[00:00:02] Speaker 03:
I'm going to switch here and see.

[00:00:51] Speaker 03:
Our third case this morning is number 17-1072 CF Crespi LLC versus Silicon Laboratories Inc.

[00:00:59] Speaker 03:
Again, Mr. Smith.

[00:01:09] Speaker 00:
Thank you, Your Honor.

[00:01:10] Speaker 00:
May it please the Court?

[00:01:10] Speaker 00:
My name is Craig Smith.

[00:01:11] Speaker 00:
I represent the appellant CF Crespi.

[00:01:14] Speaker 00:
In this case, the Court should reverse the Board's finding of

[00:01:18] Speaker 00:
obviousness because it's not supported by substantial evidence.

[00:01:22] Speaker 00:
The board was supposed to look at what the petitioner set forth in the petition and decide whether or not those arguments could be met based on the burden of proof set forth.

[00:01:34] Speaker 00:
But the board didn't address some of the dispositive issues that both of the parties considered to be relevant and instead made determinations that were separate and apart from what the arguments were made in the petition.

[00:01:47] Speaker 00:
As a result,

[00:01:48] Speaker 00:
You had the board making determinations that neither party seemed to have been pushing for or arguing over.

[00:01:56] Speaker 00:
And even on this appeal, we still have this strange dichotomy where in response to our appeal, Silicon Laboratories is arguing for its original position that was set forth in its petition, as opposed to the argument that the board set forth in its final written decision.

[00:02:16] Speaker 00:
We believe as a result, the board's decision should be reversed because it's not supported by substantial evidence.

[00:02:22] Speaker 00:
And we think this case might be similar in kind to this court's decision in in ray magnum tools, where the court cautioned the board not to be coming up with new decisions or new theories and arguments on its own that were not set forth in the petition.

[00:02:41] Speaker 00:
Claims 11 and 12 of the 585 patent are

[00:02:44] Speaker 00:
Not obvious.

[00:02:45] Speaker 00:
The board's findings with regard to claims 11 and 12 are not supported by substantial evidence.

[00:02:50] Speaker 00:
Here, none of the references that were set forth disclose storing a plurality of fur filters in memory.

[00:02:57] Speaker 00:
Silicon Labs tried to fill this gap by using a reference called Grumman to show that this element existed.

[00:03:05] Speaker 00:
But what the parties were arguing over is whether Grumman actually discloses

[00:03:10] Speaker 00:
one fur filter or multiple fur filters, and that was a key issue that was going back and forth.

[00:03:18] Speaker 00:
We submitted that Grumman doesn't disclose more than one fur filter, and it's just a single filter, not multiple fur filters.

[00:03:25] Speaker 00:
Grumman has a single fur filter having real and imaginary components, and those coefficients are part of just one filter, not multiple filters.

[00:03:36] Speaker 02:
Now...

[00:03:39] Speaker 02:
The Grumman filter, it adapts, it can switch rapidly between the two different filter functions.

[00:03:47] Speaker 00:
It is adaptive.

[00:03:48] Speaker 00:
I mean, it can switch, but it is a single filter and Grumman describes itself as a filter, not as multiple filters.

[00:03:55] Speaker 02:
But in terms of functionality, it's got multiple functions.

[00:04:00] Speaker 00:
It has multiple functions and many filters do have that ability where it can have multiple things that the filter performs.

[00:04:08] Speaker 00:
And Silicon Lab has argued that we shouldn't look at what Grumman actually says it is, which it describes itself as a single filter and not as multiple filters.

[00:04:18] Speaker 00:
But I think a person of ordinary skill in the art looking at what Grumman actually disclosed would want to know that Grumman is describing itself as a single fur filter, not as multiple fur filters.

[00:04:29] Speaker 00:
It performs the functions of multiple filters, right?

[00:04:32] Speaker 00:
No, Your Honor.

[00:04:33] Speaker 00:
It performs a function of a single filter where it allows it to adapt and switch, but it does not perform multiple filters.

[00:04:40] Speaker 03:
Didn't the board find that having a set of filter coefficients in Grumman was the equivalent of having multiple filters?

[00:04:50] Speaker 00:
No, Your Honor.

[00:04:50] Speaker 00:
The board sort of sidestepped this whole issue, even though the parties felt like it was a dispositive issue with regard to these claims.

[00:04:58] Speaker 00:
The board looked at our argument where we had

[00:05:01] Speaker 00:
presented that the evidence of Grumman and exactly what Grumman discloses is a single filter and said, well, that's not really the point here.

[00:05:10] Speaker 00:
The point is that there are other references that disclose filters.

[00:05:14] Speaker 00:
And so the board actually picked up on another reference that the petitioner hadn't argued was the multiple filters in this case.

[00:05:24] Speaker 00:
So for example, the petitioner argued, and this is at appendix 103 through 105,

[00:05:28] Speaker 00:
that Grumman disclosed the multiple fur filters and explicitly states, it says, because Grumman discloses storing coefficients for two fur filters in memory, this limitation is met by Grumman.

[00:05:41] Speaker 00:
So he specifically said that we're relying on Grumman for this.

[00:05:44] Speaker 00:
But the board didn't rely on Grumman and sort of pushed that away and said, no, we're going to rely on Harris for the filters.

[00:05:51] Speaker 00:
But that's not the argument that was set forth in the petition.

[00:05:54] Speaker 00:
Even though the board later seemingly inconsistently admits that

[00:05:58] Speaker 00:
the grumman is relied on for the fur filters as if that is also a argument that had been made.

[00:06:06] Speaker 00:
And so we have a situation where the board seems to be taking a position that the petitioner hasn't set forth.

[00:06:14] Speaker 00:
And even on appeal, we're hearing that the, the arguments that are made in support of the board's decision are arguments that go back to the petition.

[00:06:25] Speaker 00:
where they're arguing about whether Grumman discloses a single or multiple for filters as opposed to what the board actually did.

[00:06:32] Speaker 00:
In our view, the board shouldn't have created new grounds for purposes of finding the claims unpatentable.

[00:06:41] Speaker 00:
It should have looked at what was actually set forth in the petition and what was argued by the petitioner and under that analysis should have found that these claims were unpatentable.

[00:06:51] Speaker 00:
In light of this court's prior holdings, we think the board improperly took its own theories based on the prior art and substituted them for the arguments that the petitioner was actually making.

[00:07:03] Speaker 00:
This also appears in a second place in the board's decision, and this relates to the same claim, but a slightly different part of the limitation.

[00:07:13] Speaker 00:
Grumman is also used by the petitioner to show this element that says that the signal processor is going to index memory to retrieve one of a plurality of fur filters.

[00:07:25] Speaker 00:
And Grumman doesn't disclose retrieving a plurality of fur filters.

[00:07:30] Speaker 00:
And Grumman doesn't disclose a single processor for doing this.

[00:07:34] Speaker 00:
It discloses a multiple set of processors for processing the information.

[00:07:39] Speaker 00:
Our expert explained that this parallel processing concept

[00:07:44] Speaker 00:
is not a single processor, but multiple processors.

[00:07:47] Speaker 00:
But the board didn't rely on what was set forth in the petition.

[00:07:51] Speaker 00:
It relied on a different argument.

[00:07:54] Speaker 00:
So the petition relied on Grumman for the processor, not for these other references that are referred to as Thompson and Harris.

[00:08:01] Speaker 00:
But the board then looked and said, well, we're going to find that Thompson combined with Harris has this particular processor.

[00:08:08] Speaker 00:
But again,

[00:08:09] Speaker 00:
Similar to the argument I've just mentioned, the board is now pulling in its own theories and not using the theories that were set forth in the petition.

[00:08:17] Speaker 00:
We think this is inappropriate because the theories that the board is supposed to be analyzing to see whether or not the petitioner has met its burden of proof is not the arguments that the board is deciding.

[00:08:30] Speaker 00:
The board is deciding on new theories that it is coming up with independent of what the parties have set forth.

[00:08:37] Speaker 00:
As a result, we think that the board's findings with regard to claims 11 and 12 should be reversed because the board was relying on evidence and arguments that the board was coming up with as opposed to what was set forth in the petition.

[00:08:51] Speaker 00:
If the board had looked at the petition and the arguments there, it clearly would have found that there was no support for the arguments that were being made.

[00:08:59] Speaker 00:
Similarly, claims 13 and 14 of the 585 patent are not obvious.

[00:09:05] Speaker 00:
Client 13 requires a standard selection circuit.

[00:09:09] Speaker 00:
And this generates a select signal indicative of a format of said input RF signal.

[00:09:15] Speaker 00:
And then the signal processor is going to select a FIR filter in response to the select signal.

[00:09:21] Speaker 00:
Now neither Thompson nor Harris disclosed a standard selection circuit.

[00:09:26] Speaker 00:
The board previously made that determination in a separate petition that was filed in these series of actions.

[00:09:35] Speaker 00:
And our expert explained to the board that there's no need for a standard selection circuit in Thompson or Harris, because their format, they're processing this information in parallel.

[00:09:46] Speaker 00:
So in this instance, what Silicon Labs did is it tried to bring in Zenith, which is another prior art reference.

[00:09:53] Speaker 00:
And Zenith doesn't cure the problem for them, because Zenith does not teach the selection of a fur filter in response to a select signal.

[00:10:01] Speaker 00:
It teaches between selecting between two demodulators based on the presence or absence of a zinc signal.

[00:10:07] Speaker 00:
The claim requires more than this, and it requires not just determining the format of a TV signal, as they suggest, but it requires that it must select a fur filter in response to the select signal.

[00:10:20] Speaker 00:
And this is something Zenith simply does not do.

[00:10:22] Speaker 00:
As a result, I believe the board erred in coming to its determination here by looking at the references and

[00:10:31] Speaker 00:
filling in a gap that Zenith does not fill for the petitioner in this case.

[00:10:40] Speaker 00:
Finally, claims 15 and 20 are not obvious.

[00:10:43] Speaker 00:
The board's decision is also based on its interpretation of the Zenith reference, but Zenith fails to disclose detecting a carrier signal as required by the claims.

[00:10:55] Speaker 00:
Zenith discloses a sync signal that merely detects the presence or absence

[00:11:00] Speaker 00:
of an analog signal, it doesn't disclose the detection of carrier signals such as described in the 585 pattern.

[00:11:07] Speaker 00:
The 585 pattern focuses on the idea that you have to be able to detect, are you talking about an analog signal or a digital signal?

[00:11:15] Speaker 00:
Zenith talks about you just can detect whether it's an analog signal.

[00:11:19] Speaker 00:
Well, if it's not an analog, it has to be something else, right?

[00:11:21] Speaker 00:
Correct, Your Honor.

[00:11:22] Speaker 00:
It could be something else, but it doesn't have to be just one other thing.

[00:11:26] Speaker 00:
Zenith says, well, if it's not analog, it's digital.

[00:11:29] Speaker 00:
and says, well, you know that it's digital because of the absence.

[00:11:34] Speaker 00:
The 585 pattern talks about the idea that the reason we're talking about a plurality of potential carrier signals is because it talks about not just analog and digital, but there may be other signals that it would need to detect.

[00:11:46] Speaker 00:
And you need those carrier signals in order to be able to know what it is.

[00:11:50] Speaker 00:
And so the mere absence of a signal isn't a signal.

[00:11:54] Speaker 00:
And as a result,

[00:11:55] Speaker 00:
CNIT doesn't disclose the idea that it is detecting signals that are specific to analog, to digital, or to any other type of signal that could be in the stream.

[00:12:05] Speaker 00:
As a result, we believe the board's decision is not supported by substantial evidence.

[00:12:11] Speaker 03:
Thank you.

[00:12:11] Speaker 03:
Let me just sit down.

[00:12:14] Speaker 03:
I'm looking at page 28 of the board's decision.

[00:12:18] Speaker 03:
And the paragraph at the top of the page where we describe the petitioner's argument

[00:12:24] Speaker 03:
was well known to implement FIR filters by storing sets of FIR filter coefficients in memory, et cetera, et cetera.

[00:12:32] Speaker 03:
And then they say, we agree with petitioner.

[00:12:35] Speaker 00:
Isn't that adopting the petitioner's argument?

[00:12:38] Speaker 00:
I don't think it's adopting all of the petitioner's argument in whole, because the board later goes on in the next page and says, we disagree with the patent owner's framing of the dispositive issue as whether Grumman discloses

[00:12:51] Speaker 00:
a single two-part fur filter or two fur filters.

[00:12:55] Speaker 00:
So it basically says, we don't really think that's the issue here.

[00:12:59] Speaker 00:
We think we're going to look at the issue from the context of another reference.

[00:13:03] Speaker 00:
And so it brought in another reference and said, we think Harris is what discloses the fur filters.

[00:13:10] Speaker 00:
But if you go to the petition, the petition never says that Harris is the one that's bringing in the filters.

[00:13:16] Speaker 00:
It explicitly relies on Grumman for the filters.

[00:13:21] Speaker 00:
OK.

[00:13:22] Speaker 00:
Thank you, Your Honor.

[00:13:23] Speaker 00:
Thank you.

[00:13:24] Speaker 03:
Mr. Ayers?

[00:13:28] Speaker 03:
Yes, Your Honor.

[00:13:51] Speaker 01:
Peter Ayers on behalf of the Appalee Silicon Labs, and may it please the court.

[00:13:57] Speaker 01:
In this case, Your Honor, the board's decision was amply supported by substantial evidence.

[00:14:03] Speaker 01:
The board did not come up with a new theory on its own.

[00:14:06] Speaker 01:
It, as Judge Dyke pointed out, accepted the petitioner's theory that the combination of Thompson, Harris, and Grumman renders claim 11

[00:14:21] Speaker 01:
and 12 obvious.

[00:14:23] Speaker 01:
Moreover, this distinction of one filter versus two is really semantic and doesn't really join the issues.

[00:14:33] Speaker 01:
And that's what the board found, because it essentially ignores the claim language.

[00:14:38] Speaker 01:
And if I could turn to the claim language, because I think it's really dispositive in this case of this single versus multiple filters.

[00:14:47] Speaker 02:
Let me ask you a question on something that's kind of been baffling me.

[00:14:51] Speaker 02:
Aren't you barred from your arguments with respect to claims 13 and 14, because these were denied already, institutions denied by the board?

[00:15:03] Speaker 01:
No, Your Honor.

[00:15:04] Speaker 01:
The board did not make any kind of finding with respect to whether or not the selection circuit was obvious in light of Zenith.

[00:15:14] Speaker 01:
All it did was said that

[00:15:17] Speaker 01:
the combination of Thompson and Harris didn't inherently disclose claim 13.

[00:15:22] Speaker 01:
And in that case, the theory that we put forward to the board was that Thompson, as you may know, there's an adaptive filter in there, and it has an input that says TV standard.

[00:15:36] Speaker 01:
And the Thompson receiver is capable of processing both analog and digital formats.

[00:15:42] Speaker 01:
So our theory was, well,

[00:15:46] Speaker 01:
It has an input there that the filter is responsive to.

[00:15:50] Speaker 01:
And so a person of ordinary skill in the art would know that there's a circuit that would have to generate that.

[00:15:55] Speaker 01:
The board didn't accept that.

[00:15:56] Speaker 01:
And so it exercised its discretion in the second IPR and took up our theory that Thompson and Harris, in combination with Zenith, found that combination obvious.

[00:16:10] Speaker 01:
There was no finding a fact.

[00:16:12] Speaker 01:
We were never given the opportunity to appeal that institution

[00:16:16] Speaker 01:
decision, and so there's no collateral estoppel.

[00:16:18] Speaker 02:
Is there a reason to raise the zenith reference in the prior IPR?

[00:16:26] Speaker 01:
That is potentially true, your honor.

[00:16:29] Speaker 01:
However, the issue is whether or not the board could exercise its discretion under 325, and it said, yes, we can and we will.

[00:16:39] Speaker 01:
So while the board may have been permitted to decline to

[00:16:45] Speaker 01:
entertain our theory.

[00:16:47] Speaker 01:
Now that it has, it's properly before this court, and we're not barred from raising it here.

[00:16:53] Speaker 01:
And does that answer your question?

[00:16:57] Speaker 01:
And let me just get back to the claims, because I think this really drives home the point about this filter and why the board didn't really accept or really engage in the patent owner's argument about one filter versus two, because

[00:17:13] Speaker 01:
If we look at claim one itself, it says a channel filter.

[00:17:17] Speaker 01:
That's one of the main elements.

[00:17:19] Speaker 01:
And what does that filter have?

[00:17:20] Speaker 01:
It has another filter in it, an anti-aliasing filter, and then a signal processor that itself performs multiple filter functions.

[00:17:29] Speaker 01:
So the fact that something's referred to as a filter is not this positive of whether or not that filter can perform multiple filter functions.

[00:17:38] Speaker 01:
And that's what we have with Grumman.

[00:17:40] Speaker 01:
We have an adaptive filter.

[00:17:43] Speaker 01:
and a filter that can perform multiple for filter functions.

[00:17:48] Speaker 01:
And that's what the board found.

[00:17:50] Speaker 01:
And that's why that's really the dispositive issue, not whether or not Grumman describes itself as an adaptive filter or refers to the filter.

[00:17:59] Speaker 01:
We're not saying that the teaching of Grumman is irrelevant, but you have to ground yourself in the claim language.

[00:18:06] Speaker 01:
And in this case, it's clear that the filter is a functional limitation.

[00:18:12] Speaker 01:
You start with claim 10, for example.

[00:18:13] Speaker 01:
It says, wherein said signal processor applies one of a plurality of finite impulse response filters.

[00:18:22] Speaker 01:
It's performing a filtering function.

[00:18:24] Speaker 01:
The patent owner itself concedes that a filter is a functional limitation in this case, if you look at their reply brief.

[00:18:33] Speaker 01:
And then what does claim 11 require?

[00:18:36] Speaker 01:
All it requires

[00:18:37] Speaker 01:
is that these filters be stored in a memory.

[00:18:40] Speaker 01:
And we know from the specification that the only thing that is stored in a memory are the coefficients.

[00:18:46] Speaker 01:
And so we said, Grumman describes an adaptive filter that has two memory banks.

[00:18:53] Speaker 01:
And we refer the core to our expert's block diagram, where he lays this out very clearly, as we've shown in our brief, that there are two independent sets of filter functions

[00:19:07] Speaker 01:
that each perform a separate fur filter, and that that combination satisfies the limitation of claim 11.

[00:19:21] Speaker 01:
With respect to this question about a single processor in Grumman, I think it's telling that they're not challenging, in this case, claim 12 that was also invalidated by the board.

[00:19:37] Speaker 01:
That would expose the inconsistencies in their theory, because claim 12 itself says that the signal processor comprises a first computing unit and a second computing unit, showing that the signal processor doesn't have to be a single component.

[00:19:55] Speaker 01:
And that's clear from the specification.

[00:19:58] Speaker 01:
Now, with respect to claims 13 and 14,

[00:20:07] Speaker 01:
The question about the standard, really let me turn to claim 15, which depends on claim 13, the notion of detecting carrier signals.

[00:20:21] Speaker 01:
Again, this argument about that the standard selection circuit has to detect two different types of sync signals is not only inconsistent with

[00:20:35] Speaker 01:
the claim language, but also with their own specification.

[00:20:39] Speaker 01:
The claim language, if you look at claim 15, it says that the standard selection circuit generates said select signal by detecting carrier signals, identifying one of said formats of said input RF signal.

[00:20:58] Speaker 01:
And the reason why it says that is because the specification makes clear that it

[00:21:04] Speaker 01:
It can detect carrier signals in one case, and it can infer the absence of carrier signals in another.

[00:21:13] Speaker 01:
That's the so-called auto detection embodiment that they have.

[00:21:18] Speaker 01:
So the fact that Zenith admittedly doesn't detect two different types of carrier signals, again, does not distinguish Zenith from claims 15 and claim 20.

[00:21:34] Speaker 01:
it's entirely consistent with the specifications teaching that this standard selection circuit can either detect the presence of a carrier signal or infer the format from the absence of a carrier signal.

[00:21:57] Speaker 01:
So I believe I've addressed all of the arguments here today, but I'm happy to answer any further questions

[00:22:04] Speaker 01:
Thank you.

[00:22:08] Speaker 00:
I'd just like to make two quick points on rebuttal.

[00:22:24] Speaker 00:
The first is, it's a little surprising to hear that the issue of the

[00:22:29] Speaker 00:
one versus multiple filters was an issue, since if you look through all the briefing, both below and even on appeal, that seems to be the dispositive issue that the parties were arguing as to whether or not it's accurate to be multiple filters.

[00:22:43] Speaker 00:
Now, the board sort of sidestepped that issue in our view, but that was really the crux of the issue.

[00:22:48] Speaker 00:
And the claims that we're looking at refer to a plurality of fur filters.

[00:22:53] Speaker 00:
So this isn't something we are

[00:22:55] Speaker 00:
asking the court or the board to suggest that there's something to be added to this.

[00:23:00] Speaker 00:
No, the claims specifically require a plurality of fur filters.

[00:23:03] Speaker 00:
And Grumman is explicit.

[00:23:05] Speaker 00:
It says it is disclosing a complex weight fur filter.

[00:23:09] Speaker 00:
This is at 235.

[00:23:10] Speaker 00:
And then at 236, it goes on and says, we're talking about the two halves of the filter.

[00:23:17] Speaker 00:
So when it talks about the real and imaginary components that we were talking about earlier, it says those are two halves of one filter.

[00:23:25] Speaker 00:
This specifically says that that's what it's discussing when it talks about the separation between the real and imaginary portions of it.

[00:23:32] Speaker 00:
Plainly on the signal processing issue, the 585 patent actually describes different ways of doing the processing.

[00:23:39] Speaker 00:
It can do it with one unit or it can do it with multiple units.

[00:23:42] Speaker 00:
The fact that the claims are written so that they can capture either doing it with one signal processor or multiple signal processors is not surprising since the specification specifically

[00:23:53] Speaker 00:
contemplates that difference.

[00:23:55] Speaker 00:
Thank you, Your Honors.