. The first one I'm
`going to address is whether a skilled person would have
`combined Hirsch with Martin.
` The second set of issues relate to whether Martin
`teaches the techniques that are reflected in some of the
`Dependent Claims 4, 6 and 10.
` And the third disputed issue is whether a skilled
`person would have looked to a variety of publications
`describing conventional techniques of signal processing to
`find the (inaudible).
` If you could go to Slide 4. And the first issue is
`just would a person skilled in the art have considered Martin
`along with Hirsch and we think the evidence is pretty clear
`the answer is yes.
`
`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
`
`
`
`Cases IPR2017-00626 and IPR2017-00627
`Patent 6,363,345 B2
`
` First, Hirsch actually points to Martin. It cites
`to Martin to describe techniques that were known to the art
`that are relevant to a person describing the scheme. That's
`the citation I give you on Slide 4.
` And then if you go to Slide 5, we also provided
`testimony from our expert explaining why beyond the literal
`citation to Martin a skilled person would have consulted --
`I'm sorry -- a literal citation to Hirsch why a skilled
`person would have considered Martin and that's because Martin
`describes techniques that were known in the art and used
`previously in the signal processing, particularly (inaudible)
`subtraction.
` They also actually pointed to a number of attributes
`of the Martin scheme that were not addressed or tested in the
`first scheme such as stationary versus non-stationary noise
`settings. Hirsch only tested it in a stationary noise
`setting. And Martin is also identifying attributes that
`would make it simpler to do things like (inaudible) speech
`detection within the same process of the scheme.
` So what we see from the basic question is whether a
`person skilled would have consulted Martin and considered
`whether it had relevance and could be integrated with Hirsch,
`I think there's ample evidence to suggest a skilled person
`would have done that.
` JUDGE ZECHER: Can you go back to Slide 4 for me
`real quick?
`
`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
`
`
`
`Cases IPR2017-00626 and IPR2017-00627
`Patent 6,363,345 B2
`
` MR. KUSHAN: Sure.
` JUDGE ZECHER: I know Patent Owner makes an argument
`that at least with respect to that last sentence there in
`your blurb the disadvantages (inaudible) bits of noisy speech
`that that disparages Martin in some shape or form. How do
`you respond to that?
` MR. KUSHAN: Sure. Well, the response is fairly
`simple. Martin is illustrating his technique with one
`duration of single processing. He used, I think, a .625
`second example. What Martin is doing -- what Hirsch is doing
`by pointing to Martin is identifying that as a known
`technique and what we have shown with our expert testimony in
`particular is that a skilled person would look at what Martin
`has in it and see if it can be adapted and used in the Hirsch
`setting.
` This is not a teaching (inaudible). This is an
`observation about an older scheme. Obviously the Hirsch
`description or Hirsch technique being described in Hirsch is
`an advance, as there have been many, many advances to this
`scheme, and so it's a natural thing to point to some of these
`common techniques and see how they might be adapted or be
`relevant to the Hirsch scheme.
` So we don't think this rises to the level of
`teaching away where disparagement might be relevant here.
`This is a (inaudible) observation about an older scheme which
`has an element that we've seen is relevant and useful in 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
`
`
`
`7
`
`
`
`Cases IPR2017-00626 and IPR2017-00627
`Patent 6,363,345 B2
`
`Hirsch scheme.
` If you could go to Slide 4. I'm sorry. Go to Slide
`6. So the next dispute I want to address focuses on whether
`the Martin technique of using a noise estimate (inaudible) is
`something that would meet the requirements of Claim 4 and I
`think the most important thing to take away from all of the
`discussion around this is the claim language in Claim 4.
` Claim 4 is modifying the threshold detector element
`on one. That's the threshold for noise. (Inaudible)
`using that technique as set out in Claim 4. But what's
`notable about Claim 4, it doesn't set restrictions on the
`duration or the sample you have to use. It doesn't set any restrictions
`on how you derive the current minimum value from the
`future minimum value. It has no boundaries really in terms
`of the parameters that might relate to the (inaudible) arguments).
` I'm sure you've seen happen under (inaudible).
` So there's no minimum or maximum sample size,
`duration of the signal. There's no restrictions on how you
`actually derive the current minimum from the future minimum.
`And when you look at what Martin describes, that falls
`cleanly within the scope of Claim 4.
` If you go to Slide 7. These are the passages we've
`addressed in our petition and in our reply that are in Martin
`and what you see Martin explaining is that its technique is
`using a mechanism or a technique to estimate a noise or that
`is going to derive that noise far from the minimum value in a
`
`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
`
`
`
`Cases IPR2017-00626 and IPR2017-00627
`Patent 6,363,345 B2
`
`portion of the signal. And there are a couple of variables
`reflected in the bottom (inaudible) on Slide 7 where its
`explaining that you start with a window of the sample and
`then in normal operation of it it uses sub windows to then
`parse portions of that window.
` But the more important message being sent by these
`two observations in Martin is that setting a noise floor on a
`minimum value to set the current -- (inaudible) current
`minimum and that's an inescapable conclusion about what
`Martin is doing.
` If you could go to Slide 8, please. Both experts
`agree that the techniques shown in Martin are using a minimum
`value or a portion of the signal to set the current minimum
`and what I --
` JUDGE PLENZLER: Let me stop you real quick just --
` MR. KUSHAN: Sure.
` JUDGE PLENZLER: I don't think there's any real
`disagreement, correct me I'm wrong, between yourself and
`Patent Owner that Martin uses a minimum in some way. I think
`the real dispute is about the sub windows, correct?
` MR. KUSHAN: Well --
` JUDGE PLENZLER: So maybe if you could focus on that
`a little bit and tell me where, if anywhere, does Martin
`contemplate not using sub windows? Because effectively
`you're saying W equals sub windows, right, W equals 1, that's
`effectively no sub window, wouldn't you agree?
`
`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
`
`
`
`Cases IPR2017-00626 and IPR2017-00627
`Patent 6,363,345 B2
`
` MR. KUSHAN: Yeah, that would just be one window
`corresponding with one sub window.
` JUDGE PLENZLER: Right. So does Martin ever talk
`about not using sub windows? It seems like the whole point is talking
`about (inaudible) slow varying or rapid varying noise power and it's
`looking for this monotonic increase, whether or not that's occurring.
` MR. KUSHAN: Well, I think I would take issue with
`one way you portrayed that. I think the main point of Martin
`is using a noise floor to estimate a noise and then using that in a spectral
`strategy technique. The issue relating to whether you're detecting
`monotonically increasing or not increasing signals is really not the heart
`of Martin, it's an element of the Martin scheme.
` And what I think is helpful -- and I'll focus on why
`our expert focused on the one sub window scenario. First,
`obviously Martin has a formula that has that as a
`possibility, but the main point of doing that was to
`illustrate that the operation in either of the two paths,
`whether you're monotonically increasing or not, is to use the
`PN value which is the future minimum that has been derived
`from the set of samples to set the noise floor which is PN.
` So whether you're taking a no or yes branch on the
`algorithm looking at the monotonically increasing variable
`here, you're going to have a minimum value which is a future
`minimum and you're going to derive the noise floor PN which
`is a current minimum from that determination and it's --
` JUDGE PLENZLER: Quick question.
`
`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
`
`
`
`Cases IPR2017-00626 and IPR2017-00627
`Patent 6,363,345 B2
`
` MR. KUSHAN: Go ahead.
` JUDGE PLENZLER: The minimum value being depicted
`depends on whether or not you have a monotonically increasing
`sound or not, right?
` MR. KUSHAN: Correct. Correct.
` JUDGE PLENZLER: I mean, if it's monotonically
`increasing the minimum is going to be one thing, right, for
`the future minimum, all data recurring, and if it's not
`monotonically increasing then something would be different.
` MR. KUSHAN: Well, let me make it somewhat -- I
`think a clearer distinction. In the scenario that was
`exemplified in Martin you had four sub windows and in the
`setting where the signal is not monotonically increasing it
`performs an operation to find four PN values which would be
`the future minimum. So you have four options.
` It then picks the lowest value of those four and
`sets the noise floor, the PN, to that lowest value. And
`Mathematically it's indistinguishable from not doing anything
`with the window and the sub window because if you ran through
`the window (inaudible) from one to 5,000 samples you would
`end up with the same number, right. That's the point of the
`non-monotonically increasing arm of this algorithm.
` When you take the arm that goes off and it is
`monotonically increasing it has a different outcome. One of
`those four sub windows is going to pick the noise floor and
`set it to the highest value which would be the last sub
`
`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
`
`
`
`Cases IPR2017-00626 and IPR2017-00627
`Patent 6,363,345 B2
`
`window. But if you put up the claim language again we think
`that is of no consequence because we looked at the claim
`language and it captures that scenario as well.
` So there's no requirement in Claim 4 and you pick
`the lowest value of an arbitrarily defined portion of the
`signal. It says literally the current minimum being derived
`in accordance with a future minimum value. So that captures
`even the monotonically increasing scenario in Martin where
`you pick the highest of four values out of your PMINs.
` And I think that's the best way to think about this
`because the Martin scheme, the central focus of the Martin
`scheme is trying to figure out how to set a noise floor or
`doing the estimation of noise in the signal and it's using a
`lowest value of a sample, whether it's the window or a sub
`window, to set that noise floor.
` JUDGE PLENZLER: So for, like you just explained,
`the claim may be brought in a way that is (inaudible) current
`minimum with future minimum and how the future minimum is
`used. But, and I think you kind of touched on a little bit
`that Martin, even if you don't have -- and I think you talk
`about it in #5, just one sub window or no sub windows, rather
`if you have something more still -- I think you argued that
`it still meets the broad language of the claim, right?
` MR. KUSHAN: We do.
` JUDGE PLENZLER: What I'm wondering though and I
`tried to look through and maybe you can help find it or
`
`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
`
`
`
`Cases IPR2017-00626 and IPR2017-00627
`Patent 6,363,345 B2
`
`something, but it potentially looked like you were just
`looking at the scenario where W equals 1, right. So
`effectively there's no sub windows.
` MR. KUSHAN: Well --
` JUDGE PLENZLER: Do you think that you can go in
`your -- if you don't have it in your petition, why can you go
`in your reply now and change your course?
` MR. KUSHAN: Well, I think we need to take -- we're
`not agreeing with the suggestion of a hidden address. The
`central point of Martin in our petition, if you look at
`petition pages 38 to 40, we discuss the general approach of
`of Martin before we go into the example where we just
`(inaudible). And remember, our expert was using the one sub
`window scenario to illustrate kind of the overall operation of
`Martin which is find the minimum value and find a set of ---
`- a portion of the signal and then use that minimum value to
`set the noise floor which would be the current minimum. And
`by (inaudible) --
` JUDGE PLENZLER: Can you point me to some exact
`language that you have in those pages 38 to 40 of the
`petition where you're talking about how the scenarios where
`you have more than sub window, where that would be the claim?
`Because I see -- I know you have a general discussion of
`Martin, right, but as far as really getting into actually
`discussing the limitations of Claim 4 (inaudible), on page 39, all I'm
`really seeing is this specific discussion for the example
`
`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
`
`
`
`Cases IPR2017-00626 and IPR2017-00627
`Patent 6,363,345 B2
`
`like you said, W equals 1. So, I mean, now would be the
`time, if you can, to kind of point me to what I'm missing.
` MS. DUNN: So if you look at the bottom of page 38
`we've explained that the noise floor estimation process uses
`a current minimum and a future minimum and we identified the
`current minimum as PNI and then the future minimum PM-MIN
`which is then compared using an iterative process to the PXI
`value. That's the magnitude you're assessing each sample.
`And then we go literally into the algorithm where we've laid
`out those parameters and highlighted the portions of the
`algorithm where those parameters wouldn't apply. And then we
`explain at the bottom of 39 during each period (inaudible)
`of samples the Martin algorithm tracks the (inaudible) power
`level of the signal as far as the minimum value as PMIN, the
`red box, and then moves on to do the assessment of whether
`(inaudible) PN value.
` So we did it at a high level because we think that's
`what the Martin reference is describing. It's describing the
`technique at the heart of it which is using the minimum value
`of the defined portion of the signal to set the minimum
`value, the current minimum from that future minimum.
` And as I go through that example we then take on the
`question of how do we simplify this complicated algorithm
`down to illustrate the central point of it which is setting a
`floor using a minimum value to define the portion of the
`sample. And we used the illustration of the single sub
`
`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
`
`
`
`Cases IPR2017-00626 and IPR2017-00627
`Patent 6,363,345 B2
`
`window because that is permissible under the Martin scheme
`where you have not divided the window into sub windows, you
`just have one sub window and, by the way, there would be
`never any zero sub windows because you would have at least
`one sub window that's (inaudible) window.
` And then we also, I think it's fair to say -- when
`you look at what we illustrated in the algorithm, we pointed
`to the two options in the blue box that's on the bottom right
`corner of the algorithm. And again, when you look at it from
`the non-monotonically increasing scenario, whether you
`have one sub window or four sub windows, you get the same
`number. You're going to pick the lowest value within the
`window. That's what the operation is describing right there.
` So I think we've adequately identified the basis for
`our position that the Martin scheme is meeting the
`requirements of Claim 4 especially given how broad and unconditioned
`Claim
`4 is.
` JUDGE PLENZLER: I guess if we just -- our next
`question here is focusing on the (inaudible). Can you ever
`determine with one sub window whether or not something is
`monotonically increasing or not?
` MR. KUSHAN: With one sub window I think you would -
`- you could do that in the first instance when you go from --
`like starting out the algorithm where you obviously have to
`start with nothing. But again, we were not trying to bring
`
`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
`
`
`
`Cases IPR2017-00626 and IPR2017-00627
`Patent 6,363,345 B2
`
`out the question of whether monotonically increasing signals
`or not are implicated because that's not really relevant to
`Claim 4. Claim 4 doesn't have any requirement for one kind
`of signal or the other.
` It's also important to recognize that the -- whether
`a set of data is exhibiting a monotonically increasing signal
`or not is -- that's the value of the data. It's not the
`algorithm. The algorithm is taking the data and performing
`an analysis to see what the lowest value and portion of that
`signal is and then doing exactly as Claim 4 requires which is
`setting the future or the current minimum based on that
`future minimum was determined from that set of data points.
` If I can I'm just going to touch on two other topics. I mean, if you
`have any further questions I can try to address them (inaudible).
` JUDGE ZECHER: Can I kind of direct you to a topic
`that I'm interested in?
` MR. KUSHAN: Sure.
` JUDGE ZECHER: Particular language "periodically".
` MR. KUSHAN: Sure.
` JUDGE ZECHER: It seems in one petition you kind of
`take a position that it doesn't matter what that construction
`is based on the two dictionary definitions, but in the second
`petition I believe you kind of -- it's a two-fold argument
`that it would need it -- I think it's based on -- was it Helf
`in the second petition.
` MR. KUSHAN: Yes.
`
`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
`
`
`
`16
`
`
`
`Cases IPR2017-00626 and IPR2017-00627
`Patent 6,363,345 B2
`
` JUDGE ZECHER: If it was from time to time but not
`if it was a predetermined period of time when you would then
`need Martin. I mean, obviously there's ground to get there,
`but I'm more concerned with what is the correct construction
`of that term in the context of this patent?
` MR. KUSHAN: So I think there's two things I want to
`address about your question. First one, what's the correct
`construction? I think the correct construction is the
`dictionary definition (inaudible) which has (inaudible) two
`concepts (inaudible). And we have the definition on Slide 15
`and, you know, they've accepted that definition, that
`dictionary definition as a perfect starting point. They just
`want to use one half of it which I think isn't a fair way to
`go about this from a broad sort of interpretation
`respectfully.
` The reason it's not implicated in the Hirsch
`(inaudible) Martin shows the regular interval of applying
`(inaudible) noise floor. In the algorithm we mentioned a
`minute ago you have a fixed set of -- or constant set of --
`constant size set of data values and at the end of each
`iteration which is going to be the same period of time you're
`going to apply the noise floor operation to it. And what
`that does, it means it's going to be a regular application in
`the setting of the noise floor. So it's not an issue that's
`implicated in the Hirsch (inaudible) proceeding in
`combination with Martin because Martin shows that technique.
`
`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
`
`
`
`Cases IPR2017-00626 and IPR2017-00627
`Patent 6,363,345 B2
`
` In the Helf scheme it has some implication because
`Helf does have one scenario where it's not a regular setting
`interval or the noise floor is more dependent on some of the
`data flow.
` I think the one thing I would like to bring out, if
`you could look at Slide 12, to support their version of the
`definition you have to believe that "periodically" as used
`would be limited to periodic kind of equal interval settings
`of the noise floor. And you can see just from their expert's
`analysis an example in the 345 Patent that you don't have an
`identical window of time during which you're setting the
`noise floor.
` You can see, for example, the five to ten window has
`a much larger period of time where is set the floor and
`hasn't changed it and if you look out to a point of 18 and
`forward you see it being updated much more frequently and
`that's not the same interval of time.
` So I think the more natural reading is the
`definition (inaudible) by either a regular interval or from
`time to time because as Dr. Douglas' explanation of the 345
`Patent shows, that is a time to time adjustment of the noise
`floor.
` JUDGE ZECHER: What about Patent Owner's direction
`to the spec where it says every five seconds? That seems to
`be a regular interval.
` MR. KUSHAN: Well, that's an example. I think if
`
`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
`
`
`
`Cases IPR2017-00626 and IPR2017-00627
`Patent 6,363,345 B2
`
`you look at -- and this is why we wanted to show you Dr.
`Douglas' analysis. He's seeing a pattern of adjusting the
`noise floor which isn't in a fixed interval. And even if it
`is a fixed interval we think that doesn't make it patentable
`because that's exactly the technique shown in Martin, sort of
`a regular interval at which it will adjust the noise floor.
` So I think at the end of the day this is essentially
`an inconsequential dispute because (inaudible) is that it is
`an obvious variation from a time to time application.
` JUDGE ZECHER: Okay. But you do admit that if it is
`at a regular interval Helf would account for that (inaudible).
` MR. KUSHAN: Helf would not anticipate that.
` JUDGE ZECHER: Yeah.
` MR. KUSHAN: Helf would I think, as we pointed out
`in our petition, Helf would probably make it obvious because
`that would be the other way of setting the noise floor on a
`regular (inaudible) periodically time to time frequency.
` In the interest of time I would like to just hit briefly Slide 26. They
`dispute a number of the (inaudible)-- obviously
`dependent limitations and what's notable is that they're not
`disputing that any of the references we cited are known as
`conventional signal processing techniques or that they've
`been combined in various ways with spectral subtracting
`(inaudible).
` What they're disputing is really literally whether a
`person of skill would have looked to and incorporated those
`
`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
`
`
`
`19
`
`
`
`Cases IPR2017-00626 and IPR2017-00627
`Patent 6,363,345 B2
`
`references in (inaudible) combination. And we think at the
`root of it -- if you could go to Slide 27, please. This is
`just our summary of the evidence we've laid out for each of
`the sets of dependent claims on the basis of the issue and
`the records we relied on for that conditional technique.
` At the root of it we see two reasons why you
`shouldn't accept the argument (inaudible). The first one is
`that Hirsch itself invites the skilled person to go to these
`other conventional techniques and references that teach them
`which is 518 versus describing a technique which fully
`envisions that it will be combined with other techniques and
`invites the skilled person to look out beyond these four
`corners to find conventional signal processing techniques.
` The second issue is that under the law -- so if you
`go to Slide 28, (inaudible) KSR is clearly met by this level
`of direction and guidance and these are all, as I said, it's
`not disputed that these are conventional known techniques
`that were previously used for the same purpose as is being
`proposed in the dependent claims and therefore without
`something special, for lack of a better word, those elements
`which are conventional techniques used in signal processing
`would be incorporated as Hirsch proposed and we think that's
`adequate to establish a motivation or a reason to consult
`those additional references and the consequence of doing that
`is obvious (inaudible).
` We have about a minute-and-a-half left and I would
`
`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
`
`
`
`20
`
`
`
`Cases IPR2017-00626 and IPR2017-00627
`Patent 6,363,345 B2
`
`invite Mr. Broughan up to touch on any questions you have
`about the Helf grounds unless you have any further questions
`on the Hirsch grounds. (Inaudible).
` MR. BROUGHAN: Good afternoon, Your Honors. I'm Tom
`Broughan. So I'd like to start off by just focusing in on
`what I think is the heart of the dispute as raised by Andrea
`in its briefs and here we're looking at Slide 39 which shows
`how Helf's running minimum estimator works and Helf's running
`minimum estimator works by (inaudible). Helf's running
`minimum estimator will calculate the minimum value of the
`signal throughout that whole interval. And what it does is
`that it calculates -- for each frequency band it calculates
`the magnitude of that band and then it averages that
`magnitude over the preceding eight values that were in that
`interval and over the ten second period Helf will store the
`smallest value, smallest average value that it observes as
`the minimum estimate.
` Go to Slide 40, please. Helf then will make a
`determination as to whether it should update its background
`noise estimate with that minimum estimate. And if two
`conditions are satisfied it will update that estimate.
` Now, if we go to Slide 41, Andrea's argument seems
`to be that Helf can't meet Claim 4 because the average value
`is not a minimum value. But if you look at the language of
`Claim 4 and if you look at Figure 7 of the 345 Patent nothing
`precludes an average value from being a future minimum.
`
`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
`
`
`
`21
`
`
`
`Cases IPR2017-00626 and IPR2017-00627
`Patent 6,363,345 B2
`
`Figure --(Inaudible).
` JUDGE PLENZLER: These are a bunch of averages,
`right? I mean there are a lot of data points. These are a
`lot of averages of (inaudible) values, right, and Helf is
`taking the minimum of those average values, correct?
`
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
