PLANET DEPOS
`» Anywhere
`
`Transcript of MICHAEL S. TRIANTAFYLLOU, Sc.D,
`VOLUME 2
`
`Date: May 23, 2015
`
`Case: PETROLEUM GEO-SERVICES INC., ET AL v. WESTERNGECO LLC
`
`Planet Depos
`Phone: 888-433-3767
`
`Fax: 888-503-3767
`
`Email: transcripts@p|anetdepos.com
`Internet: www.p|anetdepos.com
`
`Worldwide Court Reporting | Interpretation | Trial Services
`
`-_I;_ _ 2___|; _
`
`__|
`
`PGS Exhibit 1092, Pg. 1
`PGS v. WestemGeco (IPR2014-00688)
`
`PUBLIC REDACTED
`
`

`
`DEPOSITION OF MICHAEL S. TRIANTAFYLLOU, Sc.D, VOLUME 2
`CONDUCTED ON SATURDAY, MAY 23, 2015
`
`423
`
`425
`
`1 (Pages 423 to 426)
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
` UNITED STATES PATENT AND TRADEMARK OFFICE
`
` BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`---------------------------------x
`
`PETROLEUM GEO-SERVICES INC. : Cases
`
`and ION GEOPHYSICAL CORPORATION : IPR2014-00687
`
`AND ION INTERNATIONAL S.A.R.L., : (U.S. Patent No. 7,162,967)
`
` Petitioners,: IPR2014-00688
`
` v. : (U.S. Patent No. 7,080,607)
`
`9 WESTERNGECO, LLC, : IPR2014-00689
`
` Patent Owner.: (U.S. Patent No. 7,293,520)
`
`---------------------------------x
`
` Volume 2
`
` Deposition of MICHAEL S. TRIANTAFYLLOU, Sc.D
`
` Alexandria, Virginia
`
` Saturday, May 23, 2015
`
` 8:31 a.m.
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
` A P P E A R A N C E S
`
` ON BEHALF OF PETITIONER:
`
` DAVID I. BERL, ESQUIRE
`
` THOMAS S. FLETCHER, ESQUIRE
`
` WILLIAMS & CONNOLLY LLP
`
` 725 Twelfth Street, N.W.
`
` Washington, D.C. 20005
`
` (202) 434-5000
`
` ON BEHALF OF THE PATENT OWNER:
`
` MICHAEL L. KIKLIS, ESQUIRE
`
` CHRISTOPHER RICCIUTI, ESQUIRE
`
` OBLON, SPIVAK, McCLELLAND, MAIER &
`
` NEUSTADT, LLP
`
` 1940 Duke Street
`
` Sixth Floor
`
` Alexandria, Virginia 22314
`
` (710) 413-3000
`
` A P P E A R A N C E S C O N T I N U E D
`
` ON BEHALF OF THE PATENT OWNER:
`
`426
`
` RYAN KANE, ESQUIRE
`
` KIRKLAND & ELLIS LLP
`
` 601 Lexington Avenue
`
` New York, New York 10022
`
` (212) 446-4800
`
` ALSO PRESENT:
`
` KEVIN M. HART, Petroleum Geo-Services, Inc.
`
`20
`
`21
`
`22
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8 9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`Job No.: 83210
`
`Pages: 423 - 664
`
`Reported by: Leslie A. Todd
`
` Deposition of MICHAEL S. TRIANTAFYLLOU, Sc.D, held
`
`at the offices of:
`
`424
`
` OBLON, SPIVAK, McCLELLAND, MAIER &
`
` NEUSTADT, LLP
`
` 1940 Duke Street
`
` Sixth Floor
`
` Alexandria, Virginia 22314
`
` (710) 413-3000
`
` Pursuant to Notice, before Leslie Anne Todd,
`
`Court Reporter and Notary Public in and for the
`
`Commonwealth of Virginia, who officiated in
`
`administering the oath to the witness.
`
`20
`
`21
`
`22
`
`1
`
`2
`
`3 4 5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`PLANET DEPOS
`888.433.3767 | WWW.PLANETDEPOS.COM
`
`PGS Exhibit 1092, pg. 2
`PGS v. WesternGeco (IPR2014-00688)
`
`PUBLIC REDACTED
`
`

`
`DEPOSITION OF MICHAEL S. TRIANTAFYLLOU, Sc.D, VOLUME 2
`CONDUCTED ON SATURDAY, MAY 23, 2015
`
`427
`
`429
`
`2 (Pages 427 to 430)
`
` C O N T E N T S
`
`EXAMINATION OF MICHAEL S. TRIANTAFYLLOU, Sc.D PAGE
`
` By Mr. Berl 428
`
` E X H I B I T S
`
` (Attached to transcript)
`
`DEPOSITION EXHIBIT PAGE
`
`Exhibit 1084 Sketch drawn by the witness 451
`
`Exhibit 1085 Manual of Offshore Surveying for
`
` Geoscientists and Engineers 457
`
`Exhibit 1086 Drawing 500
`
`Exhibit 1087 Article from E&P, March 2011 516
`
`Exhibit 1088 Article "Cable Positioning with
`
` IRMA" 621
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`that right?
`
` A B and C use a model to predict.
`
` Q A behavior predictive model?
`
` A Yes. To be successful.
`
` Q And A does not.
`
` A A, may not. So it can be ad hoc.
`
` Q A, you said is a noise filter?
`
` A Yes.
`
` Q And by filtering noise, the Kalman filter
`
`provides an estimate of the variable in question, for
`
`example, location?
`
` A It provides an estimate. It can be
`
`location or it can be something, whatever.
`
` Q Whatever the variable is, it provides an
`
`estimate of the actual location.
`
` A The estimate of -- an estimate.
`
` Q The estimate of the actual location if
`
`the Kalman filter is working on locations.
`
` A Yes. In the generalized sense of
`
`estimate.
`
` Q Now, how does one term whether the Kalman
`
`filter is being used for A, B or C?
`
`1
`
`2
`
`3
`
`4 5 6 7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`1
`
` P R O C E E D I N G S
`
`428
`
`430
`
` ---------------------
`
`1
`
` A The major distinguishing feature is the
`
`2 model that is used for the Kalman filter. So in a
`
` MICHAEL S. TRIANTAFYLLOU, Sc.D,
`
`3 model-based prediction, you are using a model which
`
` having been previously duly sworn, was
`
` examined and testified as follows:
`
` FURTHER EXAMINATION BY COUNSEL FOR PETITIONER
`
`BY MR. BERL:
`
` Q Good morning, Doctor.
`
` A Good morning.
`
` Q We were discussing yesterday your
`
`declaration, paragraph 137, the uses of Kalman
`
`filters. If you could turn back to that.
`
` A You are talking about my declaration?
`
` Q Yes. Paragraph 137. It should be on the
`
`top.
`
` A Oh, yes.
`
` Q Paragraph 137. Do you recall yesterday
`
`at the end of the day we discussed A, B and C in
`
`paragraph 137 where you say: "Kalman filters can be
`
`used either A, B or C"?
`
` A Yes.
`
` Q And C and B use behavior prediction; is
`
`4
`
`5
`
`6
`
`7
`
`8
`
`has been based on some physical laws, whether simple
`
`or complicated. That's how you derive the Kalman
`
`filter structure.
`
` In the case of the filter, it -- it still
`
`can be a model-based, just to clean the noise. Or it
`
`9 may be something that you concoct just to remove the
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`noise.
`
` Q So the model in B and C is taking account
`
`in this context of SPD locations for physical forces
`
`that act upon the SPDs.
`
` A It can be a variety of things. So, it
`
`depends on the sophistication of the user.
`
` Q But in order for it to be a model that is
`
`based on physical laws, it would account for physical
`
`forces on the SPD locations.
`
` A It doesn't necessarily have to be forces.
`
`Because, for example, there can be implicit
`
`assumptions like the force is constant. Okay. So it
`
`can have a much more broader interpretation what the
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`PLANET DEPOS
`888.433.3767 | WWW.PLANETDEPOS.COM
`
`PGS Exhibit 1092, pg. 3
`PGS v. WesternGeco (IPR2014-00688)
`
`PUBLIC REDACTED
`
`

`
`DEPOSITION OF MICHAEL S. TRIANTAFYLLOU, Sc.D, VOLUME 2
`CONDUCTED ON SATURDAY, MAY 23, 2015
`
`431
`
`433
`
`3 (Pages 431 to 434)
`
`1 model is. Okay.
`
`2
`
`3
`
`4
`
` Q And you're now distinguishing a model
`
`from a behavior predictive model.
`
` A No. It always -- always models will be
`
`5 models; they will be approximations.
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
` Q What is the difference between a model
`
`and a behavior predictive model?
`
` A A behavior-based model -- let's leave
`
`outside the word "prediction" which can be the
`
`subsequent step. But a model based on behavior is
`
`you use some laws, whether it's physical or chemical
`
`or whatever laws, to derive at whatever the model.
`
`That model can be simple or it can be very
`
`complicated. It depends on what you approximate. So
`
`there is no cut and dry to say the model has to be so
`
`sophisticated or less sophisticated. There will
`
`always be an approximation.
`
` Q It depends in part on the complexity of
`
`the system that is being modeled.
`
` A Exactly.
`
` Q And if the Kalman filter is being used as
`
`a filter to remove noise, I take it that the output
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
` So the modeling part allows you a very
`
`wide latitude of what to do.
`
` Q If you use a Kalman filter with a model,
`
`then you can have the output as either the same units
`
`or different units than the inputs.
`
` A You may. Depending on the model you're
`
`using.
`
` Q But we're -- well, let's turn to Workman
`
`again. I think -- it's right there marked as 1004,
`
`next to you -- no, right there.
`
` You know that Workman discloses the use
`
`of a Kalman filter, correct?
`
` A Correct.
`
` Q Let's go to that area of Workman. It's
`
`in column 3. And it's also, if you would like to
`
`look at Figure 2, obviously you are free to do that.
`
`That shows what the numbers mean graphically.
`
` It says: "The network solution system,
`
`10, implements a Kalman filter solution" --
`
` A Okay, let me get there. Where are you?
`
` Q Oh, sorry. Line 46.
`
` A Line 46. "Typically."
`
`432
`
`434
`
`1
`
`of the Kalman filter is in the same unit of
`
`2 measurement as the inputs. In other words, if --
`
`now, let's use the example of predicting -- or of
`
`using a Kalman filter for SPD locations. If
`
`filtering is going on, then the Kalman filter would
`
`output an estimate of the SPD locations based on the
`
`input of the measurements of SPD locations, right?
`
` A It depends. That's -- what you are
`
`saying in the pure filtering sense, yes, that's --
`
`that's what you may mean for it -- usually that's
`
`what you will mean for filter, that you put certain
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
` Q Yes. It says: "Typically the network
`
`solution system, 10, implements a Kalman filter
`
`solution on the signals it receives from the vessel
`
`positioning system, 20, and location sensing devices,
`
`15."
`
` Do you see that?
`
` A Yes.
`
` Q So that explains when a Kalman filter is
`
`used in Workman, right?
`
` A Yes.
`
` Q Okay. And it explains which signals or
`
`units in and you get certain units out, but it can be
`
`12 measurements go into the Workman Kalman filter,
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`different too. It all depends on what -- to the
`
`degree that you are using a model and to the degree
`
`to which you are using a filter.
`
` Q And if you are using a model, then the
`
`units can change.
`
` A Yes. You can go from forces to
`
`displacements. But also you can use -- they can be
`
`in the same units too because someone may model
`
`21 motion of a location of the streamer and get motion
`
`22
`
`in another point.
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`right?
`
` A Yes.
`
` Q And the inputs into the network solution
`
`system are the locations and the vessel positioning
`
`system, correct?
`
` A That's what it says. From the vessel
`
`positioning system, the signals it receives from the
`
`vessel positioning system and location sensing
`
`devices.
`
` Q And the vessel positioning system, among
`
`PLANET DEPOS
`888.433.3767 | WWW.PLANETDEPOS.COM
`
`PGS Exhibit 1092, pg. 4
`PGS v. WesternGeco (IPR2014-00688)
`
`PUBLIC REDACTED
`
`

`
`DEPOSITION OF MICHAEL S. TRIANTAFYLLOU, Sc.D, VOLUME 2
`CONDUCTED ON SATURDAY, MAY 23, 2015
`
`435
`
`437
`
`4 (Pages 435 to 438)
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`other information, can provide the velocity of the
`
`system, correct?
`
` A Where do you see that?
`
` Q Well, no, I'm just asking you. The
`
`vessel positioning system can provide information
`
`about the velocity, correct?
`
` A I have to -- to remember whether that's
`
`part of the system or not.
`
` Q Well --
`
` A Are you saying you can derive it from the
`
`position, the velocity?
`
` Q The vessel positioning system information
`
`can be used to obtain information about the velocity
`
`of the vessel, correct?
`
` A The velocity -- you are talking about the
`
`vessel itself, the ship.
`
` Q Yes.
`
` A So some sensing device that can
`
`provide the ship. So if the ship has a sensing, you
`
`can sense its velocity, yes.
`
` Q And the streamers are towed by -- or
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`column 3.
`
` A It will use the latest signals it has.
`
`That's the interpretation here when you say "the
`
`position."
`
` Q But those signals are not identified as
`
`realtime in that sentence, correct?
`
` A They must be realtime. I mean it's a
`
`system that works realtime.
`
` Q Well, does that sentence, column 3, lines
`
`46 through 48, identify the signals being received
`
`from the location sensing devices as realtime
`
`signals?
`
` MR. KIKLIS: Objection. Asked and
`
`answered.
`
` THE WITNESS: In the absence of a device
`
`that will store them, we have to assume that, yes,
`
`they are realtime.
`
`BY MR. BERL:
`
` Q It doesn't say that they are realtime,
`
`does it?
`
` A It does not specify whether they would be
`
`stored, so in the absence of storing, the signals
`
`attached to the ship, correct?
`
`436
`
`438
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
` A Yes.
`
` Q And the location measurements that are
`
`the inputs into the Kalman filter in the sentence we
`
`just read, column 3, lines 36 through 48, are not
`
`described as realtime location measurements, correct?
`
` A Can you specify what you mean by
`
`"realtime location measurements"?
`
` Q Well, it doesn't identify the
`
`9 measurements in the sentence we just read about the
`
`10
`
`inputs into the Kalman filter as realtime
`
`11 measurements, right?
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
` A Realtime versus something that was done a
`
`year ago?
`
` Q Something that's not realtime.
`
` A Well, if they were measured sometime
`
`earlier, yes. But it doesn't -- it doesn't say
`
`anything about the history of such signals, if that's
`
`what you are asking.
`
` Q No, what I'm asking is, it does not
`
`identify the signals that are the inputs into the
`
`Kalman filter as realtime measurements, right? That
`
`phrase is not used in the sentence we just read in
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`can't stay in thin air. It would have to have a
`
`separate system to somehow store them if they are not
`
`realtime.
`
` Q Now, let's maybe look at column 2. That
`
`can perhaps help us. If we look at the paragraph
`
`that begins on line 10 of column 2. Do you see that
`
`it says: "Location sensing devices and methods for
`
`determining the positions of the seismic sources and
`
`seismic streamer cables are also well known in the
`
`art"?
`
` Do you see that?
`
` A I see that.
`
` Q And you agree with that?
`
` A It depends on what location sensing
`
`devices and methods for determining means, whether
`
`these were to locate the -- they were used for
`
`locating the hydrophones for the purposes of knowing
`
`where the streamers were. Specifically location
`
`sensing devices, I wouldn't say that it was something
`
`that was practiced or established in the -- in the
`
`art because there was no such system working at the
`
`time.
`
`PLANET DEPOS
`888.433.3767 | WWW.PLANETDEPOS.COM
`
`PGS Exhibit 1092, pg. 5
`PGS v. WesternGeco (IPR2014-00688)
`
`PUBLIC REDACTED
`
`

`
`DEPOSITION OF MICHAEL S. TRIANTAFYLLOU, Sc.D, VOLUME 2
`CONDUCTED ON SATURDAY, MAY 23, 2015
`
`439
`
`441
`
`5 (Pages 439 to 442)
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
` Q Well, it then says: "For example, both a
`
`global positioning system as described in," and it
`
`cites a patent, "and a network of acoustic elements,"
`
`as described in another patent, "may be deployed on
`
`the vessel streamer cables and tail buoy."
`
` Do you see that?
`
` A Yes.
`
` Q Do you disagree with that?
`
` A I would say that there were these patent
`
`existing methods, but at the time when this is
`
`written, they were not practiced because there was no
`
`such system to determine the position of the -- of
`
`the buoys.
`
` Q There were --
`
` A Of the birds.
`
` Q There was no system that had acoustic
`
`elements deployed on a tail buoy; is that your
`
`testimony?
`
` A No. I'm saying there was not a working
`
`system that had multiple birds to be determined.
`
` Q That's not what I've asked. You agree
`
`that this sentence doesn't say anything here about
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
` Q And when you say that the hydrophones
`
`would send signals as part of this acoustic elements,
`
`how often would the signals typically be sent?
`
` A That's where the big difference is
`
`between a system which does exploration, so there are
`
`hydrophones which are there for the real job which is
`
`to find the oil, and then there are hydrophones to
`
`determine where the position of the system is. And
`
`those may take several seconds for the practical
`
`but not to fill the space with noise due to the
`
`positioning aspect. So there may be some time
`
`between one step and the next.
`
` Q So the acoustic elements that are used to
`
`determine the position sends signals every several
`
`seconds?
`
` A Typically.
`
` Q Okay. The next sentence says: "These
`
`devices and methods may then be used to determine the
`
`realtime position of the seismic sources and seismic
`
`streamer cables by computing a network solution to a
`
`Kalman filter as disclosed by United States Patent
`
`5,353,223."
`
`440
`
`442
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`that.
`
` This sentence says: "For example, both
`
`the GPS," as described in one patent, "and the
`
`network of acoustic elements," as described in
`
`another patent, "may be deployed on the vessel
`
`streamer cables and tail buoy."
`
` You don't disagree that that had been
`
`disclosed in the art, right?
`
` MR. KIKLIS: Objection. Form.
`
` THE WITNESS: It's referenced in those
`
`patents. Whether it has any practical applicability
`
`or whether it's a working system or anything else,
`
`that remains to be considered individually.
`
`BY MR. BERL:
`
` Q Okay. And the acoustic networks that are
`
`referenced in this sentence, how do those provide
`
`information about location?
`
` A The acoustic networks depends on the
`
`implementations, but the idea is to have hydrophones
`
`which sends signals, and by the time of travel
`
`underwater you can determine with triangulation where
`
`the various devices are.
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
` Does it say that?
`
` A It says that.
`
` Q And what does that mean?
`
` A He presumes that by using such a
`
`positioning -- acoustic positioning system, you can
`
`determine the realtime position of the seismic
`
`sources and streamer cables.
`
` Q Using a Kalman filter.
`
` A He presumes that.
`
` Q He states that.
`
` A He presumes it. He assumes that that's
`
`the case. He doesn't prove anything.
`
` Q I'm not asking about proof. That's what
`
`he says.
`
` A That's what he thinks.
`
` Q And I'm trying to understand what the
`
`Kalman filter is doing in this context.
`
` A The Kalman filter in this context may be
`
`doing -- for example, it depends on the user, but
`
`since it's not talking about a model-based, it may be
`
`a curve fitting. It takes the data and passes
`
`certain curves through the points.
`
`PLANET DEPOS
`888.433.3767 | WWW.PLANETDEPOS.COM
`
`PGS Exhibit 1092, pg. 6
`PGS v. WesternGeco (IPR2014-00688)
`
`PUBLIC REDACTED
`
`

`
`DEPOSITION OF MICHAEL S. TRIANTAFYLLOU, Sc.D, VOLUME 2
`CONDUCTED ON SATURDAY, MAY 23, 2015
`
`443
`
`445
`
`6 (Pages 443 to 446)
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
` Q It takes the data from the acoustic
`network?
`
` A From the acoustic network and passes
`certain curve fitting. It depends on what
`approximation you want to use.
` Q But it takes the data about locations
`from the acoustic network, right?
`
` A The data -- the sensing data, yes. The
`acoustic, let's say, position data.
` Q And then you say would produce a curve
`fitting.
`
` A Or it will filter them.
` Q Or a curve fitting to estimate the actual
`locations.
`
` A Trying to estimate the actual locations.
` Q And what it says is, produced by that
`process, that's used to determine the realtime
`position of the seismic sources and cables, correct?
`
` A That's what it says.
` Q And when you say "curve fitting," can you
`describe what the curve represents? Is the curve a
`depiction of location along the streamer?
`
`1
`
` A That's one way that you assume that --
`
`444
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`1
`
`2
`
` A Curve fitting is curve fitting.
`
` Q That means something to you. It may mean
`
`something different to me who doesn't have your
`
`expertise. I just want to make sure I understand.
`
` A Yes. And I don't want to confuse
`
`estimation which is based on models with estimate
`
`based on curve fitting. Curve fitting means you make
`
`quick and dirty assumptions, so to speak, so you get
`
`some answer. So it's as legitimate as, you know, the
`
`case may be.
`
` Q And when you say "as legitimate," it may
`
`be as precise or accurate --
`
` A Yeah, it may be very inaccurate. That's
`
`what I'm saying.
`
` Q And it may help, just so we understand,
`
`if you could draw for me what you understand to be
`
`the outputs of the Kalman filter in this column 2
`
`disclosure so that we understand what you mean by
`
`curve fitting.
`
` A Okay.
`
` Q Would you be able to do that?
`
` A Sure. So let's say you are receiving
`
`data, and this is the real line.
`
`446
`
`2
`
`you know, usually you have -- you don't have too many
`
` Q When you say "the real line," what do you
`
`3 measurements along the line. So what happens in
`
`3 mean by that?
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`between, you assume some continuous curve. So it --
`
`in the most sort of elemental way, you would put
`
`straight lines, but that cannot be true because it
`
`cannot be a broken line. Or it can be a quadratic
`
`curve or other smoother curves trying to arrive at an
`
`estimate.
`
` Q And one ends up with a curve that
`
`represents the positions of every point along the
`
`streamer as a function of where in the streamer you
`
`are?
`
` A And there is severe uncertainty in the
`
`in-between because it depends on the model, on the
`
`assumption you used.
`
` Q And what you're trying to do is provide
`
`an estimate of the actual locations using this Kalman
`
`filter, correct?
`
` A A curve fitting.
`
` Q A curve fitting to provide an estimate of
`
`the actual locations.
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
` A That is how the streamer array is right
`
`now. The true streamer array.
`
` Q And so can you maybe label that "A."
`
` A So let's put the ship here (indicating),
`
`and somehow it came out with an otter board here and
`
`a tail buoy here, and it has a few birds.
`
` Q And so I just want to be clear since this
`
`is going to be in the transcript. You've drawn the
`
`vessel at the top in a triangle --
`
` A So an otter board or a front buoy and a
`
`tail buoy to keep the tension in the line.
`
` Q Okay. And then --
`
` A And then this is the ship. And then this
`
`is number 1.
`
` Q That's bird number 1?
`
` A Bird number 1. And then this is the
`
`bird N. And there is an N plus 1 and N minus 1.
`
` Q Okay.
`
` A So some acoustical signal sends you the
`
`PLANET DEPOS
`888.433.3767 | WWW.PLANETDEPOS.COM
`
`PGS Exhibit 1092, pg. 7
`PGS v. WesternGeco (IPR2014-00688)
`
`PUBLIC REDACTED
`
`

`
`DEPOSITION OF MICHAEL S. TRIANTAFYLLOU, Sc.D, VOLUME 2
`CONDUCTED ON SATURDAY, MAY 23, 2015
`
`447
`
`449
`
`7 (Pages 447 to 450)
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`information. So it tells you bird N is here.
`
` Q And you put an X in your --
`
` A Yes.
`
` Q -- to the left of bird N.
`
` A Yes, measurement. Measurement N. And it
`
`gives you -- for one, let's say you are lucky here
`
`and you got a really good estimate.
`
` Q Were you've put an X --
`
` A Measurement in number 1.
`
` Q And you put that next to bird number 1?
`
` A Right.
`
` Q Okay.
`
` A So now you say not yet, I have somehow to
`
`do something about finding where my curve is. Okay.
`
` Q Where your streamer is.
`
` A Where my streamer is.
`
` MR. KIKLIS: Objection. Form.
`
`BY MR. BERL:
`
` Q Okay.
`
` A So you are going to connect with some
`
`kind of curves. So the simplest thing would say --
`
` Q Do we have another color? Let's get
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
` A Yes. So that gives --
`
` MR. KIKLIS: Objection. Form.
`
` THE WITNESS: That gives you an idea of
`
`the general picture of curve fitting.
`
`BY MR. BERL:
`
` Q And that curve fitting, as you've
`
`described it where you've connected the two
`
`8 measurements, the two Xs, that's what's done by the
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`Kalman filter.
`
` A No.
`
` Q No.
`
` A The curve fitting depends on the user.
`
`It could be a straight line or it could be something
`
`else. I'm just saying one simple way would be this,
`
`which will be real bad because the -- it's missing
`
`any information and anything else. And it shows that
`
`the -- there is this process where curve fitting is
`
`curve fitting. So ...
`
` Q And my question is, what is the Kalman
`
`filter that's described in column 2 that's being used
`
`to determine the realtime positions? What is that
`
`doing in the context of what you've drawn?
`
`448
`
`450
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`another color so that it's comprehensible.
`
` MR. KIKLIS: What is the question?
`
` THE WITNESS: Yeah, where are we leading
`
`to this? You want to see how a curve fitting works?
`
`BY MR. BERL:
`
` Q Yeah. We don't have another color, so
`
`sorry. Continue.
`
` A So the estimate would be to say, Here it
`
`is.
`
` MR. KIKLIS: Objection. Form. Is there
`
`even a question?
`
`BY MR. BERL:
`
` Q And so you've drawn a line that connects
`
`the two Xs that were in the measurements.
`
` A For example.
`
` Q And that's --
`
` A -- one way of -- of connecting the
`
`points. Okay. So you can see from this that the
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
` MR. KIKLIS: Objection. Form.
`
` THE WITNESS: The Kalman filter, we don't
`
`have any idea because it's not disclosed. It's not
`
`disclosing anything here. We're doing a hypothetical
`
`of what the Kalman filter which -- or how it can be
`
`used. It's not the Kalman filter that is in this
`
`disclosure. It has nothing to do with it
`
`necessarily. It may have nothing to do with it. But
`
`the Kalman filter is, in its simplest form, in its
`
`filtering form, it's least squares.
`
`BY MR. BERL:
`
` Q And what I'm trying to understand is what
`
`a person would understand the Kalman filter that's
`
`described here in Workman's column 2 that's being
`
`used to determine the realtime positions, what does
`
`it mean that it is being used to determine the
`
`realtime positions?
`
` A It's unspecified. It doesn't say
`
`immediate thing that will happen is that it has
`
`19 model-based, so it could be this or a million other
`
`this -- this curve which is now traveling down.
`
` Q It's missed the curve that's between
`
`bird 1 and bird N?
`
`20
`
`21
`
`22
`
`things which have nothing to do with, you know --
`
`it's least squares fitting. That's how the Kalman
`
`filter is used when you are saying it. You know what
`
`PLANET DEPOS
`888.433.3767 | WWW.PLANETDEPOS.COM
`
`PGS Exhibit 1092, pg. 8
`PGS v. WesternGeco (IPR2014-00688)
`
`PUBLIC REDACTED
`
`

`
`DEPOSITION OF MICHAEL S. TRIANTAFYLLOU, Sc.D, VOLUME 2
`CONDUCTED ON SATURDAY, MAY 23, 2015
`
`451
`
`453
`
`8 (Pages 451 to 454)
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`1
`
`2
`
`least squares -- perhaps I should explain that.
` Q No, I understand that. And it's being
`
`used in one form or another to estimate what the
`
`locations are along the streamer.
`
` A To do a least squares --
` MR. KIKLIS: Objection. Form.
`
` THE WITNESS: To do a least squares fit.
`
`That's the simplest interpretation.
`
` MR. BERL: Why don't we mark what you've
`
`drawn --
`
` MR. KIKLIS: It's least squares.
`
` THE WITNESS: Least.
`
` MR. BERL: Why don't we mark what you've
`
`drawn as Exhibit 1084.
`
` (Exhibit No. 1084 was marked for
`
` identification.)
`
`BY MR. BERL:
`
` Q So let's go back then to column 3 again
`
`of Workman. And let's look at the next sentence
`
`after the one we read before. So the sentence
`
`beginning at line 48 of column 3.
`
` "The network solution system, 10, outputs
`
`realtime streamer cable shapes, streamer cable
`
`positions, and streamer cable separations."
`
`452
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`1
`
`2
`
`3
`
` MR. KIKLIS: Objection. Form.
` THE WITNESS: It does not output the
`realtime streamer cable shapes. It produces some
`guess at those, at all those things.
`BY MR. BERL:
` Q It says that it outputs realtime streamer
`cable shapes. Are you just saying that that's not
`true?
` A That's what he hopes they produce.
` Q Well, that's what it's disclosed to
`produce. You're just saying you disbelieve the
`disclosure.
` A Do you believe everything you read?
` MR. KIKLIS: Objection. Form.
`BY MR. BERL:
` Q Well, I'm asking the questions here.
` MR. KIKLIS: Objection to form.
`BY MR. BERL:
` Q You can tell me if you don't believe what
`it says, then you don't believe what it says. Is
`that your testimony?
` A My testimony is that the -- there is an
`
`estimation process, and without the model, in the
`
`general case this is totally untrue. If he says that
`
`454
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
` Is that right?
`
` A That's what it states.
`
` Q And those are different than the inputs
`
`into the Kalman filter.
`
` A Different in what sense?
`
` Q Well, the inputs that we looked at were
`
`the signals received from the vessel positioning
`
`system and location sensing devices, right?
`
` A Right.
`
` Q And the outputs are not the locations or
`
`something from the vessel positioning system, but
`
`rather realtime streamer cable shapes, streamer cable
`
`positions, and streamer cable separations, right?
`
` A So different means it's not what got in?
`
` Q Yes.
`
` A It states that they are the realtime
`
`streamer cable shapes, cable positions and streamer
`
`capable separations, so it doesn't reproduce the
`
`input which is -- what it is.
`
` Q Which is true, right?
`
`he is going to output realtime streamer cable shapes,
`
`4 meaning they're real, what he means is he will output
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`some real cable shapes which he hopes have something
`
`to do with the real, with the actual. So in that
`
`sense, in the strict sense of your question, the
`
`statement is wrong.
`
`BY MR. BERL:
`
` Q So because in order to produce realtime
`
`streamer cable shapes, in your view you would need to
`
`use a model.
`
` MR. KIKLIS: Objection. Form.
`
` THE WITNESS: In the general case, there
`
`is noise that you will never be able to reproduce the
`
`actual. There will always be some deviation. The
`
`question is whether it will be reasonably close or
`
`far away.
`
` What I'm saying is this not only does not
`
`guarantee that he will output the realtime streamer
`
`cable shapes, which is in general impossible because
`
`there will be always some inaccuracies, okay, it may
`
`PLANET DEPOS
`888.433.3767 | WWW.PLANETDEPOS.COM
`
`PGS Exhibit 1092, pg. 9
`PGS v. WesternGeco (IPR2014