PTAB

IPR2018-01606

Microsoft Corp v. Saint Regis Mohawk Tribe

Log In
Key Events
Petition
petition

1. Case Identification

2. Patent Overview

  • Title: Multi-Adaptive Processing Systems and Techniques for Enhancing Parallelism and Performance of Computational Functions
  • Brief Description: The ’800 patent describes a method for processing a "data driven" calculation on a reconfigurable computing system, such as one using Field Programmable Gate Arrays (FPGAs). The method involves simultaneously processing different "data dimensions" of a calculation using "clusters of functional units" configured to execute computational loops in parallel.

3. Grounds for Unpatentability

Ground 1: Anticipation and Obviousness over Splash2 - Claims 1 and 15 are anticipated by, or alternatively obvious over, Splash2.

  • Prior Art Relied Upon: Splash2 (a 1996 book titled "Splash 2, FPGAs in a Custom Computing Machine" by Buell et al.).
  • Core Argument for this Ground:
    • Prior Art Mapping: Petitioner argued that Splash2, which describes a complete, operational reconfigurable computing system, discloses every limitation of claims 1 and 15. The book detailed the implementation of a systolic array for performing an "Edit Distance Algorithm" to compare genetic sequences. Petitioner mapped the claim limitations as follows: the Splash 2 system is the "reconfigurable computing system"; its FPGAs are the "reconfigurable processor"; and the configured Processing Elements (PEs) within the FPGAs are the "plurality of functional units." The process of compiling the algorithm into a hardware configuration for the FPGAs met the limitation of "transforming an algorithm into a data driven calculation." The systolic processing, where PEs operate upon the arrival of data, was argued to be inherently "data driven." The concurrent operation of multiple PEs on different time steps of the data streams satisfied the limitation of utilizing first and second functional units on "subsequent" and "previous" data dimensions, respectively. The direct communication links between adjacent PEs in the array met the "pass computed data seamlessly" limitation. For claim 15, the streaming of data between PEs established the required "stream communication connection."
    • Motivation to Combine (for §103 ground): As an alternative to anticipation, Petitioner argued obviousness. The motivation to combine the system architecture described in Chapters 1-6 of Splash2 with the systolic array algorithms of Chapter 8 was self-evident, as the book explicitly stated that the algorithms were implemented on the described system.
    • Expectation of Success (for §103 ground): A POSITA would have had a very high expectation of success because Splash2 described a system that had already been successfully built and used for the exact type of computation claimed.

Ground 2: Obviousness of Dependent Claims over Splash2 and Chunky SLD - Claims 7, 17, and 24 are obvious over Splash2 in view of Chunky SLD.

  • Prior Art Relied Upon: Splash2 (1996 book), Chunky SLD (a 1997 paper titled "Automated Target Recognition on SPLASH 2").
  • Core Argument for this Ground:
    • Prior Art Mapping: This ground built upon the teachings of Splash2 for the base system of claim 1 and added Chunky SLD to teach the specific application limitations of the dependent claims. Chunky SLD explicitly disclosed its implementation on the Splash 2 system. It described an automatic target resolution algorithm, which is a "search algorithm for an image search" (meeting claim 17). The algorithm was specifically applied to process data from Synthetic Aperture Radar (SAR), thereby teaching a "synthetic aperture radar imaging calculation" (meeting claim 7). Petitioner contended that the algorithm, which involved applying a template (filter) to image data, was a form of a "signal filtering application" (meeting claim 24).
    • Motivation to Combine: The motivation was exceptionally strong, as Chunky SLD expressly identified the Splash 2 system as the platform on which its algorithm was implemented. A POSITA seeking to implement the image search algorithm from Chunky SLD would have been directly motivated to use the well-documented and successful systolic array processing techniques described in the Splash2 book on the same Splash 2 hardware.
    • Expectation of Success: Success would have been predictable, as the combination involved using a known algorithm (Chunky SLD) on its stated hardware platform (Splash 2) using that platform's known programming and operational paradigms.

Additional Grounds: Petitioner also asserted that claims 1 and 15 are obvious over Splash2 in view of Gaudiot (a 1987 paper on data-driven multiprocessors). Gaudiot was introduced to provide an explicit teaching of scheduling operations "upon the availability" of operands to bolster the argument that the claimed calculation was "data driven."

4. Key Claim Construction Positions

  • "functional unit": Petitioner proposed construing this term to mean "a computational unit configured to perform a specific task." This construction was based on the patent’s description of such units in both conventional CPUs and reconfigurable processors.
  • "data driven": Petitioner argued this term means "the scheduling of operations upon the availability of their operands." This was based on an external technical dictionary and the Gaudiot reference, as the ’800 patent itself provided no explicit definition.
  • "form": Petitioner proposed this term means "create, such as by configuring, a particular structure." This argument was based on linking the term to "instantiate" used in the parent patent and defined during its prosecution, suggesting the terms were interchangeable.

5. Relief Requested

  • Petitioner requests institution of an inter partes review and cancellation of claims 1, 7, 15, 17, and 24 of the ’800 patent as unpatentable.