[00:00:00] Speaker 04:
Our next case is the reverse UNM Rainforest Innovations versus IZELL Communications, 2023-12-96.

[00:00:08] Speaker 04:
Mr. Key, if you're ready.

[00:00:16] Speaker 00:
Good morning, Your Honors, and may it please support Cecil Key on behalf of the Appellate University of New Mexico, Rainforest Innovations.

[00:00:24] Speaker 00:
I'd like to focus my presentation today on the board's erroneous constructions of the terms compression and quantization, which form the basis for both the board's invalidation of the challenged original claims of the 204 patent and the denial of UNMRI's contingent motion to amend.

[00:00:43] Speaker 00:
Our position is the board fundamentally erred in failing to follow the long established framework for claim construction that this court set forth in Phillips.

[00:00:53] Speaker 00:
It manufactured constructions that are at odds with the specification of the 204 patent.

[00:00:59] Speaker 00:
And even worse, the board used the specification of Zyxel's asserted prior art, Huey, to interpret the 204 patent and then erroneously concluded that to avoid Zyxel's asserted prior art an express disavowal was needed.

[00:01:15] Speaker 00:
Now, the invention at issue here, your honor, involves the operations of quantization and compression.

[00:01:21] Speaker 00:
on what's called the channel state information, or CSI, for the purpose of improving performance of multi-channel communication systems.

[00:01:31] Speaker 00:
It was known in the art that quantization is an operation typically performed on individual pieces of data whereby a finite number of bits are mapped to a data sample.

[00:01:43] Speaker 00:
Compression, on the other hand, is the operation where a set of data

[00:01:48] Speaker 00:
is represented using a reduced amount of bits.

[00:01:52] Speaker 00:
Now the board itself recognized that these differences were known in the art.

[00:01:58] Speaker 00:
It began in its opinion on APPX 30 stating the quantization.

[00:02:04] Speaker 03:
Can you say again what you think quantization

[00:02:08] Speaker 03:
Yes, Your Honor, quantization is an object where you're quantifying X. Can you say what X is and what the activity of quantization does to X?

[00:02:21] Speaker 00:
Yes, X is a data sample or a coefficient

[00:02:28] Speaker 00:
Quantization then is assigning bits, ones and zeros, to be used by the digital system.

[00:02:36] Speaker 00:
So, for example, if you take a continuous analog signal, it's discretized, broken into pieces, you then must assign bit numbers to the pieces.

[00:02:47] Speaker 00:
that are given to be used in the digital.

[00:02:49] Speaker 03:
And what I understand that either the core or at least part of the issue here is whether you have options about how many bits to use so that, you know, quantization of x equals y. You have different numbers of bits that are going to be in y. And if you choose to use fewer rather than more,

[00:03:13] Speaker 03:
that that was viewed by the board as one form of compression.

[00:03:20] Speaker 00:
Yes, Your Honor.

[00:03:21] Speaker 03:
So the board confused... That's right.

[00:03:25] Speaker 03:
I mean, if that's a correct description of what the board concluded, why is that wrong for either the original or the... So first of all, Your Honor, there is a difference between compression, which is trying to take a set of data

[00:03:40] Speaker 00:
and then be able to represent it using a few number of bits versus any other operations such as quantization that just may result in data loss.

[00:03:52] Speaker 03:
So any data reduction does not necessarily equal... I thought there was lossless compression and lossy compression, but they're both compression.

[00:04:03] Speaker 00:
Yes, Your Honor, but they're both compression that are performed on sets of data.

[00:04:08] Speaker 00:
So for example, I think a good way to visualize this, if you look at figure three of the 204 pattern, it shows where there are a lot of these channel taps, and it shows a slope, all the y-intercepts in the slope from left to right.

[00:04:22] Speaker 00:
Essentially, using the least squares method, you can represent all of these various channel taps using two bits, the one on the far left and the one on the far right.

[00:04:32] Speaker 00:
So that is being able to take the set of data, assign bits to it, and a reduced number of bits.

[00:04:41] Speaker 00:
Whereas quantization, really the idea is just to assign bits to it.

[00:04:49] Speaker 00:
Perhaps I think there was argument below about whether or not it might have incidental loss or what have you, but it can also end up an increase.

[00:04:56] Speaker 00:
It is not necessarily a compression technique, especially in the two-fold pattern.

[00:05:02] Speaker 03:
I thought that the board acknowledged that it wasn't necessarily a compression technique, but that sometimes it is, and that's enough.

[00:05:11] Speaker 00:
Yes, Your Honor.

[00:05:12] Speaker 00:
That is exactly what they said.

[00:05:14] Speaker 00:
But our position is it is not that way in the 204 patent.

[00:05:18] Speaker 00:
If they had followed the Phillips framework, you would have been able to see that, if I may.

[00:05:25] Speaker 00:
If you look at the claims of the 204 patent, looking, for example, at APPX 1664, claim one is structured as follows, the original claims.

[00:05:38] Speaker 00:
A sample is selected.

[00:05:41] Speaker 00:
Compression is performed and then the compressed data is sent to the administration.

[00:05:47] Speaker 00:
That's claim one.

[00:05:49] Speaker 00:
In dependent claims nine and ten, quantization can be performed on specific data, the time delays and the phases.

[00:05:59] Speaker 00:
In dependent claims two and seven, quantization can also be performed on the compressed data.

[00:06:06] Speaker 00:
So first of all, I should mention in dependent claims nine and ten, quantization is part of the sending step I mentioned, not part of the compression step.

[00:06:16] Speaker 00:
Throughout those claims, Your Honor, compression and quantization are two separate operations used in two separate ways, on their face indicating

[00:06:26] Speaker 00:
They are not overlapping at all.

[00:06:29] Speaker 00:
Now the board relied heavily on the sites to column four of the 204 pattern, and specifically on lines 11 through 19.

[00:06:40] Speaker 00:
But if you read that column starting a couple of sentences ahead, what you will see is starting around line seven, there is a selection process that's described.

[00:06:51] Speaker 00:
starting around line nine, it describes quantizing those time delays and phases.

[00:06:58] Speaker 00:
Then the next sentence, it says compression of the magnitude and then quantizing of the compressed information.

[00:07:07] Speaker 00:
And then the final sentence says you send the quantized information to the base station.

[00:07:14] Speaker 00:
That is exactly what the structure that's used in the claims

[00:07:19] Speaker 00:
Now, the ward seemed to believe that there is a phrase in that third sentence I mentioned.

[00:07:25] Speaker 00:
It says the base station may further compress.

[00:07:29] Speaker 00:
And the ward reading that together with the prior sentence read that to mean that quantization is a form of compression.

[00:07:38] Speaker 00:
You quantize and then you may further compress, meaning you've already compressed.

[00:07:43] Speaker 00:
It already does assume compression.

[00:07:44] Speaker 00:
And you're going to do it again.

[00:07:48] Speaker 00:
As opposition is, had they truly paid attention to the specification, you would have noticed like on column five, starting around line 59 and carrying forward to column 10, again, it expressly says that there's a compression followed by a separate quantization step.

[00:08:06] Speaker 00:
Again, throughout the intrinsic evidence, compression and quantization are used as two separate operations.

[00:08:14] Speaker 00:
There's no dispute or there was no evidence presented or anything from the prosecution history impacting on this.

[00:08:20] Speaker 00:
So the four corners of the intrinsic evidence consistently says, end the 204 pattern.

[00:08:28] Speaker 00:
Quantization is one process that is separate from compression.

[00:08:32] Speaker 00:
Now, the only way we can see that the board got to saying anything different

[00:08:39] Speaker 00:
is because it was led into using the asserted Huey patent as an interpretation aid.

[00:08:50] Speaker 00:
And one of our best indications there is on APPX 60, which is the board's decision, the board said the specification of the 204 patent contains an express definition of quantization.

[00:09:08] Speaker 00:
There's no citation to anywhere in the record on that sentence.

[00:09:12] Speaker 00:
There's not even a citation back to its own analysis.

[00:09:16] Speaker 00:
The appellee provided no explanation.

[00:09:18] Speaker 00:
The only place we can find in the record where those type of statements are made are in appellee's brief and the statements of its expert, Dr. Roy.

[00:09:31] Speaker 02:
we've been talking at length about the claim construction issue.

[00:09:37] Speaker 02:
To what extent does the way we decide that issue impact the question of the patentability of the existing claims and the proposed amended claims?

[00:09:54] Speaker 02:
Does it take care of it or is there still more work to be done with respect to the Wii reference?

[00:10:01] Speaker 00:
I believe, Your Honor, it takes care of it, but conceivably perhaps there'll be some further remand necessary.

[00:10:09] Speaker 00:
But the reason we believe it takes care of it is this.

[00:10:11] Speaker 00:
The board basically said that quantization equals compression and then referred to Huey, or at least in certain instances, and referred to Huey.

[00:10:22] Speaker 02:
I'm sorry, Huey is the name, is the way you pronounce it?

[00:10:26] Speaker 02:
I want to make sure so we're talking about the same thing.

[00:10:30] Speaker 00:
I use Huey, it may be more appropriately for those with the correct dial.

[00:10:33] Speaker 00:
I like to say Huey.

[00:10:35] Speaker 00:
Okay, well that's fine.

[00:10:36] Speaker 00:
I just want to make sure we're talking about the same thing.

[00:10:37] Speaker 00:
We are talking about the same way the 368 patent, I believe it is.

[00:10:42] Speaker 00:
But the board's decision, both on the original claims and the amended claims, hints on this claim construction.

[00:10:49] Speaker 00:
And we believe that if you correct the kind of construction, and you note that quantization is not compression in the 204 pattern, there are different operations.

[00:10:59] Speaker 00:
Then you look to HUI, and you'll notice that there's no real compression of the type that's disclosed in the 204 described anywhere in HUI.

[00:11:09] Speaker 00:
What HUI is describing, as we put it in our brief, is adaptive vector quantization.

[00:11:15] Speaker 00:
And specifically for HUI,

[00:11:18] Speaker 00:
What he's trying to do, assuming a he, is trying to do is to assign more bits to more significant channel taps, less bits to less significant, in order to improve the resolution.

[00:11:34] Speaker 00:
But in the passage that is cited in page 16 of Appellee's brief, it's a quote from CUE.

[00:11:41] Speaker 00:
It's around column five, line nine to 16 of the

[00:11:45] Speaker 00:
where QSL basically says what they're doing is quantizing individual coefficients, and remember that's what I said, quantization is typically done on individual pieces of data, and this process is to be performed for the total number of available bits

[00:12:03] Speaker 00:
There's no suggestion that there's really any reduction of data at all, even assuming that that was the definition.

[00:12:10] Speaker 00:
But there's nothing that we saw in Huey that describes the type of compression, which is this calculated operation to take the set of data and be able to represent

[00:12:22] Speaker 00:
with a smaller number of bits.

[00:12:25] Speaker 00:
In the data coefficients, you know, mathematical form, you have to have it converted to data, excuse me, to bits, and for it to be operated on in the digital system.

[00:12:35] Speaker 04:
Until you enter the rebuttal time, you may save it or use it.

[00:13:03] Speaker 01:
Good morning, Your Honors.

[00:13:04] Speaker 01:
I'm Jonathan Dietrich for Seisel Communications.

[00:13:06] Speaker 01:
May it please the Court?

[00:13:08] Speaker 01:
The Board's analysis in its final written decision that both the original challenge claims of the 204 patent and the proposed substitute claims are anticipated and rendered obvious by the Hui reference is well reasoned and supported by substantial evidence.

[00:13:24] Speaker 01:
The dispute between the parties centers on whether compression, specifically compressing estimated channel state information,

[00:13:31] Speaker 01:
excludes compression performed using what they need called quantization.

[00:13:35] Speaker 01:
So I'll start there.

[00:13:37] Speaker 01:
Like the 204 patent, we pertain to a multiple input, multiple output antenna communication system, and we recognize that as the number of antenna pairs increase, the amount of channel state feedback information that needs to be transmitted increases drastically.

[00:13:56] Speaker 01:
And so to reduce the amount of that information that needs to be transmitted,

[00:14:00] Speaker 01:
we disclose as compressing it before transmission using a technique called quantization.

[00:14:06] Speaker 01:
Here, UNM contends that this method of compressing the estimated CSI isn't within the scope of that term under the 204 pattern.

[00:14:15] Speaker 01:
But that contention is flatly contradicted by the specification of the 204 pattern, which provides an example in which quantizing portions of the estimated CSI is described as compressive.

[00:14:27] Speaker 01:
So let's turn our attention to that.

[00:14:30] Speaker 03:
That's a column for a material.

[00:14:31] Speaker 01:
Exactly.

[00:14:33] Speaker 01:
But before I read that, I'd like to first explain that the tool of four patent describes that a channel response includes channel taps with each channel tap corresponding to a time delay and having a complex value that includes a magnitude and a phase.

[00:14:54] Speaker 01:
So the channel state information

[00:14:56] Speaker 01:
refers to information regarding those three values.

[00:15:00] Speaker 01:
The time delays it, the time delays the magnitudes and the phases of the channel tabs.

[00:15:08] Speaker 01:
And so with that context, we can understand the example provided in column four, which quantizes two of those values, the time delays and phases, and describes that as compressing the estimated CSA.

[00:15:23] Speaker 01:
So I'll briefly just read that portion.

[00:15:27] Speaker 01:
which states, beginning at line 11, the mobile station 104 may quantize information regarding time delays and phases of the selected channel tabs.

[00:15:38] Speaker 01:
The mobile station 104 may further compress the estimated CSI, and then it goes on.

[00:15:45] Speaker 01:
You're saying the word further means there's already been some compression.

[00:15:49] Speaker 01:
Exactly.

[00:15:49] Speaker 01:
If the word further wasn't there, then maybe there would be a little more difficult issue to decide.

[00:15:56] Speaker 01:
But with the word further there, as the board found, the disclosure that the mobile station 104 may further compress the estimated CSA

[00:16:08] Speaker 01:
must mean that the preceding action performed by the mobile, spatial 104, which was quantizing information regarding the time delays and phases of the selected channel taps, two of the components of the CSI, that that is an action that was regarded as compressive, and therefore is an action that compresses at least a portion of the estimated CSI.

[00:16:35] Speaker 01:
So the board was correct to find that the intrinsic evidence did not support UNM's argument that compression achieved using quantization should be excluded from the scope of the compressing terms.

[00:16:47] Speaker 03:
If that sentence had been written so that the word further was moved to the

[00:16:54] Speaker 03:
front of the sentence followed by a comment and it said further the mobile station may compress the estimated CSI

[00:17:05] Speaker 03:
would that, that would take the, I don't know, the helpful character of the word further out of the equation.

[00:17:16] Speaker 03:
It could mean also just, you know, where would you be if that sentence were rewritten that way?

[00:17:26] Speaker 01:
I mean, as it's a rewriting.

[00:17:27] Speaker 01:
Right.

[00:17:27] Speaker 01:
I agree that would change the meaning of that sentence.

[00:17:30] Speaker 01:
I don't think that,

[00:17:35] Speaker 01:
it would change the result here.

[00:17:37] Speaker 01:
The plain and ordinary meaning of compression would still include quantization.

[00:17:46] Speaker 01:
And I was just about to get to the extrinsic evidence where UNN's expert Dr. Voydridge admitted that quantization results in a reduction of data bits, which is consistent with the plain and ordinary meaning of compressing.

[00:18:04] Speaker 01:
Just as in the purpose in the prior art reference, we, quantization is performed to reduce the number of data bits that need to be transmitted.

[00:18:14] Speaker 01:
That's compression under any ordinary meaning of the term, even if we didn't have that example in column four of the specification.

[00:18:28] Speaker 01:
I'd also like to address

[00:18:31] Speaker 01:
council's description that the board's decision aired by equating quantization and compression.

[00:18:37] Speaker 01:
That's not the case.

[00:18:38] Speaker 01:
That's an argument that UNM had made down below and the final written decision took care to point out that it was not equating quantization with compression.

[00:18:48] Speaker 01:
Rather, quantization, you know, where did the board say that quantization is equivalent or the only way to compress?

[00:18:57] Speaker 01:
Rather, the board found, quote,

[00:19:00] Speaker 01:
Compression is a broad and basic term that encompasses any form of compression, including quantization, so long as the quantization results in a reduction in data size.

[00:19:11] Speaker 01:
That's page 30 of the joint appendix.

[00:19:16] Speaker 01:
So with that, with the current construction in which the plain and ordinary meaning of compression does not exclude quantization, here I don't think there's any dispute

[00:19:30] Speaker 01:
that we anticipate both the original challenge claims as well as the proposed substitute claims.

[00:19:39] Speaker 01:
And so if this court were continuing to reject that argument, that compression achieved via quantization should be excluded from the scope of compression, I don't think there's any dispute.

[00:19:53] Speaker 01:
that we discloses the compressing terms at issue, and that the board's findings are supported by substantial evidence.

[00:20:00] Speaker 02:
You know, I'm saying if we adopt your claim construction, then the board did.

[00:20:06] Speaker 02:
Then it would resolve both the original claims, the three original claims, and the proposed amended claims that were rejected.

[00:20:15] Speaker 01:
That's right, Your Honor.

[00:20:15] Speaker 01:
I didn't see any special argument with respect to the proposed substitute claims apart from its claim construction argument.

[00:20:23] Speaker 01:
But I would further submit that all it's required is just to affirm that the plain and ordinary meaning of compression applies and that there's not a special card out to exclude quantization.

[00:20:40] Speaker 01:
But just going to the substantial evidence supporting anticipation, the only disputed limitation in the original claims was the compressing the estimated CSI limitation.

[00:20:54] Speaker 01:
And so the board found that WE's mobile station compresses the estimated CSI by adaptively quantizing channel technical coefficients.

[00:21:03] Speaker 01:
And we see that in WE at column three, lines 54 through 57, which describes using vector quantization techniques to reduce the channel state feedback.

[00:21:14] Speaker 01:
And regarding the substitute claims,

[00:21:18] Speaker 01:
Substantial evidence also supports the board's determination that we discloses the limitations of those.

[00:21:27] Speaker 03:
Can you just translate one thing for me?

[00:21:29] Speaker 03:
Is it reducing the channel state feedback?

[00:21:32] Speaker 03:
Yes.

[00:21:32] Speaker 03:
Why is that compression?

[00:21:36] Speaker 03:
So maybe feedback means noise to me, which I think it doesn't mean that in this document.

[00:21:45] Speaker 03:
So that's the explanation I'm looking for.

[00:21:47] Speaker 01:
Thank you, Your Honor.

[00:21:48] Speaker 01:
Yes.

[00:21:49] Speaker 01:
So this goes back to – well, to answer your question directly first, feedback refers to the number of bids in the channel state information feedback signal that needs to be transmitted.

[00:22:02] Speaker 01:
And this goes back to

[00:22:04] Speaker 01:
what we recognized as the problem to be solved, which is the identical problem the 204 pattern was trying to be solved, which is that in a MIMO system, as the number of pairs of antennas increase, channel state feedback information for each pair needs to be transmitted.

[00:22:20] Speaker 01:
So as those pairs increase, the required channel state information that needs to be sent back increases drastically.

[00:22:27] Speaker 01:
So HUE's objective was to reduce the number of data bits in that channel state information feedback signal.

[00:22:33] Speaker 01:
Okay.

[00:22:41] Speaker 01:
Turning back to the proposed substitute claims, the only limitation that UNM challenges was or recites the compressing further comprises generating a plurality of parameters to represent information regarding magnitudes of the select channel tabs and quantizing the plurality of parameters.

[00:23:03] Speaker 01:
And as the board determined, we discloses just that, specifically

[00:23:08] Speaker 01:
Wee discloses generating a plurality of complex-valued estimated channel cap coefficients, which necessarily have a magnitude and a phase, and then quantizing those coefficients.

[00:23:21] Speaker 01:
And you see that in column two lines five and six, columns four lines 31 and 38, and columns six lines two and three of Wee.

[00:23:31] Speaker 01:
For example, in that last site, column six, lines two through three, we states, let A sub MK capital R, meaning real, and A sub MK capital I, meaning the imaginary part, denote the real and imaginary parts of the estimated channel tabs, A, MK, prospectively.

[00:23:55] Speaker 01:
And although those are real imaginary, Dr. Roy explained that those representations of complex numbers is equivalent to a magnitude and phase representation through a simple mathematical transformation.

[00:24:10] Speaker 01:
Therefore, the plurality of those coefficients, as claimed, do represent information regarding the magnitude

[00:24:20] Speaker 01:
the selected channel tabs and then we discloses quantizing those coefficients so The board's determination that we discloses the that limitation is supported by substantial evidence Do you have any other questions for me I'll see the remainder of my time.

[00:24:45] Speaker 04:
Thank you counsel Mr. Key

[00:24:55] Speaker 00:
Just a few very quick points.

[00:24:58] Speaker 00:
Follow up on my colleague's response to just Toronto's question about moving further to the beginning of the sentence.

[00:25:09] Speaker 00:
Mr. Dietrich said that that would change the meaning of the sentence.

[00:25:12] Speaker 00:
I'm not sure if that is true, but it would not change the meaning of the intrinsic evidence, which must be read as a whole.

[00:25:19] Speaker 00:
And read as a whole, again, looking at the claims and claim construction, the dependent claims, et cetera, as we've gone through, it's clear that these are two separate operations.

[00:25:30] Speaker 00:
We submit that that's exactly what was meant by that sentence.

[00:25:34] Speaker 00:
And in fact, it meant further processing after quantizing than by compressing, and that's consistent with

[00:25:41] Speaker 00:
column 5, lines 59 through column 6, line 4.

[00:25:49] Speaker 00:
I also, if I understood him correctly, I heard my colleague to say that we should be considering extrinsic evidence.

[00:25:57] Speaker 00:
Well, first of all, no, we should not because the intrinsic evidence is very clear here.

[00:26:02] Speaker 00:
There's no reason to go to extrinsic evidence.

[00:26:04] Speaker 00:
But if I understood him correctly,

[00:26:07] Speaker 00:
He mentioned some purported admission by Dr. Vojcic and then something about what the Huey patent shows.

[00:26:16] Speaker 00:
That is exactly the problem.

[00:26:18] Speaker 00:
Huey, even if you consider extrinsic evidence, is not the type of objective evidence that's to be considered as extrinsic evidence under Phillips.

[00:26:27] Speaker 00:
And also even the expert testimony is less critical.

[00:26:33] Speaker 00:
The most important intrinsic evidence here is some of the scientific papers that were presented that the board looked at and said they agreed based on that evidence that quantization is not compression.

[00:26:49] Speaker 00:
The argument that there is no special carve out in the patent for saying compression is whatever without quantization, that's not required.

[00:27:02] Speaker 00:
When you follow Phillips, the whole idea is it will tell you what the boundaries and the limitations of the terms are.

[00:27:09] Speaker 00:
as we walk through, Phillips here points out very clearly that following the Phillips framework, that quantization and compression are two separate terms.

[00:27:20] Speaker 00:
And I'll point to one other thing I'll note.

[00:27:23] Speaker 00:
Again, this issue of the board's statement without citation, that the specification has an expressed definition of quantization, and remember I mentioned

[00:27:36] Speaker 00:
But there's no citation, what have you.

[00:27:39] Speaker 00:
But I will note that if you look at appendix APP 1663 to 64, that's paragraph 59, where Dr. Roy says the following.

[00:27:52] Speaker 00:
Compression is accomplished by quantizing the parameters of the channel responses that represent the magnitudes, phases, and time delays of the channel taps, citing the exact same column four, lines 11 through 19.

[00:28:06] Speaker 00:
It does not say that.

[00:28:08] Speaker 00:
It does not say you quantize it.

[00:28:10] Speaker 04:
If we were to quantize your time for argument, we would say it is over.

[00:28:15] Speaker 00:
All right.

[00:28:15] Speaker 00:
Very good, Your Honor.

[00:28:16] Speaker 04:
Thank you to both counsel.

[00:28:17] Speaker 04:
The case is submitted.