`
`
`Ex. PGS 1018
`EX. PGS 1018
`(EXCERPTED)
`(EXCERPTED)
`
`
`
`
`
`
`
`
`
`
`
`Case 4:09-cv-01827 Document 449 Filed in TXSD on 07/30/12 Page 1 of 390
`
`1209
`
`UNITED STATES DISTRICT COURT
`SOUTHERN DISTRICT OF TEXAS
`HOUSTON DIVISION
`
`09-CV-1827
`Houston, Texas
`
`7:39 a.m.
`July 27, 2012
`
`*
`*
`
`**
`
`***
`
`WESTERNGECO LLC
`VS.
`ION GEOPHYSICAL
`CORPORATION, FUGRO
`GEOTEAM, INC., ET AL
`
`JURY TRIAL
`Volume 5
`
`BEFORE THE HONORABLE KEITH P. ELLISON
`UNITED STATES DISTRICT JUDGE
`
`APPEARANCES:
`FOR THE PLAINTIFF:
`Lee L. Kaplan
`SMYSER, KAPLAN & VESELKA, LLP
`700 Louisiana, Suite 2300
`Houston, Texas 77002
`713.221.2300
`
`Gregg F. LoCascio
`KIRKLAND & ELLIS LLP
`655 Fifteenth Street Northwest
`Washington, DC 20005
`202.879.5290
`
`Sarah Tsou
`Timothy K. Gilman
`KIRKLAND & ELLIS LLP
`Citigroup Center
`153 East 53rd Street
`New York, New York 10022
`212.446.6435
`
`Johnny C. Sanchez, RMR, CRR - jcscourtreporter@aol.com
`
`123456789
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`Ex. PGS 1018
`
`
`
`Case 4:09-cv-01827 Document 449 Filed in TXSD on 07/30/12 Page 68 of 390
`Direct-Triantafyllou
`
`1276
`
`and where you're going with it. In every day language
`usually we say, now is -- the beginning is now, and future
`is ahead. But if you defined the beginning to be the
`depression, the future was rise forward from the
`depression. Okay?
`If we can go to the next slide. Can you give us an
`Q.
`example of some types of software that do this prediction
`to determine based on information you know, what a system
`will or what the mathematics at now or some point after
`the measurement?
`A.
`So there are a number of methodologies. Some of them
`you're going to hear. That's why I put some of them down.
`These are mathematical methods to do exactly what we spoke
`before, take something from a time and move it forward.
`So there's a system called an observer because it observes
`the system and looks forward.
`There's an ultimate observer, which is
`called the Kalman filter because Professor Rudy Kalman
`developed it.
`We've heard that term I think a little bit yesterday,
`Q.
`something called a Kalman filter?
`A.
`Yes.
`And that's this optimal observer?
`Q.
`A.
`Yes. It's an optimal observer.
`And it's named after a scientist?
`Q.
`
`123456789
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`Johnny C. Sanchez, RMR, CRR - jcscourtreporter@aol.com
`
`09:00:27
`
`09:00:44
`
`09:01:01
`
`09:01:14
`
`09:01:24
`
`Ex. PGS 1018
`
`
`
`Case 4:09-cv-01827 Document 449 Filed in TXSD on 07/30/12 Page 71 of 390
`Direct-Triantafyllou
`
`1279
`
`We have no idea.
`
`So if you have the Kalman filter on your
`car, what would that Kalman filter do? Would get your
`speedometer to see how fast you go, and which direction
`you go, west, north and south, and will say, based on your
`speed you have to be where this green place is. Makes a
`very simple calculation, velocity 30 miles per hour, times
`half an hour, so much in which direction.
`So it finds -- it gives you an estimate.
`It makes a prediction. Why is it a prediction? Why do
`you call it prediction? You don't have any data to
`corroborate. All you do is you take your own speed. You
`say, hey, I must be there.
`So what happens? The GPS comes in and the
`GPS is the yellow dot. The GPS is very accurate and the
`Kalman filter says hey, last week this guy changed the
`tires, so the speedometer doesn't work very well. The GPS
`is what I will believe. So, yes, I'm here. You got lucky
`this time.
`
`In the other example on the right, the GPS
`comes and shows you you are there. There's no way you're
`on the highway. You know that that's not the case.
`What I predict this green spot is good
`enough for me. So you feel confidence at least of where
`you are. That's what a predictor does for you. Okay? It
`
`123456789
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`Johnny C. Sanchez, RMR, CRR - jcscourtreporter@aol.com
`
`09:04:34
`
`09:04:52
`
`09:05:06
`
`09:05:26
`
`09:05:43
`
`Ex. PGS 1018
`
`
`
`Case 4:09-cv-01827 Document 449 Filed in TXSD on 07/30/12 Page 75 of 390
`Direct-Triantafyllou
`
`1283
`
`buy a television they give you the manual. They have such
`detailed manuals, which are very well written, both by
`Q-FIN and as we see from DigiFIN, they write very nice
`manuals and explain how things work.
`So in this case they say explicitly, "Due
`to low sample rate of the position observations," meaning
`the GPS is another example, doesn't come very often. "The
`software runs also a position predictor. The Kalman
`filter predicts where you are."
`So looking at the sentence that you've highlighted,
`Q.
`the low sample rate of the position observations, is that
`explaining that you don't always get an actual measurement
`of where the bird is?
`A.
`Exactly. It takes several seconds before the new
`observation comes. In the meanwhile, as we know those
`waves don't wait for you. They keep traveling. So you do
`this prediction to see where everything is to keep tab.
`And based on your analysis, was the WesternGeco
`Q.
`Q-Marine system covered by its own patent?
`A.
`Yes. So it contains those elements.
`If we can now look at the next slide, did you take a
`Q.
`look at ION's Q systems?
`A.
`Yes. I looked at the manuals of the DigiFIN and the
`related technology to the DigiFIN. And also, I went to
`the Websites of ION and looked at the product. And also I
`
`123456789
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`Johnny C. Sanchez, RMR, CRR - jcscourtreporter@aol.com
`
`09:10:42
`
`09:10:57
`
`09:11:12
`
`09:11:28
`
`09:11:51
`
`Ex. PGS 1018
`
`
`
`Case 4:09-cv-01827 Document 449 Filed in TXSD on 07/30/12 Page 132 of 390
`Direct-Triantafyllou/By Mr. LoCascio
`
`1340
`
`controller, and it sends the FIN angles from the lateral
`controller to the inwater devices, the fins.
`We turn to the next slide.
`Q.
`Did you also see documents that showed
`that the calculation of what that FIN angle would be to
`send that location information, that's done in the lateral
`controller?
`A.
`Yes. So if we look at this graph, I mean it's
`basically very simple, what you see on the bottom is the
`area, how far you are. The more you are, the more force.
`So you see that the curve goes up, and that is the angle
`you want. The more you are away, the more angle you want,
`and it's just put in the form of a graph and in the form
`of an equation, an algorithm, that does this.
`So this distance is how far away you are from where
`Q.
`you want the streamer to be?
`A.
`Exactly. So it's the difference between the two
`location informations that we said. So the FIN angle
`requires where you are and where you want to be. The
`location information is encapsulated inside the FIN angle.
`Now, if we turn to the next slide, was it indicated
`Q.
`again by ION's manager, Mr. MacNab, where this calculation
`actually takes place?
`A.
`Yes. And he concurs that the location information is
`sent to the lateral controller, and the lateral controller
`
`123456789
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`Johnny C. Sanchez, RMR, CRR - jcscourtreporter@aol.com
`
`10:27:02
`
`10:27:18
`
`10:27:34
`
`10:27:49
`
`10:28:06
`
`Ex. PGS 1018
`
`
`
`Case 4:09-cv-01827 Document 449 Filed in TXSD on 07/30/12 Page 137 of 390
`Direct-Triantafyllou/By Mr. LoCascio
`
`1345
`
`C and D. The Claim 15 is just A, B and C.
`Is there is this substantive difference or
`is there they just broke the paragraphs at different
`places?
`A.
`Well, it's a method, so it explains one more item
`that is needed to exercise.
`And did you look for all of those?
`Q.
`A.
`Yes.
`If we could furnish to your next slide. Can you
`Q.
`explain how ION system uses predictor software?
`A.
`So ORCA, which is the intelligence officer, so to
`speak, of the DigiFIN overall system, uses a specific
`methodology. It's called a Kalman filter. It's probably
`one of the most widely used for the -- for various.
`Is this the Kalman filter you talked about earlier?
`Q.
`A.
`It's the Kalman filter I talked earlier, and the one
`that you can put on your car to give you better
`predictions of where you are.
`And so, they use a Kalman filter to
`predict the position and velocity of each node and uses
`the measurements and predictions together to update the
`position and velocity. In other words, to tell you where
`you are now. Okay?
`So let's take the scenario just quickly
`again. You take measurements. The hydrophones, they tell
`
`123456789
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`Johnny C. Sanchez, RMR, CRR - jcscourtreporter@aol.com
`
`10:33:20
`
`10:33:32
`
`10:33:57
`
`10:34:10
`
`10:34:25
`
`Ex. PGS 1018
`
`
`
`Case 4:09-cv-01827 Document 449 Filed in TXSD on 07/30/12 Page 138 of 390
`Direct-Triantafyllou/By Mr. LoCascio
`
`1346
`
`you where you are. But the next time they're going to
`tell you when 10 seconds from now. It's a long time. Why
`10 seconds? It takes time to process the signal and the
`like.
`
`In that 10 seconds, you don't have any
`information what's going on. All you know is where you
`were. So you take your system and you say, okay, I'll do
`do a prediction. You'll use the equations of the system.
`It's not just taking a wild guess. And it tells you,
`You're going to be here at now. That's where you are.
`That's a prediction.
`Then the signals come in, and the Kalman
`filter has an ingenious algorithm to decide what is noise
`because some of these measurements are out. It's like a
`GPS that gives you crazy readings. But it's not doing
`just that. It combines that. So that doesn't throw out
`the measurements. It does a careful job. So it's a good
`algorithms to combine prediction and actual noisy
`measurements. There's no clear measurement in such field.
`They're all noisy.
`If we go to the next page, does the patent itself
`Q.
`talk about using predictor software to estimate this
`actual location of the bird?
`A.
`So if we look at the patent, which is in the upper
`left, it's one of the pages of the patent. It says you do
`
`123456789
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`Johnny C. Sanchez, RMR, CRR - jcscourtreporter@aol.com
`
`10:34:40
`
`10:34:56
`
`10:35:11
`
`10:35:33
`
`10:35:48
`
`Ex. PGS 1018
`
`
`
`Case 4:09-cv-01827 Document 449 Filed in TXSD on 07/30/12 Page 142 of 390
`Direct-Triantafyllou/By Mr. LoCascio
`
`1350
`
`prediction and filtering works in the accused ION system?
`A.
`Yes, indeed.
`MR. LOCASCIO: Turn to 69, Dave.
`THE WITNESS: If we turn to 69.
`BY MR. LOCASCIO:
`What did you say in there?
`Q.
`A.
`We see the Kalman filter and an arrow that says "raw"
`next to it. "Raw" means the data as they come. We
`haven't filtered and we haven't done anything to them.
`That's what "raw" means.
`And this produces an output, which is
`going to be used next. So the Kalman filter, as it says
`in the bottom, the Kalman filter part runs all the time,
`grabbing data and using that to update its position and
`velocity and streamer shape and arrow.
`And so, is the Kalman filter used by ION used to
`Q.
`estimate the current position of the birds?
`A.
`Yes, it is.
`In the next slide, are there other flow charts
`Q.
`identifying this Kalman filtering process in their
`operation document?
`A.
`Yes, it is. And here we see some of the prediction
`unit and the like, which is part of the Kalman filter. We
`don't need to go into any specific detail, but every
`Kalman filter has a predictor and an adjustment. The
`
`123456789
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`Johnny C. Sanchez, RMR, CRR - jcscourtreporter@aol.com
`
`10:39:20
`
`10:39:33
`
`10:39:54
`
`10:40:08
`
`10:40:25
`
`Ex. PGS 1018
`
`
`
`Case 4:09-cv-01827 Document 449 Filed in TXSD on 07/30/12 Page 143 of 390
`Direct-Triantafyllou/By Mr. LoCascio
`
`1351
`
`filter maintains its idea of time, meaning has a clock, so
`it can tell what time it is.
`When new data arise, the predicator is
`used to carry the state forward to time of observation.
`Time of observation is now, when the observation comes in.
`It cannot be the future in terms of the Kalman filter.
`Then the predictions of the data are
`thrown into the adjustment. So it's like a blender. You
`put them in and the algorithms comes up with an optimal
`solution. That's all there is to it.
`This passage here that you just read, do ION's own
`Q.
`documents talk about using the predictor to carry the
`state forward to time of observation, being present or
`actual time?
`A.
`Yes. Because in order to have observations, it can
`be at most now.
`We can't have the future observations?
`Q.
`A.
`The future will come later.
`If we look at the next slide, did ION's own employees
`Q.
`recognize, in your review of the deposition testimony?
`A.
`Yes. This was -- sometimes, you know, documents that
`were earlier to confirm this, and he's asked if it's part
`of the NCN, and he said yes.
`"So the Kalman filter is predicting the
`position of the DigiFIN node?
`
`123456789
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`Johnny C. Sanchez, RMR, CRR - jcscourtreporter@aol.com
`
`10:40:41
`
`10:40:56
`
`10:41:09
`
`10:41:25
`
`10:41:41
`
`Ex. PGS 1018
`
`
`
`Case 4:09-cv-01827 Document 449 Filed in TXSD on 07/30/12 Page 144 of 390
`Direct-Triantafyllou/By Mr. LoCascio
`
`1352
`
`"It's predicting the position" --
`THE COURT: No, you're going too fast.
`THE WITNESS: I'm sorry.
`BY MR. LOCASCIO:
`The second passage you refer to --
`Q.
`A.
`I'm sorry.
`That's okay.
`Q.
`-- Mr. MacNab was asked: "So the Kalman
`filter is predicting the positions of the DigiFIN node:
`And his answer was: "It's predicting the
`positions of all nodes."
`And then on the next question, he was
`asked: "The Kalman filter is predicting the position of
`the DigiFIN devices, that's the birds?"
`And he answered?
`A.
`"Yes." So the nodes correspond to the device. The
`fins, the DigiFINs that are spread around.
`So based on the information of how they use it and
`Q.
`the algorithms itself, and Mr. MacNab and other testimony
`you identified the limitation of a prediction unit as
`being met?
`A.
`Exactly.
`Now, I want to ask you about this particularly
`Q.
`because there's specific software that runs this Kalman
`filter; right?
`
`123456789
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`Johnny C. Sanchez, RMR, CRR - jcscourtreporter@aol.com
`
`10:41:46
`
`10:41:55
`
`10:42:05
`
`10:42:21
`
`10:42:33
`
`Ex. PGS 1018
`
`
`
`Case 4:09-cv-01827 Document 449 Filed in TXSD on 07/30/12 Page 186 of 390
`Cross-Triantafyllou/By Mr. Pierce
`
`1394
`
`regarding location. What did you use?
`A.
`Okay. In my analysis, I considered what -- location
`information, but then what the location system is doing
`with location information.
`So location information consists of in
`every case, where you want to be, and where you are. It
`can be the difference of the two, it can be some
`combination of the two, it does have the two components
`that don't need to exist separately, they can be combined,
`the essence of it is you need that location information
`processed in some form, raw, or processed.
`Okay. And so location information could include I
`Q.
`think we talked about latitude and longitude; correct?
`A.
`It could.
`It could include depth?
`Q.
`A.
`It could.
`It could include a lateral position?
`Q.
`A.
`Yes.
`And the FIN angle itself as I understand your
`Q.
`opinion, is location information because those types of
`those pieces that I just described, or at least some of
`them, were used in a calculation that been input of a FIN,
`is that fair?
`A.
`That's fair.
`And that calculation happened both in ORCA and in
`Q.
`
`123456789
`
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`Johnny C. Sanchez, RMR, CRR - jcscourtreporter@aol.com
`
`12:28:51
`
`12:29:10
`
`12:29:22
`
`12:29:33
`
`12:29:46
`
`Ex. PGS 1018
`
`