PTAB
IPR2017-00700
Apple Inc. v. California Institute of Technology
1. Case Identification
- Case #: IPR2017-00700
- Patent #: 7,421,032
- Filed: January 20, 2017
- Petitioner(s): Apple Inc. (Real Parties-in-Interest: Apple Inc. and Broadcom Corp.)
- Patent Owner(s): California Institute of Technology
- Challenged Claims: 11-17
2. Patent Overview
- Title: Serial Concatenation of Interleaved Convolutional Codes Forming Turbo-Like Codes
- Brief Description: The ’032 patent discloses error-correcting codes, specifically irregular repeat-accumulate (IRA) codes. The invention describes an encoder comprising an outer coder that irregularly repeats message bits, an interleaver that scrambles the repeated bits, and an inner coder (an accumulator) that processes the scrambled bits to generate parity bits.
3. Grounds for Unpatentability
Ground 1: Obviousness of Claims 11, 12, and 14-16 over Ping in view of MacKay and further in view of Divsalar
- Prior Art Relied Upon: Ping (a 1999 publication on LDPC codes), MacKay (a 1999 publication on irregular codes), and Divsalar (a 1998 publication on repeat-accumulate codes).
- Core Argument for this Ground:
- Prior Art Mapping: Petitioner argued that the combination of these references teaches all elements of the challenged claims. Ping disclosed a two-stage error-correcting code with a regular outer low-density parity-check (LDPC) coder and an inner accumulator, which satisfies the fundamental structure of the claims. MacKay taught that making such codes irregular (by using non-uniform column weights in the parity-check matrix) improves performance. Divsalar taught the use of a simple repeater as an outer coder, which Petitioner contended is a natural and obvious way to implement Ping's outer coder function. For independent claim 11, the combination created the claimed "irregular repeat accumulate" (IRA) structure. For dependent claim 12, MacKay’s further teaching of non-uniform row weights satisfied the limitation of a non-constant number of inputs to the check nodes. For dependent claims 14-16, Ping and Divsalar collectively taught the specific properties of the accumulator, such as it being a "recursive convolutional coder" with a transfer function of "1/(1+D)".
- Motivation to Combine: A Person of Ordinary Skill in the Art (POSITA) would combine Ping and MacKay because they address the same field (LDPC codes), and MacKay explicitly demonstrated the superior performance of irregular codes over regular ones like Ping’s. A POSITA would combine this with Divsalar because using a simple repeater, as taught by Divsalar, was a well-known, cost-effective, and straightforward method for implementing the outer coder functionality described in Ping.
- Expectation of Success: A POSITA would have a reasonable expectation of success because the combination involved applying a known improvement (irregularity from MacKay) to a known architecture (Ping's LDPC-accumulate code) using a standard component (a repeater from Divsalar).
Ground 2: Obviousness of Claim 13 over Ping in view of MacKay, Divsalar, and Luby97
- Prior Art Relied Upon: Ping, MacKay, Divsalar, and Luby97 (a 1997 publication that introduced irregularity and discussed data streams).
- Core Argument for this Ground:
- Prior Art Mapping: This ground built upon the combination in Ground 1 to meet the additional limitations of claim 13. Claim 13 requires the encoder to operate on message bits from a "source data stream." Petitioner asserted that while Ping taught block-based coding, Luby97 explicitly taught receiving data as a stream which is then partitioned into blocks for encoding. The argument was that the base combination of Ping, MacKay, and Divsalar taught the required low-density generator matrix (LDGM) coder and accumulator, while Luby97 supplied the missing "stream" limitation.
- Motivation to Combine: The motivation to combine Ping, MacKay, and Divsalar remained the same as in Ground 1. A POSITA would be motivated to incorporate the teaching of Luby97 because processing data streams was a common and practical requirement for encoders. Adapting a block-based encoder to accept streamed input was presented as a simple and obvious design choice.
- Expectation of Success: Incorporating stream-based processing into a block-based encoder was a routine task for a POSITA, leading to a high expectation of success.
Ground 3: Obviousness of Claim 17 over Ping in view of MacKay, Divsalar, and Pfister
- Prior Art Relied Upon: Ping, MacKay, Divsalar, and Pfister (a 1999 publication on Repeat-Accumulate-Accumulate codes).
- Core Argument for this Ground:
- Prior Art Mapping: This ground built upon the combination for claim 12 to meet the additional limitation of claim 17, which requires "a second accumulator." Petitioner argued that Pfister explicitly taught Repeat-Accumulate-Accumulate (RAA) codes, which are simply Repeat-Accumulate (RA) codes (like those in Divsalar) with an additional accumulator chained in series. Therefore, adding a second accumulator from Pfister to the system of Ping/MacKay/Divsalar would satisfy the claim 17 limitation.
- Motivation to Combine: The motivation to add a second accumulator was explicitly provided by Pfister itself. Pfister compared the performance of its RAA codes to Divsalar's RA codes and concluded that the extra accumulator improved performance. A POSITA would thus have been directly motivated by Pfister's own teachings to apply this improvement to the base code structure.
- Expectation of Success: Since Pfister demonstrated that adding a second accumulator was a successful method for improving performance in a highly similar system, a POSITA would have a strong expectation that the same modification would work predictably in the proposed combination.
4. Key Claim Construction Positions
Petitioner asserted that two terms were central to the invalidity arguments and proposed constructions consistent with a prior Board decision (IPR2015-00060) involving a related patent.
- "irregular": Petitioner adopted the prior construction that "irregular" refers to "the notion that different message bits or groups of message bits contribute to different numbers of parity bits." This construction was crucial for applying the teachings of MacKay, which focused on creating irregular codes with non-uniform column weights.
- "Tanner graph": Petitioner advocated for the Board's prior, detailed construction of this term. A key element of this construction was the constraint that "a parity bit is determined as a function of both information bits and other parity bits." Petitioner argued this construction was supported by the ’032 patent's specification and was met by the recursive nature of the accumulator taught in the Ping reference.
5. Relief Requested
- Petitioner requested that the Board institute an inter partes review and cancel claims 11-17 of the ’032 patent as unpatentable under 35 U.S.C. §103.