PTAB

IPR2020-01041

DISH Network LLC v. Sound View Innovations LLC

Log In
Key Events
Petition
petition

1. Case Identification

2. Patent Overview

  • Title: Method for Ensuring Quality of Service in a Computer System
  • Brief Description: The ’456 patent relates to techniques for providing a desired quality of service (QoS) for applications running in a computer system. The invention discloses methods for integrating proportional share schedulers into operating systems to provide QoS guarantees, particularly for real-time applications competing for resources like CPU, memory, and network bandwidth.

3. Grounds for Unpatentability

Ground 1: Obviousness over Durand in view of Bennett - Claim 13 is obvious over Durand in view of Bennett.

  • Prior Art Relied Upon: Durand (Patent 6,338,072) and Bennett (a 1997 IEEE/ACM article on hierarchical packet fair queueing).
  • Core Argument for this Ground:
    • Prior Art Mapping: Petitioner argued that Durand teaches a UNIX-based system for dynamically controlling resource allocation using an application programming interface (API) to manage processes grouped into "dimensions," each assigned a relative weight. This system provides QoS by adjusting process priorities based on these weights. Petitioner contended that while Durand discloses proportional sharing, its system is peer-based rather than hierarchical. Bennett was argued to supply the missing hierarchical structure, disclosing an idealized Hierarchical Generalized Processor Sharing (H-GPS) model that allocates resources using a tree structure. Bennett explicitly teaches parent-child node relationships, service shares (weights), and minimum guaranteed service rates for each node, fulfilling the hierarchical limitations of claim 13.
    • Motivation to Combine: A POSITA would combine Durand and Bennett to address the common problem of resource allocation in a more efficient and predictable manner. Petitioner asserted a POSITA would be motivated to improve Durand's peer-based dimension structure with Bennett's superior hierarchical scheduling model. Bennett's approach, which provides guaranteed minimum resource allocations for every node, represented a known technique for enhancing the proportional sharing taught by Durand to achieve more robust QoS control.
    • Expectation of Success: A POSITA would have a high expectation of success in the combination. Both references address proportional resource sharing, making the integration of Bennett's hierarchical concepts into Durand's system straightforward. A POSITA could predictably migrate from Durand's relative weight approach to Bennett's hierarchical service share model using well-known UNIX-based hierarchical inheritance concepts, as the underlying principles of allocating deterministic resource shares are analogous.

Ground 2: Obviousness over Goyal in view of Bennett - Claim 13 is obvious over Goyal in view of Bennett.

  • Prior Art Relied Upon: Goyal (a 1996 USENIX paper on hierarchical CPU scheduling) and Bennett (the same 1997 IEEE/ACM article).
  • Core Argument for this Ground:
    • Prior Art Mapping: Petitioner argued that Goyal discloses a hierarchical CPU allocation framework in a UNIX-derivative operating system to provide QoS guarantees, particularly for multimedia applications. Goyal teaches a tree-like scheduling structure where CPU bandwidth is partitioned among application classes based on weights, and applications use an API to communicate requirements to a QoS Manager. Petitioner asserted that while Goyal teaches hierarchical partitioning and weights, Bennett provides a more explicit and robust teaching of guaranteed service rates and the specific mathematical relationships for resource allocation between parent and child nodes. Bennett's disclosure ensures that the sum of minimum resources for child nodes is at least equal to the minimum resources of the parent node.
    • Motivation to Combine: A POSITA would combine Goyal and Bennett because they address the same fundamental problem of hierarchical resource allocation. Goyal notes the desirability of providing "bounds on minimum throughput," and Bennett explicitly solves this problem by providing a "guaranteed service rate" for each node in the hierarchy. A POSITA would be motivated to enhance Goyal's system by incorporating Bennett's specific mechanism for guaranteeing minimum resource levels, thereby improving the predictability and robustness of the QoS framework.
    • Expectation of Success: A POSITA would have a high expectation of success because both Goyal and Bennett describe similar hierarchical, proportional-sharing systems. Integrating Bennett’s explicit guaranteed service rate mechanism into Goyal’s existing hierarchical framework would be a predictable modification. The combination would simply apply a known technique (Bennett's guaranteed rates) to a similar system (Goyal's scheduler) to achieve a predictable improvement in performance.

4. Key Claim Construction Positions

  • "correspond to a reference to an object": Petitioner argued this term should be construed to include indirect references, such as file descriptors and pointers to computing objects (e.g., processes, files, memory objects). This construction was central to their argument that the prior art's use of file descriptors to point to files containing weights or process identifiers (PIDs) to reference processes meets this claim limitation.

5. Arguments Regarding Discretionary Denial

  • Petitioner argued against discretionary denial under 35 U.S.C. §314(a) and §325(d). They contended this petition is not cumulative to other IPRs filed against the ’456 patent because it relies on different primary prior art (Durand and Goyal) and presents distinct theories of obviousness. Furthermore, Petitioner argued the cited references were not substantively considered during the original prosecution; Durand was never before the examiner, and Goyal, though cited in the background, was never applied in any rejection.

6. Relief Requested

  • Petitioner requests institution of an inter partes review and cancellation of claim 13 of Patent 6,725,456 as unpatentable.