`____________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`_____________
`
`
`CISCO SYSTEMS, INC., DISH NETWORK, LLC,
`COMCAST CABLE COMMUNICATIONS, LLC,
`COX COMMUNICATIONS, INC.,
`TIME WARNER CABLE ENTERPRISES LLC,
`VERIZON SERVICES CORP., and ARRIS GROUP, INC.,
`Petitioner
`
`v.
`
`TQ DELTA, LLC,
`Patent Owner
`
`_____________________
`
`Case IPR2016-010201
`Patent 9,014,243
`
`
`
`
`PETITIONER’S RESPONSE TO PATENT OWNER’S MOTION FOR
`OBSERVATION ON CROSS-EXAMINATION
`
`
`
`
`
`
`1 DISH Network, LLC, who filed a Petition in IPR2017-00254, and Comcast Cable
`Communications, LLC, Cox Communications, Inc., Time Warner Cable
`Enterprises LLC, Verizon Services Corp., and ARRIS Group, Inc., who filed a
`Petition in IPR2017-00418, have been joined in this proceeding.
`
`
`
`Petitioner’s Response to Patent Owner’s Motion for Observation
`
`IPR2016-01020
`This response is submitted in view of the Scheduling Order (Paper 8) and the
`
`Trial Practice Guide, 77 Fed. Reg. 48756, 48767–68 (Aug. 14, 2012). This paper
`
`responds to Patent Owner’s Motion for Observation on Cross-examination (Paper
`
`27) filed on June 30, 2017, in the present inter partes review. Patent Owner
`
`presented ten observations on the June 20, 2017, deposition testimony of Dr.
`
`Tellado (Ex. 2013). Petitioner responds to each of Patent Owner’s observations
`
`below.
`
`Response to Observation # 1:
`
`TQ Delta omits portions of Dr. Tellado’s testimony where he explained that
`
`his simulation is not limited to just one combination, but instead applies to many
`
`combinations of high attenuation and/or high crosstalk:
`
`Q. You did use them in the simulation?
`A. I’m saying, and I repeat, the simulation is a transmitter
`simulation. It’s modeling the case for which we have 182
`random carriers and 52 Shively carriers. It includes and it
`models many combinations of high attenuation and/or high
`crosstalk.
`
`***
`
`There is many loops that would come up with the answer 182
`pseudorandom carriers and 52 Shively carriers.
`There is lots of combinations of lengths, crosstalk, attenuation,
`noise that would lead to that answer. And one example is that a
`
`2
`
`
`
`Petitioner’s Response to Patent Owner’s Motion for Observation
`
`IPR2016-01020
`standard with that loop length and that gauge and that noise
`floor.
`Ex. 2013, 34:21-35:2 & 84:14-21. Dr. Tellado also explained that Shively’s
`
`technique is not limited to 18,000-foot loops, but instead applies to many different
`
`combinations of line length and other factors:
`
`Q. Okay. And the example that Shively provides for high
`attenuation is long loops of 18,000 feet or greater; right?
`MR. McDOLE: Objection; form.
`THE WITNESS: Shively includes cables for which the
`relationship between attenuation and/or crosstalk or noise is
`such that some bits are stressed and need replication. It
`includes a plurality of combinations across cable types, gauges,
`taps, lengths, crosstalk. There is a lot of combinations that
`Shively could apply to.
`
`***
`Q. And the example he provides is an 18,000-foot loop; right?
`MR. McDOLE: Objection; form.
`THE WITNESS: He includes many combinations, and he
`includes a sentence that says “order of 18,000 feet.” “Order
`of.”
`BY MR. McANDREWS:
`Q. Or more?
`A. But “order of 18,000 feet” includes cables that are less than
`18,000 feet.
`Q. Right.
`
`3
`
`
`
`Petitioner’s Response to Patent Owner’s Motion for Observation
`
`IPR2016-01020
`Ex. 2013, 25:13-23 & 27:9-19. Dr. Tellado’s testimony is relevant to Petitioner’s
`
`argument that “Shively is not limited to cable lengths of 18,000 feet.” Petitioner’s
`
`Reply, Paper 17, p. 22.
`
`Response to Observation # 2:
`
`TQ Delta mischaracterizes Dr. Tellado’s testimony because the cited
`
`statements from Dr. Tellado do not discuss the noise characteristics used by Dr.
`
`Short. See Ex. 2013, 46:1-5. Dr. Tellado did not agree that Dr. Short’s analysis
`
`was based on a loop with high noise. Dr. Tellado’s declaration provided annotated
`
`graphs showing that ADSL systems could have noise levels much higher than
`
`“-140 dBm/Hz, which was the ‘background noise’ level shown in the attenuation
`
`graph relied upon by Dr. Short.” Ex. 1026, ¶ 9; see also id., ¶¶ 10-13.
`
`Response to Observation # 3:
`
`TQ Delta’s observation is misleading because Dr. Tellado did not rely on the
`
`18,000-foot “quick estimate” to state in paragraph 29 of his declaration (Ex. 1026)
`
`that “Dr. Short’s analysis is flawed….” PO’s Motion for Observation, p. 4. Rather,
`
`to show that Dr. Short’s use of a Gaussian approximation was flawed, Dr. Tellado
`
`relied on the simple math and logic explained in paragraphs 15-28 of his
`
`declaration (Ex. 1026). TQ Delta also ignores Dr. Tellado’s testimony explaining
`
`with “Graph 2” of Ex. 1026 (p. 30) that a system using Shively’s technique cannot
`
`be modeled accurately with a Gaussian approximation:
`
`4
`
`
`
`Petitioner’s Response to Patent Owner’s Motion for Observation
`
`IPR2016-01020
`A. Can you see graph 2? You see the solid red line?
`Q. Yes.
`A. That is the Gaussian process. You see Scenario 1 with 250
`QAM-4 carriers? Doesn’t it look very similar to the Gaussian
`process, the round, blue circles?
`Q. Yes.
`A. So I call this following a Gaussian process. What about the
`cyan Scenario 4 curve? Does it -- is it tight with the Gaussian
`process Scenario 2 curve? It's diverging. It’s worse than. The
`Gaussian process has a lower PAR than Scenario 4. It’s not a
`good model. This shows you a Scenario 4 that has Shively
`carriers cannot be modeled accurately with a Gaussian process
`in Scenario 2.
`Ex. 2013, 49:13-50:5.
`
`Response to Observation # 4:
`
`TQ Delta’s observation is misleading because Dr. Tellado did not rely on the
`
`18,000-foot “quick estimate” to state in paragraph 29 of his declaration (Ex. 1026)
`
`that “Dr. Short’s analysis is flawed….” PO’s Motion for Observation, p. 5. Rather,
`
`to show that Dr. Short’s use of a Gaussian approximation was flawed, Dr. Tellado
`
`relied on the simple math and logic explained in paragraphs 15-28 of his
`
`declaration (Ex. 1026). TQ Delta also ignores Dr. Tellado’s explanation for why
`
`he could not guess how a non-Gaussian process, such one using Shively’s bit-
`
`spreading technique, would compare to a Gaussian process:
`
`5
`
`
`
`Petitioner’s Response to Patent Owner’s Motion for Observation
`
`IPR2016-01020
`Q. Well, would it be to the left of the Scenario 1 line that you
`show here in graph 2?
`A. … So when you have a Gaussian process, it drops like the
`red line; if you have a non-Gaussian process, it could have a
`different slope. And it could look better at very high clipping
`rates, but it could be worse at low clipping rates. So I don’t
`remember exactly where it crossed. It didn’t have the same
`slope as a Gaussian process. That’s why I said the Gaussian
`approximation was not a good approximation.
`
`
`Ex. 2013, 55:8-56:2.
`
`Response to Observation # 5:
`
`TQ Delta’s observation is misleading because Dr. Tellado did not rely on the
`
`18,000-foot “quick estimate” to state in paragraph 29 of his declaration (Ex. 1026)
`
`that “Dr. Short’s analysis is flawed … .” PO’s Motion for Observation, p. 7.
`
`Rather, to show that Dr. Short’s use of a Gaussian approximation was flawed, Dr.
`
`Tellado relied on the simple math and logic explained in in paragraphs 15-28 of his
`
`declaration (Ex. 1026). TQ Delta also ignores Dr. Tellado’s testimony explaining
`
`with “Graph 2” of Ex. 1026 (p. 30) that a system using Shively’s technique cannot
`
`be modeled accurately with a Gaussian approximation:
`
`A. Can you see graph 2? You see the solid red line?
`Q. Yes.
`
`6
`
`
`
`Petitioner’s Response to Patent Owner’s Motion for Observation
`
`IPR2016-01020
`A. That is the Gaussian process. You see Scenario 1 with 250
`QAM-4 carriers? Doesn’t it look very similar to the Gaussian
`process, the round, blue circles?
`Q. Yes.
`A. So I call this following a Gaussian process. What about the
`cyan Scenario 4 curve? Does it -- is it tight with the Gaussian
`process Scenario 2 curve? It's diverging. It’s worse than. The
`Gaussian process has a lower PAR than Scenario 4. It’s not a
`good model. This shows you a Scenario 4 that has Shively
`carriers cannot be modeled accurately with a Gaussian process
`in Scenario 2.
`
`
`Ex. 2013, 49:13-50:5.
`
`Response to Observation # 6:
`
`TQ Delta mischaracterizes Dr. Tellado’s testimony by omitting Dr.
`
`Tellado’s explanation for why a non-Gaussian process (such as in Shively) cannot
`
`be intuitively compared to a Gaussian process:
`
`A. … So when you have a Gaussian process, it drops like the
`red line; if you have a non-Gaussian process, it could have a
`different slope. And it could look better at very high clipping
`rates, but it could be worse at low clipping rates. So I don’t
`remember exactly where it crossed. It didn’t have the same
`slope as a Gaussian process. That’s why I said the Gaussian
`approximation was not a good approximation.
`
`
`Ex. 2013, 55:19-56:2.
`
`7
`
`
`
`Petitioner’s Response to Patent Owner’s Motion for Observation
`
`IPR2016-01020
`Q. So you wouldn’t -- let me ask it this way: Would you know
`intuitively that 88 random carriers and 16 Shively carriers
`would be to the left on the graph of Scenario 2?
`MR. McDOLE: Objection; form.
`THE WITNESS: As I mentioned earlier, the repeating -- the
`repeating -- the replicated bits that Shively introduces has a
`non-Gaussian behavior and it separates from the Gaussian
`decay. So I cannot tell if this is correct.
`Ex. 2013, 111:14-23.
`
`Response to Observation # 7:
`
`TQ Delta’s cited testimony—which compares two Gaussian processes—is
`
`not relevant to Dr. Short’s analysis, which considered a non-Gaussian system
`
`employing Shively’s bit-spreading technique. Dr. Tellado explained that Shively’s
`
`bit-spreading technique is non-Gaussian:
`
`THE WITNESS: As I mentioned earlier, the repeating -- the
`repeating -- the replicated bits that Shively introduces has a
`non-Gaussian behavior and it separates from the Gaussian
`decay….
`Ex. 2013, 111:19-22.
`
`Q. Okay. So did you run a Gaussian process for 88 plus 16?
`A. I ran -- I compared a Gaussian process versus the 88 plus 16
`Shively.
`Q: So you ran a Gaussian process for 104 carriers?
`***
`
`8
`
`
`
`Petitioner’s Response to Patent Owner’s Motion for Observation
`
`IPR2016-01020
`A: 88 [random carriers] plus 16 [Shively carriers] is not equal
`to 104 Gaussian.
`Ex. 2013, 52:25-53:5 & 53:8.
`
`Response to Observation # 8:
`
`TQ Delta mischaracterizes Dr. Tellado’s testimony by ignoring the full
`
`context of the testimony and failing to include portions of the testimony where Dr.
`
`Tellado explains that Shively is not limited to a 1-bit threshold:
`
`Q. Why would you use a 2-bit threshold when Shively says to
`use a 1-bit threshold?
`A. Shively doesn’t say you have to. Shively includes examples
`with 1-bit thresholds.
`Ex. 2013, 71:17-20. TQ Delta also ignores Dr. Tellado’s explanation that he used
`
`a 2-bit threshold for random carriers based on the ADSL T1.413-1995 standard:
`
`A. Now we’re focusing our discussion on the T1 -- ADSL
`T1.413-1995. I believe the minimum constellation size has 2
`bits. That’s why the 2-bit threshold.
`Q. Okay. But you understand that Shively is using BPSK on the
`Shively’s carriers; right?
`A. We’re talking about the random – or pseudorandom QAM
`symbols.
`Q. And you’re referencing
`justification for using --
`A. The 2-bit threshold.
`Ex. 2013, 71:1-11.
`
`
`the T1.413 specification as
`
`9
`
`
`
`Petitioner’s Response to Patent Owner’s Motion for Observation
`
`IPR2016-01020
`
`Response to Observation # 9:
`
`TQ Delta’s observation ignores Dr. Tellado’s repeated testimony that
`
`Stopler randomizes individual QAM symbols:
`
`Q. Is it your opinion that -- in Stopler, that QAM symbols are
`randomized one with respect to the other within a single DMT
`symbol?
`A. Yes.
`Ex. 2013, 117:24-118:2.
`
`Q. Okay. Would it be possible to take every one of those 250
`QAM symbols and rotate them by the same amount?
`A. So my interpretation of Stopler is he is rotating QAM
`symbols individually. There is – the other interpretation does
`not provide benefits to the context in which he is operating.
`Ex. 2013, 121:20-122:1.
`
`Stopler explicitly states that the purpose of the phase scrambler
`is to “randomize the overhead channel symbols” (Ex. 1012,
`12:24),… Since overhead channel symbols are generally QAM
`symbols in a DMT system, the only way to randomize the
`overhead channel symbols is to randomize the QAM symbols.
`Applying phase randomization to DMT symbols, as TQ Delta
`and Dr. Short interpret Stopler, would not break up any
`structure or phase alignment of the overhead channel symbols,
`and thus would not randomize the overhead channel symbols.
`The interpretation by TQ Delta and Dr. Short would not achieve
`Stopler’s stated purpose for the phase scrambler.
`
`10
`
`
`
`Petitioner’s Response to Patent Owner’s Motion for Observation
`
`IPR2016-01020
`
`Ex. 1026, ¶ 57.
`
`Response to Observation # 10:
`
`TQ Delta’s observation confirms Dr. Tellado’s testimony that in the 1990s
`
`phase scrambling for randomization was trivial and well-known. For example, Dr.
`
`Tellado testified that “[a]s I said, the research community in the ’90s was doing
`
`phase-scrambling randomization to reduce PAR to make it better than Gaussian.”
`
`Ex. 2013, 16:12-14. TQ Delta also ignores Dr. Tellado’s testimony regarding the
`
`Muller (Ex. 1024) and Mestdagh (Ex. 1023) references cited in Dr. Tellado’s thesis
`
`(Ex. 1025). Ex. 2013, 149:19-150:2, 146:16-147:8. Dr. Tellado testified that both
`
`papers were examples of how phase scrambling reduced PAR. Ex. 2013, 152:2-3,
`
`148:18-20. For example, Dr. Tellado stated that Muller describes “two algorithms
`
`that use phase rotations to reduce PAR,” while Mestdagh describes how “each
`
`QAM-modulated carrier within a DMT symbol has a phase transformation to
`
`reduce PAR.” Ex. 2013, 152:2-3, 148:18-20.
`
`July 14, 2017
`
`
`
`
`
`
`
`
`
`Respectfully submitted,
`
`/David L. McCombs/
`David L. McCombs
`Counsel for Petitioner
`Registration No. 32,271
`
`
`
`11
`
`
`
`
`
`CERTIFICATE OF SERVICE
`
`The undersigned certifies, in accordance with 37 C.F.R. § 42.205, that
`service was made on the Patent Owner as detailed below.
`
`Date of service
`
`July 14, 2017
`
`
`
`
`
`
`
`Persons served
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Manner of service
`Email: pmcandrews@mcandrews-ip.com;
`twimbiscus@mcandrews-ip.com; smcbride@mcandrews-ip.com;
`cscharff@mcandrews-ip.com; akarp@mcandrews-ip.com;
`TQD-CISCO@mcandrews-ip.com
`
`Documents served
`
`Petitioner’s Response to Patent Owner’s Motion
`for Observation on Cross-Examination
`
`Peter J. McAndrews
`Thomas J. Wimbiscus
`Scott P. McBride
`Christopher M. Scharff
`Andrew B. Karp
`MCANDREWS, HELD & MALLOY, LTD
`500 West Madison Street, 34th Floor
`Chicago, IL 60661
`
`
`
`/David L. McCombs/
`David L. McCombs
`Counsel for Petitioner
`Registration No. 32,271
`
`
`
`12
`
`