PTAB

IPR2017-01240

Marvell Semiconductor Inc v. Spectra Licensing Group LLC

Key Events
Petition
petition

1. Case Identification

2. Patent Overview

  • Title: Method and Device for Reception of a Series of Digital Symbols
  • Brief Description: The ’388 patent discloses an iterative method for processing received signals in a digital communications receiver. The method, termed "turbo equalization," uses a modular architecture to repeatedly perform equalization and decoding steps, passing calculated "correction information" between iterations to progressively improve the accuracy of the output.

3. Grounds for Unpatentability

Ground 1: Claims 1-4 and 7-9 are obvious over Zhou in view of Berrou.

  • Prior Art Relied Upon: Zhou ("Decision-Feedback Equalization of Time-Dispersive Channels with Coded Modulation," a 1990 IEEE journal article) and Berrou (French Patent No. 2675971B1).
  • Core Argument for this Ground:
    • Prior Art Mapping: Petitioner argued that the combination of Zhou and Berrou taught every limitation of the challenged claims. Zhou was asserted to teach the foundational architecture of the invention: a two-stage, iterative receiver using cascaded modules, with each module performing equalization to combat inter-symbol interference (ISI) followed by decoding. However, Zhou did not explicitly disclose the specific method for computing and feeding back the "correction information" that is central to the ’388 patent's iterative improvement. Petitioner contended that Berrou supplied this missing element. Berrou disclosed the seminal "Turbo Principle" for iterative decoding, a nearly optimal method where two decoders iteratively exchange "soft" extrinsic information to enhance performance.

    • Petitioner mapped the combination to the independent claims (device claim 1 and method claim 9) by arguing that Zhou's equalizer performs the claimed "correcting inter-symbol interference" step and its soft decoder performs the "decoding" step. Berrou's method for computing extrinsic information (via its estimation block 50) and feeding it to the next iteration was argued to directly teach the claimed "computing" and "delivering" of correction information. Dependent claims were also allegedly taught; for example, claim 7's requirement for de-interleaving means between the correction and decoding means was explicitly shown in Zhou's receiver diagram.

    • Motivation to Combine: Petitioner asserted that a Person of Ordinary Skill in the Art (POSITA) would have been motivated to combine the references for several reasons. The primary motivation was to improve the performance of Zhou's known iterative equalization-and-decoding system. A POSITA would have recognized the strong analogy between the problems of equalization and decoding, as an ISI channel has a similar effect on transmitted symbols as a channel encoder. With this knowledge, it would have been an obvious and natural step to apply Berrou's well-known and highly effective Turbo Principle to Zhou's architecture. Alternatively, a POSITA could have started with Berrou's two-decoder system and replaced the first decoder with a standard equalizer, a natural substitution to address ISI.

    • Expectation of Success: The petition contended that a POSITA would have had a high expectation of success. The performance benefits of the Turbo Principle were well-documented, and applying the technique to the analogous problem of equalization would have predictably resulted in a significant performance improvement. Petitioner noted that subsequent scholarly articles described "turbo equalization" as a "natural extension" of the turbo decoding algorithm.

4. Key Claim Construction Positions

  • "weighted [values/symbols]": Petitioner argued this phrase, not a standard term of art and undefined in the patent, should be construed as "non-binary [values/symbols]." The proposed construction was based on intrinsic evidence and related patents linking the term to the common concepts of "soft output" or "soft decision." This construction was critical to the invalidity argument, as it allowed Berrou's teachings of using non-binary, soft information in its iterative process to be mapped directly onto the claim limitations.
  • "or i greater than 1": Petitioner argued that this phrase in claim 2 contained a typographical error and should be construed as "for i greater than 1" to be grammatically and logically consistent with the rest of the claim and the specification. Correcting this apparent error was necessary for the claim to be coherent.
  • Means-Plus-Function Terms: The petition proposed constructions for several means-plus-function limitations, including "inter-symbol interference correction means" (claim 1), "means for the decoding" (claim 1), and "means for the computation of an item of correction information" (claim 1). The proposed structures were identified as well-known components described in the specification, such as an equalizer, a convolutive decoder (e.g., implementing a Viterbi algorithm), and an adder/subtractor. By linking the claimed functions to these conventional structures, Petitioner's constructions supported mapping the standard components found in the Zhou and Berrou references onto the claims.

5. Relief Requested

  • Petitioner requests institution of an inter partes review and cancellation of claims 1-4 and 7-9 of Patent 6,108,388 as unpatentable.