PTAB

IPR2020-00337

Juniper Networks Inc v. Packet Intelligence LLC

Key Events

Petition

petition Intelligence

Title: Packet Flow Classification
Brief Description: The ’646 patent discloses methods and systems for monitoring and classifying network data packets into "conversational flows." The technology purports to identify and link multiple, distinct connection flows that result from a single underlying software application activity, distinguishing itself from prior art that allegedly only recognized individual "connection flows."

Prior Art Relied Upon: Riddle (Patent 6,412,000), Ferdinand (WO 92/19054), and Wakeman (Patent 5,740,175).
Core Argument for this Ground:
- Prior Art Mapping: Petitioner argued that Riddle teaches the core concept of a "conversational flow" through its disclosure of classifying traffic based on "service aggregates," which group multiple connection flows associated with a single application (e.g., FTP). Ferdinand, a network monitoring patent, was argued to teach using a database for storing flow-entry information and using buffers, while Wakeman was argued to teach employing a content-addressable memory (CAM) cache to improve lookup speed for frequently accessed data.
- Motivation to Combine: A person of ordinary skill in the art (POSITA) would combine Riddle's classification system with Ferdinand’s database to gain enhanced data management functionalities like searching and modifying flow-entries. A POSITA would further integrate Wakeman’s CAM cache into the combined Riddle/Ferdinand system to predictably reduce lookup times and improve overall performance, a common design choice in network devices.
- Expectation of Success: The combination involved applying known technologies (databases, caches) to a packet classification system for their conventional purposes, which would have yielded predictable improvements in functionality and performance.

Prior Art Relied Upon: Riddle (Patent 6,412,000), Ferdinand (WO 92/19054), Wakeman (Patent 5,740,175), and Yu (Patent 6,625,150).
Core Argument for this Ground:
- Prior Art Mapping: This ground built upon the core combination of Ground 1, adding Yu to further support the obviousness of the "conversational flow" and state processing limitations. Petitioner contended that Yu explicitly teaches classifying multiple "streams" (analogous to connection flows) into a single "flow" based on application-level criteria, directly mapping to the claimed "conversational flow." Yu also disclosed stateful packet inspection, where its logic "keeps track of the flow's state until a matching criteria is met."
- Motivation to Combine: A POSITA seeking to improve Riddle's classifier would have looked to Yu's flexible and efficient architecture for stateful packet inspection. The motivation was to integrate Yu's state-tracking logic to more robustly classify flows based on application behavior over multiple packets.

Prior Art Relied Upon: Riddle (Patent 6,412,000), Ferdinand (WO 92/19054), Wakeman (Patent 5,740,175), and RFC1945 (Request for Comments 1945).
Core Argument for this Ground:
- Prior Art Mapping: This ground also built upon the combination in Ground 1, adding RFC1945 to demonstrate that linking separate connection flows was a well-understood technique in the art. Petitioner asserted that RFC1945, the specification for HTTP/1.0, discloses using the "Referer" header field to link a new HTTP request to a prior one, effectively creating a "conversation." This argument was bolstered by pointing to the Patent Owner's expert testimony in co-pending litigation, which allegedly used this exact concept to define a "conversational flow."
- Motivation to Combine: A POSITA implementing Riddle's system for classifying HTTP traffic would have routinely consulted the foundational RFC1945 standard. It would have been obvious to use the header fields described in RFC1945, like "Referer," to identify and link related connection flows, thereby creating the claimed "conversational flow."

"Conversational Flow": Petitioner argued for a construction narrower than that used in a prior IPR's Broadest Reasonable Interpretation. Citing the Patent Owner's own statements in prior proceedings, Petitioner proposed the term means "the sequence of packets that are exchanged...as a result of specific software program activity, where such packets form multiple connection flows that are linked based on that activity." This construction is based on Patent Owner's use of examples like a single Skype call generating separate but linked audio, video, and control flows.
"The Flow" / "New Flow" / "Existing Flow": Petitioner argued that based on the claim language and logic, these terms refer to a "new conversational flow" and an "existing conversational flow," not merely a connection flow. This interpretation was supported by the prosecution history and prior Board findings.
"Flow-Entry Database": Petitioner argued this term should be construed as a "database having a separate entry for each encountered conversational flow." This was based on statements made by the patentee during prosecution to distinguish the invention from prior art that did not maintain such distinct entries.

Petitioner argued that discretionary denial under 35 U.S.C. §325(d) or §314(a) would be inappropriate. They asserted this was their first petition challenging the ’646 patent and that none of the five primary prior-art references were considered during the original examination. Furthermore, they contended that while a prior inter partes review (IPR) existed (IPR2017-00450), it relied on a different primary reference (Engel) and a different theory, and another IPR (IPR2019-01292) was terminated before any substantive review by the Board.

Petitioner requests institution of an IPR and cancellation of claims 1-3, 7, 16, and 18 of the ’646 patent as unpatentable.