PTAB

IPR2017-00701

Apple Inc. v. California Institute of Technology

Log In

1. Case Identification

2. Patent Overview

  • Title: Serial Concatenation of Interleaved Convolutional Codes Forming Turbo-Like Codes
  • Brief Description: The ’032 patent discloses irregular repeat-accumulate (IRA) error-correcting codes. The system uses an encoder with an outer coder that irregularly repeats message bits, an interleaver that scrambles the repeated bits, and an inner coder (accumulator) that processes the scrambled bits to generate a sequence of parity bits for transmission.

3. Grounds for Unpatentability

Ground 1: Claims 1-10 are obvious over Ping in view of MacKay, Divsalar, and Luby97.

  • Prior Art Relied Upon: Ping (a 1999 article on LDPC codes), MacKay (a 1999 article on irregular Gallager codes), Divsalar (a 1998 article on turbo-like codes), and Luby97 (a 1997 article on loss-resilient codes).
  • Core Argument for this Ground:
    • Prior Art Mapping: Petitioner argued that the combination of these four references teaches every limitation of the challenged claims. Independent claim 1 recites a method of generating a sequence of parity bits by receiving a collection of message bits and, for each parity bit, summing the previous parity bit with a sum of "randomly chosen irregular repeats of the message bits."
      • Ping was asserted to teach the fundamental two-stage encoding structure of the ’032 patent: a first stage that generates sums of information bits (using a low-density matrix) followed by a second recursive accumulation stage that uses these sums and the previously generated parity bit to create the next parity bit. However, Petitioner contended that Ping’s code is regular, as each information bit contributes to the same number of parity bits.
      • MacKay was introduced to supply the missing element of irregularity. MacKay taught that the best-performing Low-Density Parity-Check (LDPC) codes are irregular, where different information bits contribute to different numbers of parity bits (i.e., its parity-check matrix has non-uniform column weights).
      • Divsalar was argued to teach an explicit implementation of a "repeat-accumulate" (RA) encoder, comprising a repeater, an interleaver, and an accumulator. This provided a practical, well-known method for implementing the repeating and interleaving functions inherent in Ping's matrix-based design.
      • Luby97 was cited for its disclosure of processing data as a "stream" partitioned into "blocks," which Petitioner argued would make it obvious to adapt the block-based coding of the other references to operate on the "source data stream" recited in claim 1.
    • Motivation to Combine: A Person of Ordinary Skill in the Art (POSITA) would combine these references for several reasons.
      • A POSITA would combine Ping and MacKay because both address improving the performance and complexity of LDPC codes. MacKay explicitly demonstrated that irregular codes outperform the regular codes taught by Ping, providing a clear motivation to apply MacKay's teaching of irregularity to Ping's architecture to achieve better performance.
      • A POSITA would incorporate Divsalar’s repeater and interleaver into Ping's design as a straightforward and cost-effective way to implement the repeating and scrambling functions. The structures are highly analogous, as both Ping and Divsalar use a final accumulation step.
      • Finally, a POSITA would find it obvious to use the stream-based processing of Luby97 with the combined system, as this reflects a common and practical way to handle real-time data encoding.
    • Expectation of Success: A POSITA would have a reasonable expectation of success in making this combination. The proposed modifications involved applying known techniques (irregularity, repeaters, interleavers) to a known code architecture (LDPC-accumulate) to achieve a predictable improvement in performance, all within the same field of error-correcting codes.
    • Key Aspects: Petitioner mapped the combination to dependent claims as follows:
      • Claim 2 (constant number of summed message bits) was taught by Ping’s primary example.
      • Claim 3 (varying number of summed message bits) was taught by modifying Ping with MacKay's disclosure of non-uniform row weights.
      • Claim 5 (irregular repeats distributed in a random sequence) was taught by the combination of MacKay's irregularity and Divsalar's random interleaver.
      • Claim 6 (using a low-density generator matrix) was taught by Ping, whose encoding matrix is inherently low-density.
      • Claims 9 and 10 (non-systematic and systematic codes) were obvious as this was a well-known design choice, with Divsalar teaching a non-systematic implementation and Ping teaching a systematic one.

4. Key Claim Construction Positions

Petitioner argued for adopting constructions from a prior IPR proceeding (IPR2015-00060) involving a related patent, asserting they represent the broadest reasonable interpretation.

  • Equation in "generating" step of claim 1: Petitioner adopted the Board's prior construction that the parity bit xj is the sum of two components: (a) the previous parity bit xj-1 and (b) the sum of a number, a, of randomly chosen irregular repeats of the message bits. This construction was central to mapping Ping's two-stage accumulator structure to the claim.
  • "irregular": Petitioner adopted the Board's 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 key to establishing the motivation to combine Ping (regular) with MacKay (irregular).

5. Relief Requested

  • Petitioner requested the institution of an inter partes review and the cancellation of claims 1-10 of the ’032 patent as unpatentable under 35 U.S.C. §103.