EMC Corp v. PersonalWeb Technologies LLC

1. Case Identification

• Case #: IPR2013-00084
• Patent #: 7,945,544
• Filed: December 16, 2012
• Petitioner(s): EMC Corporation
• Patent Owner(s): PersonalWeb Technologies LLC
• Challenged Claims: 1

2. Patent Overview

• Title: Similarity-Based Access Control of Data in a Data Processing System
• Brief Description: The ’544 patent discloses a data storage system that uses content-based, location-independent identifiers to manage and access data items. The system generates a "substantially unique data identifier" (termed a "True Name") based on a hash of a data item's content, allowing for functions like duplicate detection regardless of the item's name or location.

3. Grounds for Unpatentability

Petitioner argued that claim 1 is unpatentable based on several anticipation and obviousness grounds, asserting that the core concept of using a "hash of hashes" for content-based identification of compound data items was well-known long before the patent's priority date.

Ground 1: Anticipation by Kantor - Claim 1 is anticipated under 35 U.S.C. §102 by Kantor.

• Prior Art Relied Upon: Kantor (“The Frederick W. Kantor Contents-Signature System Version 1.22,” a user manual dated Aug. 10, 1993).
• Core Argument for this Ground:
  • Prior Art Mapping: Petitioner argued that Kantor disclosed every element of claim 1. Kantor’s software system, FWKCS, was designed to solve the problem of duplicate files on electronic bulletin board systems (BBSs) by creating unique, location-independent “contents-signatures” based on file content. For a simple file, the signature was derived from a CRC hash of the file’s contents and its length. For a multi-part, or compound, data item (a “zipfile”), Kantor taught computing contents-signatures for each individual file within the zipfile (the first function) and then applying a second hash function ("addition modulo 2^32") to these individual signatures to create a single "zipfile contents-signature" for the entire package (the second function). FWKCS then ascertained whether two zipfiles corresponded by comparing their final zipfile contents-signatures, directly mapping to the method of claim 1.

Ground 2: Anticipation by Browne - Claim 1 is anticipated under §102 by Browne.

• Prior Art Relied Upon: Browne (“Location-Independent Naming for Virtual Distributed Software Repositories,” a technical report dated Feb. 1995).
• Core Argument for this Ground:
  • Prior Art Mapping: Petitioner asserted that Browne disclosed a system for managing distributed software repositories that anticipated claim 1. Browne addressed the shortcomings of location-based identifiers (like URLs) by creating “Location-Independent File Names” (LIFNs), which are MD5 hashes of file contents. For compound data items, which Browne called "resources" (e.g., a software package with multiple files), the system computed a LIFN for each component file (the first function). It then created a composite list of these component LIFNs and computed a final LIFN for the entire resource by performing an MD5 hash on that list (the second function). Browne used these final LIFNs to verify file integrity and check for correspondence by comparing the signatures of two resources, thereby teaching all steps of claim 1.

Ground 3: Anticipation by Woodhill - Claim 1 is anticipated under §102(e) by Woodhill.

• Prior Art Relied Upon: Woodhill (Patent 5,649,196).

• Core Argument for this Ground:

  • Prior Art Mapping: Petitioner contended that Woodhill’s disclosure of a distributed storage management system anticipated claim 1. Woodhill taught breaking large files into smaller segments called “granules.” A "contents identifier" (a first value, derived from a hash function) was calculated for each granule based on its content (the first function). The contents identifiers for all granules in a file were then collected into a “shadow file.” A “Binary Object Identifier” (a second value) for the shadow file was then calculated based on its contents—which are the granule identifiers (the second function). Woodhill disclosed comparing the Binary Object Identifiers of successive versions of a shadow file to determine if the underlying file had changed. Petitioner argued this process of creating a "hash of hashes" and comparing the final identifiers to ascertain correspondence directly anticipated the method of claim 1.

• Additional Grounds: Petitioner asserted additional anticipation challenges based on Langer (an August 1991 newsgroup post). Petitioner also presented alternative obviousness grounds, arguing that if any of Kantor, Browne, or Langer were found not to disclose a multi-part data item, it would have been obvious to combine them with Woodhill, which explicitly taught dividing files into smaller parts as a known technique for handling large files efficiently.

4. Relief Requested

• Petitioner requested institution of an inter partes review and cancellation of claim 1 of the ’544 patent as unpatentable.