Packet Intelligence LLC

Patent Trials and Appeals Board

Sep 8, 2021

IPR2020-00337 (P.T.A.B. Sep. 8, 2021)

Trials@uspto.gov Paper 48
571-272-7822 Date: September 8, 2021

UNITED STATES PATENT AND TRADEMARK OFFICE

BEFORE THE PATENT TRIAL AND APPEAL BOARD

JUNIPER NETWORKS, INC. and PALO ALTO NETWORKS, INC.,
Petitioner,
v.
PACKET INTELLIGENCE LLC,
Patent Owner.

IPR2020-00337
Patent 6,771,646 B1

Before STACEY G. WHITE, CHARLES J. BOUDREAU, and
JOHN D. HAMANN, Administrative Patent Judges.
BOUDREAU, Administrative Patent Judge.

JUDGMENT
Final Written Decision
Determining Some Challenged Claims Unpatentable
35 U.S.C. § 318(a)

IPR2020-00337
Patent 6,771,646 B1
2
I. INTRODUCTION
This is a Final Written Decision in an inter partes review challenging
the patentability of claims 1–3, 7, 16, and 18 (“the challenged claims”) of
U.S. Patent No. 6,771,646 B1 (Ex. 1003, “the ’646 patent”). We have
jurisdiction under 35 U.S.C. § 6 and enter this Decision pursuant to
35 U.S.C. § 318(a) and 37 C.F.R. § 42.73. For the reasons set forth below,
we determine that Petitioner has shown, by a preponderance of the evidence,
that claims 1, 2, 7, 16, and 18 are unpatentable, but that Petitioner has not
shown that claim 3 is unpatentable. See 35 U.S.C. § 316(e).
II. BACKGROUND
A. Procedural History
Juniper Networks, Inc. and Palo Alto Networks, Inc. (collectively
“Petitioner”) filed a Petition (Paper 3, “Pet.”) requesting inter partes review
of claims 1–3, 7, 16, and 18 of the ’646 patent pursuant to 35 U.S.C. § 311.
Petitioner supported its Petition with the Declaration of Dr. Jon B.
Weissman. Ex. 1006. Packet Intelligence LLC (“Patent Owner”) filed a
Preliminary Response. Paper 7 (“Prelim. Resp.”).1
On September 10, 2020, pursuant to 35 U.S.C. § 314(a), we instituted
trial to determine whether any challenged claim of the ’646 patent is
unpatentable based on the grounds raised in the Petition:

1 On our authorization (Paper 8), Petitioner also filed a Preliminary Reply
(Paper 9), and Patent Owner filed a Preliminary Sur-reply (Paper 10) related
to discretionary denial of institution under 35 U.S.C. § 314(a).
IPR2020-00337
Patent 6,771,646 B1
3
Claims
Challenged 35 U.S.C. §
2 References
1–3, 7, 16, 18 103(a) Riddle,3 Ferdinand,4 Wakeman5
1–3, 7, 16, 18 103(a) Riddle, Ferdinand, Wakeman, Yu6
1–3, 7, 16, 18 103(a) Riddle, Ferdinand, Wakeman, RFC19457
Paper 20, 8, 56 (“Institution Decision” or “Inst. Dec.”).8
Patent Owner filed a Response. Paper 26 (“PO Resp.”). Patent
Owner supported its Response with the Declaration of Cathleen T. Quigley.
Ex. 2061. Petitioner filed a Reply to Patent Owner’s Response. Paper 30
(“Reply”). Patent Owner filed a Sur-reply. Paper 32 (“Sur-reply”).
A combined oral hearing in this proceeding and IPR2020-00336,
involving a related patent, was held on June 9, 2021. A transcript of the
hearing is included in the record. Paper 47 (“Tr.”). The transcript of an oral
hearing held the same day in cases IPR2020-00338, IPR2020-00339, and

2 The Leahy-Smith America Invents Act (“AIA”) included revisions to
35 U.S.C. § 103 that became effective on March 16, 2013. Because the
’646 patent issued from an application filed before March 16, 2013, we
apply the pre-AIA version of the statutory basis for unpatentability.
3 Riddle et al., US 6,412,000 B1 (issued June 25, 2002) (Ex. 1008).
4 Ferdinand et al., WO 92/19054 (published Oct. 29, 1992) (Ex. 1009).
5 Wakeman et al., US 5,740,175 (issued Apr. 14, 1998) (Ex. 1014).
6 Yu, US 6,625,150 B1 (issued Sept. 23, 2003) (Ex. 1011).
7 T. Berners-Lee et al., Hypertext Transfer Protocol -- HTTP/1.0, Request
for Comments: 1945, Network Working Group (May 1996) (Ex. 1010).
8 Patent Owner filed a Request for Rehearing of the Institution Decision
(Paper 24), which we denied (Paper 27).
IPR2020-00337
Patent 6,771,646 B1
4
IPR2020-00486, also involving patents related to the ’646 patent, also is
included in the record of this proceeding.9 Paper 46.
Following oral hearing, we ordered the parties to provide additional
briefing on the claim-construction arguments presented in the briefs and at
oral hearing. Paper 41 (“Order”). Petitioner and Patent Owner each filed
respective Opening Briefs on claim construction. See Paper 42
(“Petitioner’s Opening Brief” or “Pet. Br.”); Paper 43 (“Patent Owner’s
Opening Brief” or “PO Br.”). Petitioner filed a Responsive Brief to Patent
Owner’s Opening Brief, Paper 44 (“Petitioner’s Responsive Brief” or “Pet.
Resp. Br.”), and Patent Owner filed a Responsive Brief to Petitioner’s
Opening Brief, Paper 45 (“Patent Owner’s Responsive Brief” or “PO Resp.
Br.”).
B. Real Parties in Interest
Petitioner identifies Juniper Networks, Inc. and Palo Alto Networks,
Inc. as its real parties in interest. Pet. 1. Patent Owner identifies Packet
Intelligence LLC and Packet Intelligence Holdings LLC as its real parties in
interest. Paper 6, 2.
C. Related Matters
The parties identify two district court litigations as related matters that
involve the ’646 patent: Packet Intelligence LLC v. Juniper Networks, Inc.,
3:19-cv-04741 (N.D. Cal.) and Palo Alto Networks, Inc. v. Packet
Intelligence LLC, No. 3:19-cv-02471 (N.D. Cal). Pet. 1; Paper 6, 2. The
parties also identify Packet Intelligence LLC v. NetScout Systems, Inc., 2:16-

9 The parties had no objection to entering into this record the transcript from
the oral hearing for IPR2020-00338, IPR2020-00339, and IPR2020-00486.
Tr. 7:15–8:5; see also Paper 46, 5:22–6:10.
IPR2020-00337
Patent 6,771,646 B1
5
cv-230-JRG (E.D. Tex.) and Packet Intelligence LLC v. NetScout Sys., Inc.,
No. 19-2041 (Fed. Cir.) as related matters. Pet. 1; Paper 6, 2.
The parties also identify as related matters IPR2017-00450 and
IPR2019-01292, no longer pending before the Board, which challenged
certain claims of the ’646 patent, as well as certain other earlier proceedings
that challenged claims of various related patents. Pet. 2; Paper 6, 3–5.
D. The ’646 Patent (Ex. 1003)
The ’646 patent, titled “Associative Cache Structure for Lookups and
Updates of Flow Records in a Network Monitor,” discloses a network
activity monitor with a cache subsystem. Ex. 1003, code (54), 1:42–3:14.
The ’646 patent explains that there was a need in the art for “a real-time
network monitor that can provide details as to the application programs
being used.” Id. at 1:42–47. The disclosed monitor receives packets passing
in either direction through its connection point on the network and
“elucidate[s] what application programs are associated with each packet” by
extracting information from the packet, using selected parts of the extracted
information to “build[] a signature for identifying the conversational flow of
the packet,” and performing a lookup of “a database of flow records for
previously encountered conversational flows to determine whether [the]
signature is from an existing flow.” Id. at 1:66–2:28, 4:61–5:8, Fig. 1. The
’646 patent states that due to the high speed at which packets enter the
system, it is advantageous to use a cache system for the memory containing
the flow database. Id. at 2:37–62. “One desirable property of such a cache
system is a least recently used (LRU) replacement policy that replaces the
LRU flow-entry when a cache replacement is needed.” Id. at 2:53–56.
“Replacing least recently used flow-entries is preferred because it is likely
IPR2020-00337
Patent 6,771,646 B1
6
that a packet following a recent packet will belong to the same flow.” Id.
at 2:56–58.
Figure 3 of the ’646 patent is reproduced below.

Figure 3, above, depicts various components of network packet monitor 300,
including parser subsystem 301, analyzer subsystem 303, and database of
known flows 324. Ex. 1003, 7:36–58. Parser subsystem 301 “parses the
packet and determines the protocol types and associated headers for each
protocol layer that exists in the packet 302,” “extracts characteristic portions
(signature information) from the packet 302,” and builds the “unique flow
signature (also called a ‘key’) for this flow.” Id. at 8:5–9:28, 27:66–29:61
(describing an example of how the disclosed monitor builds signatures and
flow states in the context of a Sun Remote Procedure Call (RPC), where,
IPR2020-00337
Patent 6,771,646 B1
7
after all of the required processing, “KEY-2 may . . . be used to recognize
packets that are in any way associated with the application ‘a2’”), Fig. 2.
Analyzer system 303 then determines whether the packet has a
matching flow-entry in database of flows 324, and processes the packet
accordingly, including, for example, determining whether the packet belongs
to an existing conversational flow or a new (i.e., not previously encountered)
flow and, in the case of the latter, performing state processing to determine
whether the conversational flow has been “fully characterized” and should
be finalized. Ex. 1003, 9:45–12:34, 19:46–20:2, 30:13–36:28, Fig. 11. The
’646 patent discloses that
[f]uture packets that are part of the same conversational flow
have their state analysis continued from a previously achieved
state. When enough packets related to an application of interest
have been processed, a final recognition state is ultimately
reached, i.e., a set of states has been traversed by state analysis
to completely characterize the conversational flow. The
signature for that final state enables each new incoming packet
of the same conversational flow to be individually recognized in
real time.
In this manner, one of the great advantages of the present
invention is realized. Once a particular set of state transitions has
been traversed for the first time and ends in a final state, a short-
cut recognition pattern—a signature—an [sic] be generated that
will key on every new incoming packet that relates to the
conversational flow. Checking a signature involves a simple
operation, allowing high packet rates to be successfully
monitored on the network.
Id. at 11:67–12:17.
IPR2020-00337
Patent 6,771,646 B1
8
E. The Challenged Claims
Petitioner challenges claims 1–3, 7, 16, and 18 of the ’646 patent, of
which claims 1, 7, and 16 are independent. Pet. 7. Claims 1 and 16,
reproduced below, illustrate the claimed subject matter.
1. A packet monitor for examining packets passing through a
connection point on a computer network, each packet
conforming to one or more protocols, the monitor comprising:
(a) a packet acquisition device coupled to the connection
point and configured to receive packets passing through
the connection point;
(b) a memory for storing a database comprising flow-
entries for previously encountered conversational flows to
which a received packet may belong, a conversational
flow being an exchange of one or more packets in any
direction as a result of an activity corresponding to the
flow;
(c) a cache subsystem coupled to the flow-entry database
memory providing for fast access of flow-entries from the
flow-entry database;
(d) a lookup engine coupled to the packet acquisition
device and to the cache subsystem and configured to
lookup whether a received packet belongs to a flow-entry
in the flow-entry database, the looking up being via cache
subsystem; and
(e) a state processor coupled to the lookup engine and to
the flow-entry-database memory, the state processor being
to perform any state operations specified for the state of
the flow starting from the last encountered state of the flow
in the case that the packet is from an existing flow, and to
perform any state operations required for the initial state
of the new flow in the case that the packet is not from an
existing flow.
16. A method of examining packets passing through a connection
point on a computer network, each packets conforming to one or
more protocols, the method comprising:
IPR2020-00337
Patent 6,771,646 B1
9
(a) receiving a packet from a packet acquisition device;
(b) performing one or more parsing/extraction operations
on the packet to create a parser record comprising a
function of selected portions of the packet;
(c) looking up a flow-entry database comprising none or
more flow-entries for previously encountered
conversational flows, the looking up using at least some of
the selected packet portions and determining if the packet
is of an existing flow, the lookup being via a cache;
(d) if the packet is of an existing flow, classifying the
packet as belonging to the found existing flow; and
(e) if the packet is of a new flow, storing a new flow-entry
for the new flow in the flow-entry database, including
identifying information for future packets to be identified
with the new flow-entry,
wherein the parsing/extraction operations depend on one or more
of the protocols to which the packet conforms.
Ex. 1003, 36:39–67, 39:10–40:4, 44–47 (Certificates of Correction).
III. ANALYSIS
We have reviewed the parties’ respective briefs as well as the relevant
evidence discussed in those papers. For the reasons discussed in detail
below, we determine that Petitioner has shown by a preponderance of the
evidence that claims 1–2, 7, 16, and 18 of the ’646 patent are unpatentable
under 35 U.S.C. § 103 as having been obvious, but has not shown by a
preponderance of the evidence that claim 3 is unpatentable.
A. Legal Standards
To prevail in its challenges to the patentability of the challenged
claims, Petitioner must demonstrate by a preponderance of the evidence that
the claims are unpatentable. 35 U.S.C. § 316(e); 37 C.F.R. § 42.1(d) (2019).
“In an [inter partes review], the petitioner has the burden from the onset to
IPR2020-00337
Patent 6,771,646 B1
10
show with particularity why the patent it challenges is unpatentable.”
Harmonic Inc. v. Avid Tech., Inc., 815 F.3d 1356, 1363 (Fed. Cir. 2016)
(citing 35 U.S.C. § 312(a)(3) (2012) (requiring inter partes review petitions
to identify “with particularity . . . the evidence that supports the grounds for
the challenge to each claim”)). This burden of persuasion never shifts to
patent owner. See Dynamic Drinkware, LLC v. Nat’l Graphics, Inc., 800
F.3d 1375, 1378 (Fed. Cir. 2015); see also In re Magnum Oil Tools Int’l,
Ltd., 829 F.3d 1364, 1375–78 (Fed. Cir. 2016) (discussing the burden of
proof in inter partes review).
A claim is unpatentable for obviousness if, to one of ordinary skill in
the pertinent art, “the differences between the subject matter sought to be
patented and the prior art are such that the subject matter as a whole would
have been obvious at the time the invention was made.” 35 U.S.C. § 103(a)
(2006); see also KSR Int’l Co. v. Teleflex Inc., 550 U.S. 398, 406 (2007).
The question of obviousness is resolved on the basis of underlying factual
determinations including (1) the scope and content of the prior art; (2) any
differences between the claimed subject matter and the prior art; (3) the level
of ordinary skill in the art; and (4) when in evidence, objective evidence of
nonobviousness.10 Graham v. John Deere Co., 383 U.S. 1, 17–18 (1966). A
petitioner cannot satisfy its burden of proving obviousness by employing
“mere conclusory statements.” Magnum Oil, 829 F.3d at 1380. Moreover, a
decision on the ground of obviousness must include “articulated reasoning
with some rational underpinning to support the legal conclusion of

10 The parties do not direct our attention to any evidence of objective indicia
of nonobviousness.
IPR2020-00337
Patent 6,771,646 B1
11
obviousness.” KSR, 550 U.S. at 418 (citing In re Kahn, 441 F.3d 977, 988
(Fed. Cir. 2006)).
We analyze Petitioner’s asserted grounds of unpatentability in
accordance with the above-stated principles.
B. Level of Ordinary Skill in the Art
We consider the asserted grounds of unpatentability in view of the
understanding of a person of ordinary skill in the art and, thus, begin with
the level of ordinary skill in the art. The level of ordinary skill in the art is
“a prism or lens through which . . . the Board views the prior art and the
claimed invention” to prevent hindsight bias. Okajima v. Bourdeau, 261
F.3d 1350, 1355 (Fed. Cir. 2001). In assessing the level of ordinary skill in
the art, various factors may be considered, including the “type of problems
encountered in the art; prior art solutions to those problems; rapidity with
which innovations are made; sophistication of the technology; and
educational level of active workers in the field.” In re GPAC Inc., 57 F.3d
1573, 1579 (Fed. Cir. 1995). “[O]ne or more factors may predominate.” Id.
Petitioner contends that an ordinarily skilled artisan would have “had
a bachelor’s degree in electrical engineering, computer engineering,
computer science, or a related field (or its equivalent), and one to two years
of experience working in networking environments, including at least some
experience with network traffic monitors and/or analyzers.” Pet. 7–8 (citing,
inter alia, Ex. 1006 ¶¶ 195–201). In our Institution Decision, we adopted
Petitioner’s articulation of the level of skill in the art, which we determined
was consistent with the ’646 patent and the asserted prior art. Inst. Dec. 25–
26. In its Response, Patent Owner states that it “generally does not object to
this finding.” PO Resp. 22.
IPR2020-00337
Patent 6,771,646 B1
12
We apply Petitioner’s definition of the level of ordinary skill in the
art, as we did in the Institution Decision. Inst. Dec. 25–26. We also
maintain our determination that the definition offered by Petitioner is
consistent with the teachings of the ’646 patent and the prior art of record.
Cf. Okajima, 261 F.3d at 1355 (noting that the prior art itself may reflect an
appropriate level of skill in the art).
C. Claim Construction
In interpreting the claims of the ’646 patent, we “us[e] the same claim
construction standard that would be used to construe the claim[s] in a civil
action under 35 U.S.C. [§] 282(b).” 37 C.F.R. § 42.100(b). The claim
construction standard includes construing claims in accordance with the
ordinary and customary meaning of such claims as would have been
understood by one of ordinary skill in the art and the prosecution history
pertaining to the patent. See id.; Phillips v. AWH Corp., 415 F.3d 1303,
1312–14 (Fed. Cir. 2005) (en banc). We need not explicitly interpret every
claim term for which the parties propose a construction. See 35 U.S.C.
§ 314(a) (2012); Vivid Techs., Inc. v. Am. Sci. & Eng’g, Inc., 200 F.3d 795,
803 (Fed. Cir. 1999) (“[O]nly those terms need be construed that are in
controversy, and only to the extent necessary to resolve the controversy.”);
see also Nidec Motor Corp. v. Zhongshan Broad Ocean Motor Co., 868 F.3d
1013, 1017 (Fed. Cir. 2017) (applying Vivid Techs. in the context of an inter
partes review).
1. Incorporation by reference of previous disclosures
The ’646 patent claims the benefit of U.S. Provisional Application
No. 60/141,903, filed on June 30, 1999 (Ex. 1016, “the provisional
application”). See Ex. 1003, code (60). The ’646 patent states that the
contents of the provisional application, as well as the contents of several
IPR2020-00337
Patent 6,771,646 B1
13
other U.S. patents and applications, including U.S. Patent No. 6,651,099,
issued on November 18, 2003 (Ex. 1001, “the ’099 patent”), “are
incorporated herein by reference.” See id. at 1:7–11 (incorporating by
reference the provisional application), 1:16–18 (incorporating by reference
the ’099 patent), 1:19–23 (incorporating by reference U.S. Patent No.
6,665,725), 1:25–29 (incorporating by reference U.S. Patent Application No.
09/608,126), 1:30–33 (incorporating by reference U.S. Patent Application
No. 09/608,267). Throughout their papers, the parties refer to the
disclosures of the ’099 patent and the provisional application, and state that
the ’646 patent incorporates the entire contents of those disclosures. See,
e.g., PO Resp. 2 n.1; Reply 2 n.1, 3–4.
We agree with the parties that the ’646 patent incorporates the
disclosures of the ’099 patent and the provisional application by specific
reference to those documents. Ex. 1003, code (60), 1:7–11, 1:16–18; see
also 37 C.F.R. § 1.57. Thus, we also refer to the ’099 patent and the
provisional application when discussing claim construction—not only for
consistency with the parties, but also for consistency throughout our final
decisions involving related inter partes proceedings between these parties.
See Advanced Display Sys., Inc. v. Kent State Univ., 212 F.3d 1272, 1282
(Fed. Cir. 2000) (stating that material incorporated by reference is
“effectively part of the host document as if it were explicitly contained
therein”); see also Trs. of Columbia Univ. v. Symantec Corp., 811 F.3d
1359, 1365–66 (Fed. Cir. 2016) (using statements in a provisional
application to construe claim terms in a patent incorporating the provisional
application by reference).
IPR2020-00337
Patent 6,771,646 B1
14
2. Overview
The only claim-construction dispute remaining in this proceeding
concerns the meaning of “conversational flow[s],” as recited in independent
claims 1, 7, and 16. See, e.g., Ex. 1003, 36:46–47 (claim 1 reciting “a
memory for storing a database comprising flow-entries for previously
encountered conversational flows”). At the heart of the parties’ dispute is
the following passage from the ’099 patent:
A conversational flow, on the other hand, is the sequence of
packets that are exchanged in any direction as a result of an
activity—for instance, the running of an application on a server
as requested by a client. It is desirable to be able to identify and
classify conversational flows rather than only connection flows.
The reason for this is that some conversational flows involve
more than one connection, and some even involve more than one
exchange of packets between a client and server.
Ex. 1001, 2:37–45 (emphases added). In our Institution Decision, we
preliminarily construed “conversational flow” as a “sequence of packets that
are exchanged in any direction as a result of an activity.” Inst. Dec. 29. We
declined to include the phrases “for instance, the running of an application
on a server as requested by a client” and “some conversational flows involve
more than one connection,[11] and some even involve more than one
exchange of packets between a client and server” in the construction of
“conversational flow,” as Patent Owner had requested. Id. We explained
that those passages, because they begin with the phrases “for instance,”
“where some,” and “some,” are “merely exemplary and non-limiting.” Id.

11 The ’099 patent expressly defines “connection flow”: “The term
‘connection flow’ is commonly used to describe all the packets involved
with a single connection.” Ex. 1001, 2:35–37. Neither party disputes this
definition of “connection flow.”
IPR2020-00337
Patent 6,771,646 B1
15
Petitioner does not challenge our preliminary construction. Tr. 9:20–
10:15. Patent Owner, however, argues that our construction captures only a
“part of the definition in the specification,” and improperly “excludes some
of the clarifying language that is also part of the explicit definition in the
specification.” PO Resp. 23. According to Patent Owner, “[t]hat clarifying
language is important for understanding the nature of a conversational flow.”
Id.
Specifically, Patent Owner argues that the written description of the
’646 patent provides an express definition of the term “conversational flow”
and that the express definition controls even though it contains exemplary
language. Id. at 23–24 (citing Sinorgchem Co., Shandong v. Int’l Trade
Comm’n, 511 F.3d 1132, 1137 (Fed. Cir. 2007)). Patent Owner argues that
the language following “for instance” is necessary for understanding a
“conversational flow,” because that language “makes apparent that
‘conversational flow’ must relate to a conversation” and “involves an
application activity involving the same client.” Id. at 26 (emphasis added);
see also Sur-Reply 1 (arguing that an “‘activity’ occurs between specific
entities”); PO Br. 1 (arguing that an activity “must relate to the actions of a
particular client or end user”). “Ignoring that language,” Patent Owner
argues, “risks losing the basic and fundamental requirement of a
‘conversation.’” PO Resp. 26
Patent Owner also argues our construction is incorrect because, prior
to this inter partes review, “every tribunal to have considered the proper
construction for ‘conversational flow’ has accepted . . . the same
construction advanced by Patent Owner” here:
the sequence of packets that are exchanged in any direction as a
result of an activity—for instance, the running of an application
IPR2020-00337
Patent 6,771,646 B1
16
on a server as requested by a client—and where some
conversational flows involved more than one connection, and
some even involve more than one exchange of packets between
a client and server.
PO Resp. 24–25. As an example, Patent Owner argues that the District
Court for the Eastern District of Texas adopted this construction in Packet
Intelligence LLC v. NetScout Systems, Inc., and that the Court of Appeals for
the Federal Circuit “addressed conversational flows and connection flows
when affirming the jury’s infringement verdict in NetScout, which
necessarily hinged on applying the court’s claim construction of
‘conversational flow.’” Id. at 25 (citing Ex. 2060, 3).
3. Analysis
Independent claims 1 and 7 recite a database that includes “flow-
entries” for “previously encountered conversational flows.” Ex. 1003,
36:46–47 (claim 1), 38:7–9 (claim 7). Independent claim 16 recites a
“database comprising none or more flow-entries for previously encountered
conversational flows.” See id. at 39:18–20. As explained above, the parties
agree that “conversational flow” includes “the sequence of packets that are
exchanged in any direction as a result of an activity,” but disagree about
whether the phrases “for instance, the running of an application on a server
as requested by a client” and “some conversational flows involve more than
one connection, and some even involve more than one exchange of packets
between a client and server” should limit the meaning of “conversational
flow.” Patent Owner argues that those phrases are necessary because they
make clear that a conversational flow is client-specific. Thus, the parties’
basic dispute is whether a conversational flow is limited to a specific client
or user, as Patent Owner argues, or may include multiple clients or users, as
Petitioner contends. Compare, e.g., PO Resp. 26 (arguing that “the basic
IPR2020-00337
Patent 6,771,646 B1
17
and fundamental requirement” of a “conversational flow” is that it “involves
an application activity involving the same client”), with Reply 3 (contending
that “[t]he definitional language of ‘conversational flow’ doesn’t contain any
requirement for identifying flows based on a particular user or client”).
Starting with the passage reproduced above that we identified as the
heart of the parties’ dispute (Ex. 1001, 2:37–45), we determine that the “for
instance” phrase is exemplary and does not limit a “conversational flow” to
a specific client or user. The phrase “for instance” is synonymous with “for
example,” and necessarily indicates that “the running of an application on a
server as requested by a client” exemplifies, but does not limit, the
conversational flow to a server and a specific client or user.
Our determination that the “for instance” phrase is exemplary and
non-limiting is consistent with Patent Owner’s argument in its Preliminary
Response that the “some” phrase (“some conversational flows involve more
than one connection, and some even involve more than one exchange of
packets between a client and server”) is non-limiting. Specifically, Patent
Owner argued that the use of the word “some” necessarily means that “some
conversational flows involve multiple connections, while, some involve a
single connection.” Prelim. Resp. 24. Similarly here, while “conversational
flow” certainly may involve the running of an application on a server as
requested by a single client or user, “conversational flow” is not limited to a
single client or user.
Indeed, the ’099 patent and the provisional application expressly
contemplate classifying connection flows from different clients into the
same conversational flow when those connections involve the same activity.
These documents describe, in the context of identifying packets as part of a
conversational flow, a client/server protocol known as Service Advertising
IPR2020-00337
Patent 6,771,646 B1
18
Protocol (SAP), which is “used to identify the services and addresses of
servers attached to a network.” Ex. 1016, 3:12–14; Ex. 1001, 2:49–52. The
provisional application describes a packet exchange between a client and the
server in which the client sends a SAP request to a server for print service,
and the server sends a SAP reply that identifies the print service address:
In a first exchange, a client sends a SAP request to a server, for
example, for print service. The server sends a SAP reply that
identifies a particular address, for example, SAP #5, as the print
service on that server. Such may be responses used to update a
table, for example in a router, known as the Server Information
Table.
Ex. 1016, 3:14–18; see also Ex. 1001, 2:53–58. The provisional application
then describes a client who has “inadvertently seen” the SAP reply, and
therefore does not need to make a SAP request, but may send a print request
directly to SAP #5:
A client who has inadvertently seen this reply or who has access
to the table (via the router that has the Server Information Table,
for example) would know that SAP #5 for such this [sic] server
is a print service. Therefore, in order to print data on the server,
such a client does not need to make the request for a print service,
but simply to send data to be printed specifying SAP #5.
Ex. 1016, 3:18–23; see also Ex. 1001, 2:58–64.
The provisional application explains that the packet exchange between
the client who has “inadvertently seen” the print service address and SAP #5
is not connected to the initial packet exchange, “which was with a different
client”:
This sending of data to be printed again involves an exchange of
data between a client and a server, disjoint [sic] from the previous
exchange which was with a different client setting up that
SAP #5 is a print service on this server is a second connection.
IPR2020-00337
Patent 6,771,646 B1
19
Ex. 1016, 3:23–25 (emphasis added). Similarly, the ’099 patent describes
this packet exchange as “independent of the initial exchange.” Ex. 1001,
2:64–67.
The provisional application states that “[i]t is desirable for a network
packet monitor to be able to ‘virtually concatenate’ the first exchange that
defines SAP #5 as the print service on the server with the second exchange
that uses the print service.” Ex. 1016, 3:25–28. In this case, the provisional
application explains, the two packet exchanges (the first between a client
and the server and the second between a client and SAP #5) “would then be
correctly identified as being part of the same flow if the clients were the
same,” but also, “[t]hey would even be recognized if the clients were not the
same.” Id. at 3:28–30 (emphasis added). Similarly, the ’099 patent states
that “because one features [sic] of the invention is to correctly identify the
second exchange as being associated with a print service on that server, such
exchange would even be recognized if the clients were not the same.”
Ex. 1001, 3:44–48 (emphasis added).
We interpret the SAP description in the provisional application and
’099 patent as describing two seemingly disjointed packet exchanges
involving two different clients or users (the first packet exchange between a
client and the server and the second packet exchange between the client who
has “inadvertently seen” the print service address and SAP #5) as belonging
to the same “conversational flow.” And because Patent Owner’s proposed
construction—that a “conversational flow” is client-specific—is inconsistent
with the written description, we decline to limit a “conversational flow” to a
specific client or user.
Patent Owner argues that our interpretation of the SAP example is
incorrect because “[t]he second sentence” of that example—i.e., “such
IPR2020-00337
Patent 6,771,646 B1
20
exchange would even be recognized if the clients were not the same”—
merely “illustrates how a later print request by a different client can also be
recognized as a print request activity because it uses the now known address
of the print service.” Sur-reply 7 (emphasis added). We are not persuaded.
In the provisional application, this description of the second client follows
the statement that “[i]t is desirable to be able to identify and classify
conversational flows,” Ex. 1016, 3:7–8 (emphasis added), and is followed by
statements describing “[o]ther protocols that are similar in that they may
lead to disjointed conversational flows,” id. at 4:1–5 (emphasis added). The
provisional application also states that “[p]rior art network monitors do not
presently have the ability to recognize such disjointed flows as belonging to
the same conversational flow.” Id. at 4:13–14. Similarly, in the ’099 patent,
the phrase “such exchange would even be recognized if the clients were not
the same” is soon followed by the statement that “[w]hat distinguishes this
invention from prior art network monitors is that it has the ability to
recognize disjointed flows as belonging to the same conversational flow.”
Ex. 1001, 3:48–51 (emphasis added). We view these statements, in context,
as reinforcing the notion that the disclosed invention sought to classify
disjointed connection flows—even those involving more than one client or
user—into the same conversational flow based on a specific activity, such as
print service.
We are not persuaded by Patent Owner’s argument that Sinorgchem
compels a different result here. See PO Resp. 24. In Sinorgchem, the claim
language at issue was “controlled amount,” of which the patent stated:
A ‘controlled amount’ of protic material is an amount up to that
which inhibits the reaction of aniline with nitrobenzene, e.g., up
to about 4% H2O based on the volume of the reaction mixture
when aniline is utilized as the solvent.
IPR2020-00337
Patent 6,771,646 B1
21
511 F.3d at 1136. The International Trade Commission “agreed that the
patentee had expressly defined the term ‘controlled amount’ in the
specification but held that the language ‘e.g., up to about 4% H2O based on
the volume of the reaction mixture when aniline is utilized as the solvent’
should not be considered part of that definition,” in part because the
Commission considered “the 4% limit as merely an example that did not
apply to all situations in which aniline was used as the solvent.” Id. at 1137.
The Federal Circuit rejected the Commission’s construction, holding that
“[w]hen aniline is used as the solvent, the express definition is neither
ambiguous nor incomplete—the ‘controlled amount’ is ‘up to about 4% H2O
based on the volume of the reaction mixture’—and we need look no further
for its meaning.” Id. at 1138.
Patent Owner argues that Sinorgchem stands for the proposition that
an express definition in the specification controls even if it contains
exemplary language. PO Resp. 24. According to Patent Owner, the phrase
“for instance, the running of an application on a server as requested by a
client” is the express definition of “conversational flow” that we must
accept, even though it contains the exemplary language “for instance.” Id.
We do not agree because, in Sinorgchem, the patent at issue described “at
least six different solvents,” of which aniline was just one example. The
Federal Circuit reasoned that, when aniline was used as the solvent, then the
upper limit of protic material was “up to about 4% H2O.” Sinorgchem, 511
F.3d at 1138. In other words, the use of aniline was a condition precedent
(“when aniline is utilized”) that, when satisfied, triggered the condition
subsequent upper limit of protic material (“up to about 4% H2O”). Id. The
language in the ’099 patent (“for instance, the running of an application on a
server as requested by a client”)—unlike that in Sinorgchem—is not
IPR2020-00337
Patent 6,771,646 B1
22
conditional. Thus, we find Sinorgchem to be inapposite to the facts of this
case.
Finally, Patent Owner’s argument that “[b]efore the Institution
Decisions in this and the related IPRs, every tribunal to have considered the
proper construction for ‘conversational flow’ has accepted . . . the same
construction advanced by Patent Owner,” PO Resp. 24–25, does not
persuade us that our construction here is erroneous.
Despite Patent Owner’s argument, neither the District Court nor the
Federal Circuit appears to have expressly analyzed “conversational flow”
and determined that the claim term necessarily includes the language recited
within the “for instance,” “where some,” and “some” phrases. In Packet
Intelligence LLC v. NetScout Systems, for example, the District Court merely
adopted Patent Owner’s construction without analysis after the parties
“reached agreement” at a hearing dated March 2, 2017. Ex. 1067, 6. And
the Federal Circuit, if anything, appears to have relied on a definition of
“conversational flow” lacking the additional phrases Patent Owner advances
here. Ex. 2060, 3. Specifically, in describing “conversational flows,” the
court stated that:
The specifications explain that it is more useful to identify and
classify “conversational flows,” defined as “the sequence of
packets that are exchanged in any direction as a result of an
activity.”
Id. (citing Ex. 1001, 2:45–47) (emphasis added). Finally, we also observe
that the Board’s constructions of “conversational flow” in prior proceedings
were made in institution decisions and, thus, were merely preliminary. See,
e.g., Ex. 1056, 10 (interpreting “conversational flow” “for purposes of this
Decision” (emphasis added)).
IPR2020-00337
Patent 6,771,646 B1
23
For these reasons, we have considered the constructions of the other
panels and tribunals, but are not persuaded that our construction of
“conversational flow” is mistaken or is inconsistent with those other
constructions. See 37 C.F.R. § 42.100 (b) (“Any prior claim construction
determination concerning a term of the claim in a civil action, or a
proceeding before the International Trade Commission, that is timely made
of record in the inter partes review proceeding will be considered.”
(emphasis added)).
4. Summary
In summary, having considered the totality of the arguments and
evidence anew at the close of trial, we determine that “conversational
flow[s]” means a “sequence of packets that are exchanged in any direction
as a result of an activity,” without the additional restriction that the
conversational flow must be client- or user-specific.
D. The Prior Art
Before turning to Petitioner’s asserted grounds of unpatentability, we
provide a brief summary of the asserted references.
1. Riddle (Ex. 1008)
Riddle relates to a method for automatically classifying packet flows
for use in allocating bandwidth resources by a rule of assignment of a
service level. Ex. 1008, 4:6–10. The method comprises applying individual
instances of traffic classification paradigms to packet network flows based
on selectable information obtaining from layers of a multi-layered
communication protocol in order to define a characteristic class, then
mapping the flow to the defined traffic class. Id. at 4:10–15. Figure 3 is
reproduced below.
IPR2020-00337
Patent 6,771,646 B1
24

Figure 3 illustrates a system for automatically classifying traffic.
Ex. 1008, 12:27–28.
With reference to Figure 3, traffic tree 302 classifies new traffic under
a particular member class node. Ex. 1008, 12:28–30. Traffic classifier 304
detects services for incoming traffic. Id. at 12:30–31. Knowledge base 306
contains heuristics for determining traffic classes. Id. at 12:32–33. A
plurality of saved lists 308 stores classified traffic pending incorporation into
traffic tree 302. Id. at 12:37–38.
Figure 4A is reproduced below.
IPR2020-00337
Patent 6,771,646 B1
25

Figure 4A illustrates a flowchart 401 of processing steps for
automatically classifying traffic. Ex. 1008, 12:42–43.
In step 402 shown in Figure 4A, a flow specification is parsed from
the flow being classified. Ex. 1008, 12:43–44. Then in step 404, the flow
specification parsed from the flow in step 402 is compared with the traffic
specifications in each node of the classification tree. Id. at 12:44–47. In
decisional step 406, a determination is made of whether traffic matches one
of the classes being classified. Id. at 12:48–50. If this is so, then in
step 408, an entry is made in a list of identifying characteristics, such as
protocol type, IP protocol number, server port, traffic type, MIME type, or
time of occurrence of traffic. Id. at 12:50–53.
Figure 4B is reproduced below.
IPR2020-00337
Patent 6,771,646 B1
26

Figure 4B illustrates a flowchart 403 of the processing steps for
integrating traffic classes into a classification tree. Ex. 1008,
13:36–38.
In step 420 of Figure 4B, an instance of saved traffic is received from
a saved traffic list 308. Ex. 1008, 13:40–42. Next, in decisional step 422,
the instance of saved traffic is examined to determine whether it is
well-known and a name representing its type exists. Id. at 13:42–45. If this
is so, then processing continues with a test of whether the saved traffic
belongs to a service aggregate in step 426. Id. at 13:45–47. Otherwise, in
step 423, the instance of saved traffic is examined to determine whether it
IPR2020-00337
Patent 6,771,646 B1
27
appears to be a server connection port of an unregistered IP port. Id.
at 13:47–50. If this is not so, then processing continues with the next traffic
class in the saved list in step 420. Id. at 13:51–52. In decisional step 426,
the instance of saved traffic is examined to determine whether it belongs to a
service aggregate. Id. at 13:52–54. If the traffic does belong to a service
aggregate, then, in step 428, a traffic class is created which will match all
components of the service aggregate. Id. at 13:57–59. In further step 425, a
new traffic class is created to match the instance of saved traffic. Id.
at 13:59–62.
2. Ferdinand (Ex. 1009)
Ferdinand relates to a system for “monitoring and managing
communication networks for computers.” Ex. 1009, code (54), 1:3–4.
Ferdinand discloses a monitoring system with “a Network Monitor 10 and a
Management Workstation 12.” Id. at 11:32–12:1. In monitoring the
network, Ferdinand indicates that a “statistical object represents a network
parameter for which performance information is gathered,” and that
Monitor 10 keeps information about monitored statistical objects in
“Statistics Module (STATS) 36.” Id. at 22:18–22. STATS 36 is a database
(id. at 19:5–11) and “defines the database and it contains subroutines for
updating the statistics which it keeps” (id. at 28:14–15). Examples of data
the database stores include records “per ip address,” “per ip pair,” “per udp
pair,” “per ftp control connection,” and “per ftp data connection.” Id.
at 29:3–30:7.
3. Wakeman (Ex. 1014)
Wakeman, titled “Forwarding Database Cache for Integrated Switch
Controller,” describes a local access network (LAN) network switch that
includes a random access memory (RAM) forwarding database (FDB)
IPR2020-00337
Patent 6,771,646 B1
28
containing address-to-port mappings for all devices connected to the
switch’s ports, as well as at least one CAM-cache connected to one or more
of the switch’s ports. Ex. 1014, codes (54), (57). By way of background,
Wakeman explains that LAN network switches typically include a switch
engine (SE), an FDB, and one or more dozens of ports, where the FDB may
be implemented either as a hardware CAM or as a RAM. Id. at 1:18–23,
1:55–56. According to Wakeman, a hardware CAM is “very fast,” but “can
be prohibitively expensive,” whereas RAM “can be implemented at a
fraction of the cost of such hardware CAM” but is “typically too slow to
keep up” with a network switch’s SE. Id. at 1:56–67. By including both a
RAM FDB and a CAM-cache having an access time much faster than that of
the FDB, Wakeman’s switch purportedly overcomes the problems in the
prior art. Id. at 2:22–30; see also id. at 2:15–18 (citing a need for a network
switch that is “not confined by the rigid balancing between the superior
performance of a CAM database and the cost savings of a RAM database”).
4. Yu (Ex. 1011)
Yu relates to “an architecture 100 for applying policies to network
data traffic.” Ex. 1011, 2:46–50. The architecture includes three
components: (i) an application “such as a firewall, virtual private network
(VPN), or traffic management,” (ii) a “policy engine,” and (iii) an API
between these two components. Id. at 2:51–62.
Yu describes a “flow classification specification 203a provides the
screening criteria for the flow classifier logic 204 to sort network traffic into
flows,” that “[a]ll packets that match the same flow classification
specification 203a form a flow,” that “a flow is a stream of correlated
packets to which policy decisions apply,” and that “a flow classifier 204
classifies the packet according to one or more classification
IPR2020-00337
Patent 6,771,646 B1
29
specifications 203a and finds one or more corresponding action
specifications 203b.” Ex. 1011, 3:32–59. Yu further describes that a
“stream is an ‘instantiation’ of a flow-packets that have the same source and
destination address, source and destination port, and protocol type,” and
“[p]ackets may be sorted into streams, and a flow may include one or more
streams,” where “[a]ll packets belonging to the same stream are to be
regulated by the same policy.” Id. at 4:2–9.
In Yu, “the stream classifier 207 matches the packets to a particular
stream specification 208 and then, using the corresponding action
specifications 210, activates the proper action processors 206.” Ex. 1011,
5:8–11.
5. RFC1945 (Ex. 1010)
RFC1945 is an informational specification, published by the Network
Working Group, and concerns Hypertext Transfer Protocol 1.0. Ex. 1010, 1.
The publication “reflects common usage of the protocol referred to as
‘HTTP/1.0’.” Id. The RFC defines several message headers, such as
“Referer,” which “allows the client to specify, for the server’s benefit, the
address (URI) of the resource from which the Request-URI was obtained,”
and “Server,” which “contains information about the software used by the
origin server to handle the request.” Id. at 44–45.
E. Obviousness over Riddle, Ferdinand, and Wakeman
Petitioner contends that claims 1–3, 7, 16, and 18 would have been
obvious over the combination of Riddle, Ferdinand, and Wakeman. Pet. 17–
80. Patent Owner opposes. PO Resp. 34–46, 53–57; Sur-reply 13–16, 18–
22. Having considered the totality of the arguments and evidence, we find
that Petitioner has shown by a preponderance of the evidence that claims 1,
2, 7, 16, and 18 are unpatentable as having been obvious over Riddle in view
IPR2020-00337
Patent 6,771,646 B1
30
of Ferdinand, but that Petitioner has not shown that claim 3 is unpatentable
on this ground.
1. Independent Claims 1, 7, and 16
a. Preamble
The preambles of claims 1 and 7 of the ’646 patent each recite “[a]
packet monitor for examining packets passing through a connection point on
a computer network, each packet conforming to one or more protocols.”
Ex. 1003, 36:39–41, 37:61–63, 45 (Certificate of Correction dated Sept. 21,
2004). The preamble of claim 16 similarly recites “[a] method of examining
packets passing through a connection point on a computer network, each
packets [sic] conforming to one or more protocols.” Id. at 39:10–12.
To the extent that the preamble recitations are limiting, Petitioner
persuasively shows that they are taught by Riddle. See Pet. 30–33 (citing
Ex. 1006 ¶¶ 263–269, 617–623, 686, 687). In particular, Petitioner contends
that Riddle’s classifier operates in a network-connected computer system
(e.g., in a server acting as a packet monitor and a network interface acting as
a packet acquisition device) and parses and examines (or analyzes) traffic
flow packets passing through a network. Id. at 30–31 (citing Ex. 1008,
code (57), 1:57–61, 4:6–17, 5:53–67, 7:21–24, 12:27–41, 14:22–40, Fig. 3;
Ex. 1006 ¶¶ 617, 618). Regarding “connection point[s],” Petitioner
contends, “the ’646 [p]atent acknowledges the packet monitor connects to
the network at these points,” and “Riddle’s packet monitor connects to [a]
network connection . . . via a system gateway.” Id. at 32 (citing Ex. 1003,
4:54–5:8, Fig. 1; Ex. 1008, 5:53–67, 6:9–15, 7:21–24, Figs. 1A, 1B;
Ex. 1006 ¶ 619). Further, Petitioner contends, “[f]or networks connecting
multiple clients and servers . . . , Riddle teaches examining packets via
‘network routing means’ and/or routers,” which we agree with Petitioner
IPR2020-00337
Patent 6,771,646 B1
31
also serve as “connection point[s]” as claimed. Id. (citing Ex. 1008, 7:10–
34, Figs. 1C, 3; Ex. 1006 ¶ 619). Still further, Petitioner contends that
“Riddle discloses ‘each packet conforming to one or more protocols’
because it ‘relates to digital packet telecommunications, and particularly to
management of network bandwidth based on information ascertainable from
multiple layers of OSI network model,’” and that, “[f]or packet
transmissions, Riddle’s Figure 1D shows ISO network model diagramming
the relationship between layers of the TCP/IP protocol suite (e.g., the
application, transport, network, data link, and physical layers (80–88)).” Id.
at 33 (citing Ex. 1008, 1:54–57, 7:35–8:46, Fig. 1D; Ex. 1006 ¶¶ 620–623).
Patent Owner does not substantively address these assertions or
otherwise contend that the art lacks any of the preamble features, and we are
persuaded that Petitioner has shown by a preponderance of the evidence that
they are taught by Riddle and adopt the contentions set forth in the Petition
and in Dr. Weissman’s Declaration as mapped to these undisputed
limitations of claims 1, 7, and 16 as our own findings. See In re NuVasive,
Inc., 841 F.3d 966, 974 (Fed. Cir. 2016) (explaining that the Board need not
make specific findings about claim limitations that a patent owner does not
dispute are disclosed in the prior art).
b. Packet Acquisition Device Limitations (Claims 1, 7,
and 16)
Claims 1 and 7 of the ’646 patent each recite, in part, “a packet
acquisition device coupled to the connection point and configured to receive
packets passing through the connection point.” Ex. 1003, 36:43–45, 37:65–
67. Claim 16 similarly recites “receiving a packet from a packet acquisition
device.” Id. at 39:14.
IPR2020-00337
Patent 6,771,646 B1
32
Petitioner demonstrates by a preponderance of the evidence that
Riddle teaches these limitations. See Pet. 34–36 (citing Ex. 1006 ¶¶ 263–
269, 624–626, 688), 77. In particular, Petitioner identifies various “packet
acquisition devices coupled to a connection point” in Riddle, including
network interface 40 connected to the system gateway depicted in Riddle’s
Figures 1A and 1B, as well as the “network routing means” and router 75
depicted in Riddle’s Figure 1C. Id. at 34 (citing Ex. 1008, 6:1–23, 7:21–24,
16:54–17:15 (claim 8), Figs. 1A–1C; Ex. 1006 ¶¶ 264–266). Citing
disclosure in Riddle describing automatically classifying packet flows to
help allocate bandwidth resources and “to classify a complete enumeration
of the possible traffic,” and further relying on the testimony of
Dr. Weissman, Petitioner contends that a person of ordinary skill in the art
“would have understood that Riddle’s packet acquisition device is
configured to receive packets.” Id. at 35–36 (citing Ex. 1008, code (57),
4:7–17, 4:55–60; Ex. 1006 ¶ 625). Patent Owner does not dispute
Petitioner’s contentions regarding these limitations. See generally PO Resp.
Petitioner’s assertions are supported by the cited evidence and are
persuasive. Accordingly, we are persuaded that Petitioner has shown by a
preponderance of the evidence that the recited limitations are taught by
Riddle and adopt the contentions set forth in the Petition and in
Dr. Weissman’s Declaration as mapped to these undisputed limitations of
claims 1, 7, and 16 as our own findings. NuVasive, 841 F.3d at 974.
c. Input Buffer Memory Limitation (Claim 7)
Claim 7 of the ’646 patent recites, in part, “an input buffer memory
coupled to and configured to accept a packet from the packet acquisition
device.” Ex. 1003, 38:1–2.
IPR2020-00337
Patent 6,771,646 B1
33
Relying on the testimony of Dr. Weissman, Petitioner contends, and
we agree, that Riddle discloses this limitation or renders it obvious in view
of Ferdinand, contending that a person of ordinary skill in the art would have
understood that packet queues in Riddle’s router function as buffers, or,
alternatively, that a person of ordinary skill in the art would have been
motivated and found it obvious to include an input buffer memory in
Riddle’s memory storage based either on the artisan’s own knowledge or
Ferdinand’s teachings. Pet. 36–37 (citing Ex. 1008, code (57), 2:51–55,
4:15–17, 6:1–23, 7:21–24, 16:54–17:15 (claim 8), Figs. 1A, 1B; Ex. 1003,
18:41–51; Ex. 1009, 26:2–7, 41:17–31, 49:2–12; Ex. 1006 ¶¶ 270–276, 625,
689–691; Ex. 1031, 1:54–2:3).
Petitioner further contends that the person of ordinary skill in the art
would have been motivated, based on Ferdinand’s teachings, to “modify
Riddle’s monitor with input buffer memory to temporarily store received
packets and improve performance by limiting packet drops,” and that
including such input buffer memory in a packet acquisition device in
accordance with Riddle’s and Ferdinand’s teachings “amounts to nothing
more than combining known prior-art technologies used in their ordinary
and predictable manner to queue packet traffic.” Id. at 37–38 (citing
Ex. 1006 ¶ 691). With respect to reasons to combine the teachings of Riddle
and Ferdinand, Petitioner contends that “Riddle and Ferdinand are in the
same field of endeavor,” “contain overlapping disclosures with similar
purposes,” and illustrate that it was well-known and ubiquitous in the art for
networking devices to include database storage structures, such as buffers.
Id. at 25 (citing Ex. 1008, 4:7–17, 9:14–27, 12:37–59, Figs. 3, 4A, 4B;
Ex. 1009, 12:3–9; Ex. 1006 ¶¶ 256, 257).
IPR2020-00337
Patent 6,771,646 B1
34
Patent Owner does not dispute Petitioner’s contentions regarding this
limitation. See generally PO Resp.
Petitioner’s assertions are supported by the cited evidence and are
persuasive. Moreover, we are persuaded that a person of ordinary skill in
the art would have been motivated to combine the teachings of Riddle and
Ferdinand as proposed by Petitioner and supported by the testimony of
Dr. Weissman. See Pet. 25–27 (citing Ex. 1008 ¶¶ 256–257, 259).
Accordingly, we are persuaded that Petitioner has shown by a preponderance
of the evidence that the recited limitations are taught by the combination of
Riddle and Ferdinand and adopt the contentions set forth in the Petition and
in Dr. Weissman’s Declaration as mapped to these undisputed limitations of
claim 7 as our own findings. NuVasive, 841 F.3d at 974.
d. Parser Record Limitations (Claims 7 and 16)
Claim 7 of the ’646 patent recites, in part, “a parser subsystem
coupled to the input buffer memory, the parsing subsystem configured to
extract selected portions of the accepted packet and to output a parser record
containing the selected portions.” Ex. 1003, 38:3–6. Claim 16 similarly
recites “performing one or more parsing/extraction operations on the packet
to create a parser record comprising a function of selected portions of the
packet.” Id. at 39:15–17.
Relying on disclosure in Riddle of a processor programmed to
perform parsing/extraction operations and coupled to memory, Petitioner
persuasively shows that Riddle discloses a parser subsystem, as recited in
claim 7, as well as the corresponding parsing/extraction step of claim 16.
Pet. 38 (citing Ex. 1006 ¶¶ 294–327, 691–700; Ex. 1008, 6:1–15, 16:61–62
(claim 8), Fig. 1A), 77. Specifically, Riddle teaches “a processor means
operative to: parse a packet into a first flow specification” and illustrates in
IPR2020-00337
Patent 6,771,646 B1
35
Figure 1A that processor 30 is coupled to memory subsystem 35. Ex. 1008,
16:61–62 (claim 8), 6:1–15, Fig. 1A. Riddle teaches that processor 30
comprises a parser subsystem that applies “individual instances of traffic
classification paradigms to packet network flows based on selectable
information obtained from a plurality of layers of a multi-layered
communication protocol in order to define a characteristic class, then
mapping the flow to the defined traffic class.” Id. at 4:10–15. And Riddle
teaches that the parser extracts identifying characteristics (i.e., a signature
for the flow) such as patterns and/or reference strings from headers. Id.
at 8:67–9:27, 12:42–59. Thus, we agree with Petitioner that Riddle teaches a
parser subsystem coupled to memory that extracts packet portions. Pet. 39–
40 (citing Ex. 1008, 4:10–15, 9:28–42, 12:26–41, 12:42–53, 15:55–16:14
(claim 1), 17:21–18:16 (claim 11), Fig. 4A; Ex. 1006 ¶¶ 693–696). Further,
Petitioner persuasively shows that Riddle discloses outputting parser records
containing the selected portions in the same way as disclosed by the
’646 patent. Id. at 41 (citing Ex. 1003, 9:29–39; Ex. 1008, 8:67–9:27,
12:42–59; Ex. 1006 ¶¶ 697, 698). Patent Owner does not dispute
Petitioner’s contentions regarding these limitations. See generally PO Resp.
Petitioner’s assertions are supported by the cited evidence and are
persuasive. Accordingly, we are persuaded that Petitioner has shown by a
preponderance of the evidence that these limitations are taught by Riddle
and adopt the contentions set forth in the Petition and in Dr. Weissman’s
Declaration as mapped to these undisputed limitations of claims 7 and 16 as
our own findings. NuVasive, 841 F.3d at 974.
e. Flow-Entry Database Limitations (Claims 1, 7, and 16)
Claim 1 of the ’646 patent recites, in part, “a memory for storing a
database comprising flow-entries for previously encountered conversational
IPR2020-00337
Patent 6,771,646 B1
36
flows to which a received packet may belong, a conversational flow being an
exchange of one or more packets in any direction as a result of an activity
corresponding to the flow.” Ex. 1003, 36:46–51. Claim 7 similarly recites
“a memory for storing a database of one or more flow-entries for any
previously encountered conversational flows, each flow-entry identified by
identifying information stored in the flow-entry;” and claim 16 recites
“looking up a flow-entry database comprising none or more flow-entries for
previously encountered conversational flows, the looking up using at least
some of the selected packet portions and determining if the packet is of an
existing flow, the lookup being via a cache.” Id. at 38:7–10, 39:18–22, 47
(Certificate of Correction dated Oct. 15, 2013).
(1) The Parties’ Arguments
Petitioner contends, and we agree, that Riddle alone or in view of
Ferdinand renders these limitations obvious. Pet. 42–57, 77–78. More
particularly, with respect to the recited “memory for storing . . . flow-
entries” limitation, Petitioner contends that Riddle’s monitor includes a
storage subsystem and stores flow-entry lists of previously encountered
flows in a series of lists that include a plurality of flow-entries encountered
by the monitor. Pet. 42–43 (citing Ex. 1006 ¶¶ 328–339, 628; Ex. 1008,
6:1–23, 6:43–50, 12:37–59, Figs. 1A, 1B, 3). According to Petitioner,
Riddle incorporates the saved lists into a classification tree, where each node
of the tree represents a traffic class. Id. at 43 (citing Ex. 1008, 8:47–50,
9:28–33, Fig. 3). Relying on the testimony of Dr. Weissman, Petitioner
contends that a person of ordinary skill in the art would have understood that
Riddle’s saved lists “store flow-entries in memory.” Id. at 44 (citing
Ex. 1006 ¶¶ 330–334, 628–630).
IPR2020-00337
Patent 6,771,646 B1
37
Further, Petitioner contends, “Riddle identifies a ‘service aggregate’
which is one type of traffic.” Pet. 45 (citing Ex. 1006 ¶ 331; Ex. 1008,
11:10–22, 13:53–59). Petitioner argues that Riddle’s “service aggregate
links together into a ‘conversation’ multiple connection flows based on
specific software program activity (e.g., PointCast traffic),” and,
“[a]ccordingly, Riddle teaches storing separate entries for encountered
conversational flows.” Id. (citing Ex. 1006 ¶ 331; Ex. 1008, 11:60–63).
Additionally, Petitioner contends, Riddle illustrates processing of the flow-
entries to determine whether a traffic class, such as a service aggregate,
needs to be created for the flow, and that Riddle’s monitor retrieves
previously stored data from the saved lists, tests whether the retrieved traffic
belongs to a service aggregate, and if so, creates a traffic class that “will
match all components of the service aggregate.” Id. (citing Ex. 1006 ¶¶ 332,
333, 628; Ex. 1008, 4:49–51, 13:35–14:6, Fig. 4B). According to Petitioner,
a person of ordinary skill in the art would have been motivated and would
have found it obvious to store Riddle’s lists and related tree in a flow-entry
database, based either on the artisan’s own knowledge of network devices or
on Ferdinand’s teachings, in order, for example, to increase functionalities in
furtherance of Riddle’s desired goal of determining whether the packet
monitor has received duplicate flow-entries. Id. at 46–48 (citing, e.g.,
Ex. 1006 ¶¶ 335–339, 631; Ex. 1008, 6:1–15, 12:32–35, 12:53–57, 15:1–15,
Fig. 4A; Ex. 1009, 28:14–17).
Petitioner further contends that Riddle discloses flow-entries of
“previously encountered conversational flows” in at least two ways:
“(a) classifying based on service aggregates and (b) classifying based on
PointCast.” Pet. 49 (citing Ex. 1006 ¶¶ 297–327, 632–635). Regarding the
former, Petitioner contends that “Riddle teaches identifying whether packets
IPR2020-00337
Patent 6,771,646 B1
38
are part of ‘service aggregates,’ i.e., traffic classes linking separate
connection flows based on the associated application,” and that “these
‘service aggregates’ meet the claimed ‘conversational flow.’” Id. (citing
Ex. 1006 ¶¶ 303–315, 632). Indeed, Petitioner contends, Riddle’s claims 1
and 2 teach that the service aggregates are conversational flows. Id. at 53–
54 (citing Ex. 1008, 15:56–16:14 (claim 1) (reciting “said network having
any number of flows” and “parsing a packet into a first flow specification”),
16:15–26 (claim 2) (reciting “for at least a second flow having a second flow
specification, recognizing said second flow specification and said first flow
specification to comprise together a service aggregate” and “associating said
first flow specification and said second flow specification with a newly-
created classification tree node, said newly-created classification tree type
node having a first traffic specification corresponding to said first flow
specification and a second traffic specification corresponding to said second
flow specification”); Ex. 1006 ¶ 313).
Regarding “PointCast Traffic,” Petitioner contends that the
provisional application, from which the ’646 patent claims priority and
which is incorporated by reference in the ’646 patent, expressly “specifies
that PointCast traffic flows include an identification signature, and that
identifying PointCast traffic is an example of identifying a conversational
flow.” Pet. 55 (citing Ex. 1016, 7:16–25; Ex. 1006 ¶¶ 320–322). Because
Riddle “creates a single traffic class for disjointed PointCast connection
flows by searching headers for URLs that begin with ‘/FIDO-1/,’” Petitioner
contends, “Riddle thus teaches identifying a conversational flow.” Id. at 55–
56 (citing Ex. 1006 ¶¶ 323, 324; Ex. 1008, 11:57–12:9). Indeed, Petitioner
asserts, one of the inventors of the ’646 patent previously testified that
creating a single flow to describe such disjointed flows is a type of
IPR2020-00337
Patent 6,771,646 B1
39
conversational flow. Id. at 56 (citing Ex. 1068, 55:11–57:15; Ex. 1071 ¶ 4;
Ex. 1072, 3). Finally, Petitioner contends, “Riddle teaches that one of its
autoclassification processes identifies PointCast traffic using the outside
service field of the class, specifying that “[c]ertain traffic may be
distinguished by a signature,” and accordingly, based on Riddle’s teachings,
a person of ordinary skill in the art “would have understood that Riddle
stores flow-entries for ‘previously encountered conversational flows” such
as PointCast traffic.’” Id. (citing Ex. 1008, 11:50–53, 14:54–63, 15:28–31;
Ex. 1006 ¶¶ 325, 328–339, 634, 636).
Lastly, Petitioner contends that Riddle teaches “each flow-entry
identified by identifying information stored in the flow-entry” as claimed,
because Riddle enters characteristics of the traffic flow into the saved list
(step 408), which identifying information includes “a flow specification and
indicators.” Id. at 57 (citing, e.g., Ex. 1008, 12:42–59, Figs. 4A–4B;
Ex. 1006 ¶¶ 701–704).
In response to Petitioner’s contentions, Patent Owner asserts, first,
that Riddle fails to teach “conversational flows” (PO Resp. 34–45), and
second, that the combination of Riddle and Ferdinand does not teach the
claimed flow-entry database (id. at 45–46). Regarding the first point, Patent
Owner contends that Riddle classifies traffic into generic traffic classes, but
does not perform any more granular analysis. Id. at 40. According to Patent
Owner, Riddle has no need or ability to recognize whether a particular group
of packets in one direction relate to another flow or may form part of a
conversational flow implicating an activity, unlike the claims of the ’646
patent “which require specific operations involving conversational flows.”
Id. Further, according to Patent Owner, “just because Riddle discloses a
traffic classification activity does not mean that it discloses or suggests the
IPR2020-00337
Patent 6,771,646 B1
40
specific activity required in the ’646 Patent, which is narrowly focused on
conversational flows.” Id. at 41–42. Patent Owner contends, “Riddle
combines flows based on the type of activity,” and “[a]t best, . . . combines
flows of the same type into a traffic class,” whereas “the ’646 Patent
inventions recognize separate conversational flows even when these
different flows are based on the same type of activity.” Id. at 42–43. Still
further, Patent Owner argues, “‘conversational flows’ include bidirectional
exchanges of packets,” but “Riddle, however, expressly teaches that its
traffic classes are limited to unidirectional traffic.” Id. at 44 (emphasis
omitted) (citing Ex. 1008, 8:50–53, 11:28–29, 13:57–59; Ex. 2061 ¶¶ 64, 65,
83). According to Patent Owner, “[b]ecause Riddle’s traffic classes are
unidirectional, they cannot satisfy the ‘conversational flow’ element required
in every challenged claim.” Id. at 44–45.
Regarding the recited flow-entry database, Patent Owner contends
that Riddle does not use conversational flows and accordingly has no
disclosure of the recited flow-entry database “comprising flow-entries for
previously encountered conversational flows.” PO Resp. 45 (citing Ex. 2061
¶ 87). Further, Patent Owner contends, although the Petition relies on the
combination of Riddle and Ferdinand as teaching this limitation,
“Ferdinand . . . focuses on a ‘technique of organizing and displaying
network performance statistics’” and “never discloses or suggests storing
‘flow-entries for previously encountered conversational flows.’” Id. (citing
Ex. 1009, 7:17–18, 8:3–5). “Nor would Ferdinand have reason to implement
such a solution,” according to Patent Owner, “because its goal is merely to
keep count,” which “[i]t can do . . . by the standard statistical measures
described in the specification.” Id. at 45–46.
IPR2020-00337
Patent 6,771,646 B1
41
In its Reply, Petitioner contends that Patent Owner’s arguments do not
analyze Riddle under the Board’s construction of “conversational flows,”
but instead under Patent Owner’s own incorrect construction. Reply 10–11.
Regardless, Petitioner argues, Riddle discloses “conversational flows” even
under Patent Owner’s incorrect construction. Id. at 11. First, according to
Petitioner, “Riddle specifies that its service aggregate ‘is provided for certain
applications that use more than one connection in a particular conversation
between a client and a server.’” Id. (citing Ex. 1008, 11:10–19). For
example, “Riddle describes listing service aggregates based on their source
and/or destination, e.g., ‘host1,’ where host1 can represent a client or a
server involved in an FTP flow.” Id. at 11–12 (citing Ex. 1008, 2:13–21,
7:12–21, 13:8–26, Fig. 1C; Ex. 1006 ¶¶ 311, 312, 632, 633). Further,
Petitioner contends, “Riddle teaches defining a traffic class by URI-pattern
(e.g., ‘/sales/*’) and by the specific client/server pair involved (e.g., from
‘client Z’ to ‘server Y’).” Id. at 12 (citing Ex. 1008, 8:58–9:11; Pet. 70–71,
94–95).
In response to Patent Owner’s contention that Riddle classifies
inbound and outbound flows only separately, not bidirectionally, Petitioner
contends that Riddle in fact discloses that all traffic in a particular
conversation between a client and a service is classified into a single service
aggregate. Reply 13 (citing Ex. 1008, 11:12–15; Pet. 49–50). Petitioner
cites, for example, disclosure in Riddle that the service aggregate for an FTP
session contains “each conversation” between the client and server for the
session, which Petitioner argues refers to all packets exchanged in either
direction. Id. (citing Ex. 1008, 11:13–19, 12:1–12, 13:54–59; Pet. 50–51).
Still further, Petitioner argues that “Riddle isn’t limited to flow classification
based on port or server identity,” but discloses, for example, that “flow
IPR2020-00337
Patent 6,771,646 B1
42
classification can use port numbers and can further ‘extend to examination
of the data contained in a flow’s packets’ via signature-based analysis.” Id.
at 14 (citing Ex. 1008, 11:50–53). Petitioner contends that “Telnet traffic’s
multi-packet ‘option negotiations’ ‘may indicate an appropriate class,’ an
example of classification from exchanged data, not port number.” Id. (citing
Ex. 1008, 11:66–67).
Lastly, in response to Patent Owner’s arguments that “Riddle . . . has
no disclosure of a flow-entry database” and that “Ferdinand never discloses
or suggests storing ‘flow-entries for previously encountered conversational
flows’” (PO Resp. 45–46), Petitioner argues, first, that Patent Owner does
not address the Petition’s contention that, “based upon a [person of ordinary
skill in the art’s] knowledge of network devices, a [person of ordinary skill
in the art] would have been motivated and found it obvious to store Riddle’s
classification information, such as flow-entry lists 308, in a database.”
Reply 18 (citing Pet. 46–47; Ex. 1006 ¶¶ 328–335, 629). Petitioner further
points out that the Petition alleges that the “flow-entry database” limitations
would have been obvious over “Riddle alone or in combination with
Ferdinand,” not based on Ferdinand alone, and accordingly, Petitioner
contends, Patent Owner’s attack on Ferdinand individually is irrelevant to
the trial ground. Id. at 18–19 (citing Pet. 23–27, 46–48). Citing the
arguments presented in the Petition, Petitioner argues that “[t]he ‘flow-entry
database’ limitations would have been obvious in view of Riddle’s teachings
regarding flow-entries and relational databases based on Ferdinand’s
teachings regarding statistics databases for storing flow classification
information.” Id. at 19 (citing Pet. 23–27, 46–48).
In the Sur-reply, Patent Owner argues again that Riddle fails to
disclose “conversational flows,” contending that Riddle does not classify
IPR2020-00337
Patent 6,771,646 B1
43
activities based on a particular client, that Riddle categorizes inbound and
outbound traffic separately, and that Riddle fails to distinguish between
different activities of the same type. Sur-reply 13–16. First, according to
Patent Owner, Petitioner’s contention that a service aggregate can
distinguish between the same type of activity between different clients
(Reply 11–12) “hinges on an extrapolation of how a single connection
example is classified that is not supported by the rest of the specification.”
Sur-reply 13. According to Patent Owner:
In Riddle, service aggregates are a “common traffic class” that
are “created with a plurality of traffic specifications.” EX1008
at 11:15–23. Riddle notes “[a] traffic class is broadly defined as
traffic between one or more clients and one or more servers.”
Id. at 8:48–49, 14:18–20 (“[A]nother method of identifying an
individual traffic class is to detect simultaneous connections to
the same host port from different clients.”). Thus, Riddle does
not classify traffic per user, but by the type of traffic based on
properties such as the ports being used. This understanding
reflects Riddle’s stated purpose—bandwidth management.
Petitioner also suggests that, because a traffic class can be
defined in terms of a specific source or destination IP
address, . . . Riddle discloses “classifying traffic into client-
specific conversational flows.” Reply at 11–12 (citing Riddle
at 8:58–9:11, 10:1–17). That is incorrect. A rule written
specifically for a client’s IP address would not distinguish one
client activity from another; it would simply aggregate all traffic
for that client. Defining a traffic class in terms of a specific
source or destination address is an implementation of the
conventional technique of matching one or more parts of a
5-tuple—but not a conversational flow. PO [Resp.] 6, 18–19;
EX2061 ¶¶62–65, 81–83.
Id. at 13–14 (first and second alterations in original). Further, Patent Owner
argues, “Riddle expressly teaches that its traffic classes are limited to
unidirectional traffic, and that service aggregates are traffic classes.” Id.
at 15. Finally, with regard to the recited “flow-entry database,” Patent
IPR2020-00337
Patent 6,771,646 B1
44
Owner responds that, notwithstanding Petitioner’s arguments that
implementing these limitations would have been obvious, “Petitioner
provided no evidence or argument to show it would have been ‘unusually
simple’ to add the claimed limitations, nor that the technology is
‘particularly straightforward.’” Id. at 20–21. According to Patent Owner,
“[t]he challenged claims do not recite a database generally, but recite a flow-
entry database specifically adapted for use in a packet monitor,” and a
person of ordinary skill in the art “would have understood the statistics
database of Ferdinand to serve a very different purpose than the claimed
flow-entry database.” Id. at 21.
(2) Analysis
Having considered the full trial record, we are persuaded that
Petitioner has shown by a preponderance of the evidence that the
combination of Riddle and Ferdinand renders obvious the recited flow-entry
database limitations in view of the proper construction of “conversational
flow.” See supra § III.C. Petitioner’s assertions, which we adopt, are
supported by the cited evidence and are persuasive. Moreover, we are
persuaded that a person of ordinary skill in the art would have been
motivated to combine the teachings of Riddle and Ferdinand as proposed by
Petitioner and supported by the testimony of Dr. Weissman. See Pet. 25–27
(citing Ex. 1006 ¶¶ 256–257, 259).
As an initial matter, for the reasons set forth in Section III.C above,
we conclude that the term “conversational flow” is not client- or user-
specific. Further, notwithstanding Patent Owner’s arguments that Riddle is
concerned with bandwidth management rather than network monitoring (see
PO Resp. 40–43; Sur-reply 13–14), obviousness does not require that the
prior art address the same problem identified by the patent. See KSR, 550
IPR2020-00337
Patent 6,771,646 B1
45
U.S. at 419 (“[N]either the particular motivation nor the avowed purpose of
the patentee controls . . . . What matters is the objective reach of the claim.
If the claim extends to what is obvious, it is invalid.”); Outdry Techs. Corp.
v. Geox S.p.A., 859 F.3d 1364, 1371 (Fed. Cir. 2017) (“The motivation
supported by the record . . . need not be the same motivation articulated in
the patent.”).
Further, contrary to Patent Owner’s contentions (PO Resp. 44;
Sur-reply 15), neither the Board’s construction of “conversational flow” nor
Patent Owner’s proposed construction requires “bidirectional” exchanges of
packets (see supra § III.C; PO Resp. 23–26). Both the Board’s adopted
construction and Patent Owner’s proposed construction recite, inter alia,
“the sequence of packets that are exchanged in any direction as a result of an
activity.” Supra § III.C (emphasis added); PO Resp. 23–26. The phrase “in
any direction” is not limited to bidirectional packet flows but also
encompasses unidirectional flows. As Petitioner points out, this
interpretation is consistent with the declaration of Patent Owner’s expert,
Ms. Quigley, stating that conversational flows “can,” rather than “must,”
include bidirectional exchanges. Ex. 2061 ¶ 83 (cited in Reply 13 n.41).
Regardless, we further agree with Petitioner that Riddle’s disclosure that all
traffic in a particular conversation between a client and a server is classified
into a single service aggregate teaches bidirectional exchange. Reply 13
(citing Ex. 1008, 11:12–19, 12:1–12, 13:54–59); see also Pet. 49–51
(mapping Riddle’s service aggregates to the claimed conversational flows
and stating that Riddle’s “Table 3 specifies detecting services in ‘both
directions’”).
Having concluded that Riddle teaches the recited conversational
flows, we are further persuaded by Petitioner’s arguments that a person of
IPR2020-00337
Patent 6,771,646 B1
46
ordinary skill in the art would have found it obvious in view of the combined
teachings of Riddle and Ferdinand to store flow-entries for previously
encountered conversational flows in a database. See Pet. 46–48. Riddle
discloses storing flow entries in its series of lists 308 that include a plurality
of flow-entries encountered by the monitor, and Petitioner persuasively
argues with the support of Dr. Weissman’s testimony that a person of
ordinary skill in the art would have been motivated, based either on Riddle’s
own disclosure of relational database 306 or on the teachings of Ferdinand,
to implement Riddle’s saved flow-entries in a database, in order, for
example, to increase functionalities in furtherance of Riddle’s desired goal
of determining whether the packet monitor has received duplicate flow-
entries. See id. (citing, e.g., Ex. 1008, 6:1–15, 10:1–18 (Table 2), 12:32–38,
12:53–57, 12:61–13:9, 14:1–5, 15:1–15, Fig. 4A; Ex. 1009, 23:19–23,
27:16–19, 28:14–24, 54:13–17, 60:10–15, Fig. 22; Ex. 1006 ¶¶ 335–339,
631). Because the asserted ground is based on the combination of Riddle
and Ferdinand, Patent Owner’s arguments that Ferdinand does not
individually teach storing flow-entries in a database are unpersuasive. See
Bradium Techs. LLC v. Iancu, 923 F.3d 1032, 1050 (Fed. Cir. 2019)
(holding that arguments that “attack the disclosures of the two references
individually” cannot overcome a ground premised on the combination of
those references (citing In re Merck & Co., 800 F.2d 1091, 1097 (Fed. Cir.
1986) (“Non-obviousness cannot be established by attacking references
individually where the rejection is based upon the teachings of a
combination of references.”))).
Accordingly, we are persuaded that Petitioner has shown by a
preponderance of the evidence that the recited limitations are taught by the
combination of Riddle and Ferdinand.
IPR2020-00337
Patent 6,771,646 B1
47
f. Cache Subsystem Limitations (Claims 1 and 7)
Claim 1 of the ’646 patent recites, in part, “a cache subsystem coupled
to the flow-entry database memory providing for fast access of flow-entries
from the flow-entry database.” Ex. 1003, 36:52–54. Claim 7 similarly
recites “a cache subsystem coupled to and between the lookup engine and
the flow-entry database memory providing for fast access of a set of likely-
to-be-accessed flow-entries from the flow-entry database.” Id. at 38:19–22.
Petitioner contends Riddle in view of Ferdinand or Wakeman renders
these limitations obvious, based on Riddle’s disclosure of examining flow-
entries from stored flow-entry lists (see supra § III.E.1.e), together with
Ferdinand’s or Wakeman’s cache systems. Pet. 57–59 (citing Ex. 1006
¶¶ 627–648, 668–671, 707–714; Ex. 1008, 13:40–47; Ex. 1009, 18:27–29,
28:14–21; Ex. 1014, 1:20–28, 1:55–67, 2:31–49, 3:36–45, 4:31–40, 5:22–27,
Figs. 2, 3). Patent Owner does not dispute Petitioner’s contentions regarding
these limitations. See generally PO Resp.
Petitioner’s assertions are supported by the cited evidence and are
persuasive. Moreover, we are persuaded that a person of ordinary skill in
the art would have been motivated to combine the teachings of Riddle and
Ferdinand or Wakeman as proposed by Petitioner and supported by the
testimony of Dr. Weissman. See Pet. 25–27 (citing Ex. 1008 ¶¶ 256–257,
259), 29–30 (citing Ex. 1008 ¶¶ 185–191, 613–616). Accordingly, we are
persuaded that Petitioner has shown by a preponderance of the evidence that
the recited limitations are taught by the combination of Riddle and
Ferdinand or the combination of Riddle and Wakeman.
g. Lookup Engine Limitations (Claims 1 and 7)
Claim 1 of the ’646 patent recites, in part, “a lookup engine coupled to
the packet acquisition device and to the cache subsystem and configured to
IPR2020-00337
Patent 6,771,646 B1
48
lookup whether a received packet belongs to a flow-entry in the flow-entry
database, the looking up being via the cache subsystem.” Ex. 1003, 36:55–
59, 47 (Certificate of Correction dated Oct. 15, 2013). Claim 7 similarly
recites “a lookup engine coupled to the output of the parser subsystem and to
the flow-entry memory and configured to lookup whether the particular
packet whose parser record is output by the parser subsystem has a matching
flow-entry, the looking up using at least some of the selected packet portions
and determining if the packet is of an existing flow.” Id. at 38:12–18.
Petitioner contends Riddle in view of Ferdinand or Wakeman renders
these limitations obvious. Pet. 59 (citing Ex. 1006 ¶¶ 340–344, 649–652,
705, 706). Referring back to its contentions with respect to the flow-entry
database limitations of claims 1 and 7 (see supra § III.E.1.e), Petitioner
contends that “Riddle looks up whether a flow matches a traffic class in
relation to classifying a service aggregate based on a plurality of indicators”
and “Riddle’s operations run on a processor having programming code
performing lookup functions.” Id. at 60 (citing Ex. 1008, 5:53–57, 12:42–
49, 13:42–47, 16:40–48 (claim 5), Fig. 4B; Ex. 1006 ¶¶ 340–344, 650, 692–
700). Accordingly, Petitioner argues, relying on the testimony of
Dr. Weissman, a person of ordinary skill in the art “would have understood
Riddle’s processor and memory include a lookup engine,” where such
“lookup engine is configured to determine whether a received packet
belongs to a flow-entry in the flow-entry lists.” Id. at 60–61 (citing Ex. 1006
¶¶ 340–344; Ex. 1008, 12:37–49, 16:54–17:15 (claim 8), Figs. 3, 4A, 4B).
Still further, Petitioner contends, as discussed in connection with the “cache
subsystem” claim elements (see supra § III.E.1.f), “Riddle in view of
Ferdinand or Wakeman renders obvious a cache subsystem,” and a person of
ordinary skill in the art “would have been motivated to modify Riddle’s
IPR2020-00337
Patent 6,771,646 B1
49
lookup engine to employ a cache subsystem as taught by Ferdinand or
Wakeman, as this modification would improve Riddle’s monitor by storing
source and destination flow-entry addresses” and “would have understood
that Riddle’s looking up would be done via the cache subsystem” when
including such a cache. Id. at 61 (citing Ex. 1006 ¶¶ 637–648, 653). Patent
Owner does not dispute Petitioner’s contentions regarding these limitations.
See generally PO Resp.
Petitioner’s assertions are supported by the cited evidence and are
persuasive. Moreover, we are persuaded that a person of ordinary skill in
the art would have been motivated to combine the teachings of Riddle with
Ferdinand or Wakeman as proposed by Petitioner. See Pet. 25–27 (citing
Ex. 1008 ¶¶ 256–257, 259), 29–30 (citing Ex. 1008 ¶¶ 185–191, 613–616).
Accordingly, we are persuaded that Petitioner has shown by a preponderance
of the evidence that the recited limitations are taught by the combination of
Riddle and Ferdinand or the combination of Riddle and Wakeman.
h. State Processor Limitation (Claim 1)
Claim 1 of the ’646 patent recites, in part, “a state processor coupled to the
lookup engine and to the flow-entry-database memory, the state processor
being to perform any state operations specified for the state of the flow
starting from the last encountered state of the flow in the case that the packet
is from an existing flow, and to perform any state operations required for the
initial state of the new flow in the case that the packet is not from an existing
flow.” Ex. 1003, 36:60–67, 47 (Certificate of Correction dated Oct. 15,
2013).
(1) The Parties’ Arguments
Petitioner persuasively contends that Riddle alone or in view of
Ferdinand or Wakeman renders this limitation obvious. Pet. 62 (citing
IPR2020-00337
Patent 6,771,646 B1
50
Ex. 1006 ¶¶ 654–671). First, Petitioner contends, it would have been
obvious to use a database for Riddle’s memory storing flow-entry lists, for
reasons discussed with respect to the “flow entry database” limitation of
claim 1. Id.; see supra § III.E.1.e. Further, Petitioner contends, “Riddle
discloses a state processor coupled to that memory where the processor
performs state operations for existing and new flows.” Id. at 62–63 (citing
Ex. 1006 ¶¶ 345–372, 378–393, 627–636, 654–657, 659, 660; Ex. 1008,
5:53–57, 9:28–41, 10:19–56, 12:42–13:23, 14:1–5, 15:56–16:14 (claim 1),
Figs. 2A, 2B, 3, 4A). Relying on the testimony of Dr. Weissman, Petitioner
asserts that a person of ordinary skill in the art “would have understood that
Riddle’s processor, lookup engine, and flow-entry database memory work
together to achieve the goal of classifying traffic, and therefore would be
coupled together.” Id. at 63 (citing Ex. 1006 ¶ 657). Further, Petitioner
contends, Riddle stores previously encountered flows and performs state
operations on both existing and new conversational flows. Id. at 64–66
(citing Ex. 1006 ¶¶ 627–636, 662–671; Ex. 1008, 11:25–31, 12:30–13:22,
14:1–5, 18:26–28 (claim 15), Fig. 4A).
In response to Petitioner’s arguments, Patent Owner contends that the
prior art does not teach conversational flow state-based analysis. PO Resp.
53–57. According to Patent Owner, “Riddle relates to classifying traffic
based on an individual packet in the flow, rather than on a state of the flow
(i.e., evaluation across multiple packets).” Id. at 53. More particularly,
Patent Owner contends, “Riddle views each packet as an independent and
unique occurrence and focuses on evaluating that packet—in isolation—to
determine whether a given category of traffic has encountered a ‘hit.’” Id.
(citing Ex. 2061 ¶¶ 62–65, 80–83). According to Patent Owner, “[t]he
teachings of Riddle relied on by Petitioner[] show Riddle stepping through a
IPR2020-00337
Patent 6,771,646 B1
51
hierarchical traffic class tree to classify a packet,” but “this analysis turns
solely on an individual packet’s information and ‘multiple orthogonal
classification attributes.’” Id. (citing Ex. 1008, Fig. 2A, 9:16–18, 9:20–25,
10:20–39). In contrast with claim 1, Patent Owner contends, “Riddle does
not analyze a packet based on the context provided by previously
encountered packets.” Id. Because “Riddle’s evaluation turns based on the
‘information’ and ‘attributes’ in an individual packet, as opposed to relying
on the context and relationship of various packets with one another,” Patent
Owner argues, “Riddle has no concept of ‘state of the flow.’” Id. at 55.
Finally, citing the Board’s findings in related proceedings that Riddle does
not teach state transition patterns or “the state of [a] conversational flow of
[a] packet being indicative of the sequence of any previously encountered
packets of the same conversational flow,” Patent Owner further contends
that the Board already rejected arguments mirroring those raised by
Petitioner in this proceeding with respect to this limitation. Id. at 55–57
(citing IPR2020-00335, Paper 19 at 19–20 (PTAB Aug. 27, 2020);
IPR2020-00336, Paper 21 at 48–49 (PTAB Sept. 10, 2020)).
In its Reply, Petitioner points out that, in contrast to the limitations at
issue in the related proceedings cited by Patent Owner, claim 1 of the ’646
patent recites “a conversational flow being an exchange of one or more
packets in any direction between two network entities . . . the state processor
being to perform any state operations . . . in the case that the packet is from
an existing flow, and to perform any state operations . . . in the case that the
packet is from an existing flow.” Reply 15 (alterations in original).
Petitioner contends that the limitations of claim 1 thus “refer to the [’646]
patent’s embodiments whereby information from an individual packet is
sufficient for the state-based analysis of a conversational flow” and “[n]one
IPR2020-00337
Patent 6,771,646 B1
52
of these limitations requires state transitions to occur across a sequence of
packets.” Id. Furthermore, Petitioner argues, even if the state-related
limitations required analyzing more than one packet, Riddle would meet
them by its disclosure of classifying service aggregates based on a plurality
of indicators across multiple packets. Id. at 15–17 (citing, e.g., Ex. 1008,
11:11–20, 12:42–63, 13:40–59, Fig. 4B, claim 5; Ex. 1006 ¶¶ 120, 121, 132,
353, 356, 357, 360, 361, 366–369, 628, 632–634, 665, 666).
Patent Owner contends in its Sur-reply that Petitioner has omitted
critical claim language by using ellipses to present an abbreviated version of
the recited claim limitation. Sur-reply 19. According to Patent Owner,
Petitioner thereby provides an improper invalidity analysis by dropping
several requirements of the claim language. Id. For example, Patent Owner
contends, “claim 1 requires the presence of a state processor that can
perform certain steps, which appears nowhere in Riddle,” and “Petitioner
glosses over the requirement that a state processor be used.” Id. “Even
under Petitioner’s interpretation of a conversational flow,” Patent Owner
argues, “Riddle merely tallies occurrences of certain types of packets, and
does not consider the sequence of any previously encountered packets of the
same flow.” Id. at 19–20. According to Patent Owner, “[n]othing Petitioner
cites in Riddle suggests deviating from its serial analysis, nor does Petitioner
argue that the processing of a packet in Riddle depends on earlier packets.”
Id. at 20.
(2) Analysis
Having considered the parties’ respective contentions, we find that
Petitioner’s assertions are supported by the cited evidence and are persuasive
that the combination of Riddle and Ferdinand teaches this limitation for the
reasons presented by Petitioner, which we adopt. We agree with Petitioner
IPR2020-00337
Patent 6,771,646 B1
53
that the disputed limitation differs substantially from the limitations that
were at issue in the related proceedings cited by Patent Owner and that the
differences are significant with respect to our analysis. Notably, the Board
did not find in either of those related proceedings that Riddle fails to disclose
a “state processor . . . to perform any state operations specified for . . . an
existing flow, and to perform any state operations required for the initial
state of [a] new flow,” as recited in independent claim 1 of the ’646 patent.
Instead, the Board found, for example, that Petitioner had not sufficiently
shown that the asserted prior art teaches or suggests “state transition
patterns” across multiple packets (IPR2020-00335, Paper 19 at 14–22
(emphasis added)). That limitation does not appear in claim 1 of the
’646 patent. Moreover, we agree with Petitioner that information only from
an individual packet is required for the state-based analysis recited in
claim 1 and that the claim does not require state transitions to occur across a
sequence of packets. See Reply 15. In particular, as Petitioner explains,
claim 1 here recites “a conversational flow being an exchange of one or
more packets.” Ex. 1003, 36:48–51 (emphasis added). And thus
information only from an individual packet is required (i.e., “one” packet is
covered). In contrast, the corresponding limitation that was at issue in the
related proceedings recited “a sequence of packets” or a “conversational
flow of packets” (i.e., packets in plural form). See Ex. 1001, 35:34–35
(reciting limitation in claim 1 of the ’099 patent).
Accordingly, we are persuaded that Petitioner has shown by a
preponderance of the evidence that the recited limitation is taught by the
combination of Riddle and Ferdinand.
IPR2020-00337
Patent 6,771,646 B1
54
i. Remaining Limitations of Claims 7 and 16
Supported by the testimony of Dr. Weissman, Petitioner contends, and
we agree, that Riddle discloses each of the remaining limitations of claims 7
and 16. Pet. 66–72 (citing Ex. 1006 ¶¶ 343, 668–671, 692–700, 705–706,
715–734; Ex. 1008, code (57), 4:10–15, 5:53–57, 8:47–9:27, 9:48–49, 9:64–
65, 12:26–60, 13:36–62, 15:16–27, 15:56–16:30 (claims 1–3), 16:54–17:15
(claim 8), Figs, 3, 4A, 4B), 79. Patent Owner does not dispute Petitioner’s
contentions regarding these limitations. See generally PO Resp.
Petitioner’s assertions are supported by the cited evidence and are
persuasive. Accordingly, we are persuaded that Petitioner has shown by a
preponderance of the evidence that each of the recited limitations is taught
by Riddle and adopt the contentions set forth in the Petition and in
Dr. Weissman’s Declaration as mapped to these undisputed limitations of
claims 7 and 16 as our own findings. NuVasive, 841 F.3d at 974.
2. Dependent Claim 2
Claim 2 of the ’646 patent depends from claim 1 and further recites,
“a parser subsystem coupled to the packet acquisition device and to the
lookup engine such that the acquisition device is coupled to the lookup
engine via the parser subsystem, the parser subsystem configured to extract
identifying information from a received packet,” “wherein each flow-entry is
identified by identifying information stored in the flow-entry, and wherein
the cache lookup uses a function of the extracted identifying information.”
Ex. 1003, 37:1–10.
Supported by the testimony of Dr. Weissman, Petitioner contends, and
we agree, that Riddle in view of Ferdinand or Wakeman renders obvious
each of the further limitations of claim 2 and that Riddle in view of
Ferdinand and Wakeman renders obvious the further limitation of claim 3.
IPR2020-00337
Patent 6,771,646 B1
55
Pet. 72–75 (citing Ex. 1006 ¶¶ 294–327, 627–653, 672–680, 692–704;
Ex. 1008, 4:10–15, 6:1–15, 12:26–59, 16:54–17:15 (claim 8), Figs. 1A, 4A).
More particularly, according to Petitioner, the “parser subsystem coupled . . .
to the lookup engine” clause of claim 2 would have been obvious over
Riddle and Ferdinand for similar reasons to those presented for the similar
limitations of claim 7 discussed in Sections III.E.1.d and III.E.1.g above.
See id. at 72–73 (citing id. at 38–42, 59–61). With regard to the two
“wherein” clauses recited in claim 2, Petitioner points to Step 408 of
Riddle’s Figure 4A (i.e., “enter into a saved list characteristics of the
traffic”) and the discussion of the “cache subsystem” limitation of claim 1
(see supra § III.E.1.f) and contends that a person of ordinary skill in the art
“would have understood that Riddle’s lookup engine modified with the
cache subsystem as taught by Ferdinand and/or Wakeman results in the
cache lookup using a function of the extracted identifying information.” Id.
at 74–75 (citing Ex. 1006 ¶¶ 637–653, 677–680; Ex. 1008, 12:42–53, Fig.
4A).
Patent Owner does not dispute Petitioner’s contentions regarding the
limitations recited in claim 2. See generally PO Resp.
Petitioner’s assertions are supported by the cited evidence and are
persuasive. We are persuaded that Petitioner has shown by a preponderance
of the evidence that these undisputed limitations are taught by the asserted
prior art and adopt the contentions set forth in the Petition and in
Dr. Weissman’s Declaration as mapped to these limitations as our own
findings. NuVasive, 841 F.3d at 974.
3. Dependent Claim 3
Claim 3 of the ’646 patent depends from claim 2 and further recites,
“wherein the cache subsystem is an associative cache subsystem including
IPR2020-00337
Patent 6,771,646 B1
56
one or more content addressable memory cells (CAMs).” Ex. 1003, 37:11–
13. Petitioner contends that Riddle in view of Ferdinand and Wakeman
renders obvious the recited limitation. Pet. 75–76 (citing Ex. 1006 ¶¶ 185–
191, 637–648, 681–685, 692–704; Ex. 1003, 19:56–20:2). More particular,
Petitioner contends, “[c]onsistent with the ’646 Patent’s cache design
paradigm that includes CAMs, a [person of ordinary skill in the art] would
have understood that an associative cache was a well-known cache design
paradigm”; that Wakeman teaches a CAM cache and it would have been
obvious to modify Riddle’s flow-entry with a cache subsystem as taught, for
example, by Wakeman; and that a person of ordinary skill in the art would
have understood that Wakeman’s CAM cache is fully associative. Id. (citing
Ex. 1006 ¶¶ 185–191, 637–648, 681–685). Relying on Dr. Weissman’s
testimony, Petitioner further contends that “[b]efore the ’646 Patent, an
associative cache was readily known to a [person of ordinary skill in the art]
and using an associative cache was a routine engineering design choice.” Id.
at 76 (citing Ex. 1006 ¶¶ 683–685).
Patent Owner contends that the asserted prior art fails to teach or
suggest the associative cache subsystem recited in claim 3. PO Resp. 57–59.
In particular, Patent Owner argues:
Petitioners argue that the use of an associative cache
subsystem in ’646 Claim 3 is obvious, because: “a [person of
ordinary skill in the art] would have understood that an
associative cache was a well-known cache design paradigm in
which a memory block can appear in any cache line and that the
cache line is a form of a content addressable memory.” Petition
at 75–76. Petitioners provide no support for this beyond a
citation to Dr. Weissman, and Dr. Weissman provides only a
conclusion that “Wakeman’s CAM cache can be a fully
associative cache,” despite Wakeman not identifying what kind
of cache it uses. Petitioners further argue “a [person of ordinary
IPR2020-00337
Patent 6,771,646 B1
57
skill in the art] would have understood that Wakeman’s CAM
cache is fully associative due to Wakeman not disclosing a cache
placement policy indicating there are no constraints on
placement within Wakeman’s cache.” Id. Dr. Weissman does
not explain why the absence of a discussion of a cache placement
policy in Wakeman would lead a [person of ordinary skill in the
art] to just assume that Wakeman was teaching using of an
associative cache placement policy.
First, Dr. Weissman did not provide any facts or analysis
underlying his opinion, and therefore his analysis is entitled to
little weight. 37 CFR § 42.65(a). His conclusory statement that
Wakeman could be modified to utilize an associative cache does
not demonstrate that a [person of ordinary skill in the art] would
have been motivated to implement Wakeman’s cache as a fully
associative cache. “The mere fact that the prior art could be so
modified would not have made the modification obvious unless
the prior art suggested the desirability of the modification.” In
re Mills, 916 F.2d 680, 682 (Fed. Cir. 1990) (quoting In re
Gordon, 733 F.2d 900, 902 (Fed. Cir. 1984)) (emphasis added).
Second, even if an associative cache were a “common type
of cache” as Dr. Weissman claims, its mere existence does not
mean its use with the specific combination of elements claimed
in ’646 Patent Claim 3 would have been obvious. Ex. 1006
¶ 683; Ex. 2061 ¶¶ 88–90. Associative caches are more
expensive than direct mapping caches due to the requirement that
every potential cache location must be searched in parallel when
checking whether a value is already in the cache. Ex. 2061 ¶ 89.
Wakeman asserts that “[c]hoosing an appropriate network switch
[] requires a balancing between cost and performance”—it does
not espouse the use of an associative caching policy over any
other policy, and in light of Wakeman’s emphasis on balancing
cost, a [person of ordinary skill in the art] would not simply
assume Wakeman to be teaching the use of the more expensive
associative cache (as Dr. Weissman appears to argue). EX1014
(Wakeman) at 2:1–2; Ex. 2061 ¶ 90. If the use of an associative
cache was truly as “common” in the field as Petitioners assert,
then Petitioners would have offered a reference or at least some
form of evidence demonstrating that its use was common in the
field, or that a [person of ordinary skill in the art] would have
assumed the use of an associative cache in the absence of any
IPR2020-00337
Patent 6,771,646 B1
58
discussion on cache placement policy. They did not, and for at
least these reasons Petitioners have not shown claim 3 to be
invalid.
PO Resp. 57–59.
In its Reply, Petitioner responds that Patent Owner and Ms. Quigley
“concede [that] an associative cache was a known cache design that predates
the challenged claims” and “don’t dispute that the prosecution history
illustrates that using an associative cache was readily understood in the art to
expedite memory accesses to and from processors.” Reply 27 (citing PO
Resp. 58; Ex. 2061 ¶ 88). Petitioner further contends that “Dr. Weissman
and the Petition refer to the prosecution history in which the Examiner
discussed how associative caches were known in the art,” and
“Dr. Weissman explains it would have been obvious to use associative
caches because they expedite memory access for databases, particularly
when storing Ethernet packet addresses, as taught in Riddle and Ferdinand.”
Id. at 27–28 (citing Ex. 1006 ¶¶ 643–648, 683, 684; Ex. 1019, 193–201
(’646 patent’s prosecution history); Pet. 58–59, 75–76). Still further,
Petitioner argues that Ms. Quigley’s testimony on which Patent Owner bases
its lack-of-motivation argument “is bereft of any evidentiary support and is
conclusory,” and therefore, is entitled to little weight. Id. at 28 (citing
Ex. 2061 ¶ 89; PO Resp. 58; 37 CFR § 42.65(a)). According to Petitioner,
the fact that other cache types existed does not lessen the obviousness of
using an associative cache in Riddle’s system, and Ms. Quigley’s testimony
that Wakeman teaches away from using an associative cache “overlooks
unrebutted testimony from Dr. Weissman that ‘[associative] caches were
well-known to reduce look-up times’ and thus would have improved the
performance of Riddle’s system.” Id. at 28–29 (alteration in original) (citing
IPR2020-00337
Patent 6,771,646 B1
59
In re Mouttet, 686 F.3d 1322, 1334 (Fed. Cir. 2012); Pet. 75–76; Ex. 1006
¶ 684).
Having reviewed the full trial record, we conclude that Petitioner has
not established by a preponderance of the evidence that claim 3 is
unpatentable over the combination of Riddle, Ferdinand, and Wakeman. We
agree with Patent Owner that Petitioner provides insufficient evidence to
support its contentions. Indeed, the Petition’s discussion with respect to
claim 3 does not even cite Wakeman, but instead cites only the testimony of
Dr. Weissman and the specification of the ’646 patent. See Pet. 75–76.12
And neither Dr. Weissman’s testimony nor the ’646 patent provides any
persuasive support for Petitioner’s assertion that “a [person of ordinary skill
in the art] would have understood that Wakeman’s CAM cache is fully
associative due to Wakeman not disclosing a cache placement policy
indicating there are no constraints on placement within Wakeman’s cache.”
We agree with Patent Owner that “Dr. Weissman does not explain why the
absence of a discussion of a cache placement policy in Wakeman would lead
a [person of ordinary skill in the art] to just assume that Wakeman was
teaching using of an associative cache placement policy,” and that his
testimony in this regard is entitled to little, if any, weight. PO Resp. 58;
37 CFR § 42.65(a). As Patent Owner contends, Dr. Weissman’s statement
that Wakeman could be modified to utilize an associative cache does not

12 To be sure, the discussion of claim 3 in the Petition also includes an
internal cross-reference to the “discuss[ion] regarding element 1.3,” i.e., the
“cache subsystem” limitation of claim 1, discussed at Pet. 57–59. Although
the discussion of that limitation does include citations to Wakeman (see
Pet. 58 (citing Ex. 1014, 1:20–28, 1:55–67, 2:31–49, 3:36–45, 4:31–40,
5:22–27, Figs. 2, 3)), however, notably absent from that discussion is any
assertion that Wakeman specifically suggests an associative cache.
IPR2020-00337
Patent 6,771,646 B1
60
demonstrate that a person of ordinary skill in the art would have been
motivated to implement Wakeman’s cache as a fully associative cache. Id.
(citing In re Mills, 916 F.2d 680, 682 (Fed. Cir. 1990)). Although we agree
with Petitioner that there is no indication that Wakeman disparages the use
of an associative cache (Reply 28–29), there still needs to be “some rational
underpinning to support the legal conclusion of obviousness” (KSR, 550
U.S. at 418), and we do not find such underpinning on the record before us.
Accordingly, we determine that Petitioner has not shown by a preponderance
of the evidence that claim 3 is unpatentable over Riddle in view of
Ferdinand and further in view of Wakeman.
4. Dependent Claim 18
Claim 18 of the ’646 patent depends from claim 16 and further recites,
wherein the function of the selected portions of the packet forms
a signature that includes the selected packet portions and that can
identify future packets, wherein the lookup operation uses the
signature and wherein the identifying information stored in the
new or updated flow-entry is a signature for identifying future
packets.
Ex. 1003, 40:8–14. Petitioner contends, and we agree, that these limitations
are taught by Riddle, asserting that Riddle discloses “pars[ing] packet
portions to identify a signature (i.e., identifying information) and stor[ing]
the signature extracted from [the] parsed packet portion[s] for identifying
future packets,” as well as “using identifying information that includes
selected portions of the packet to identify future packets to suppress
duplicates of previously identified packet flows.” Pet. 79–80 (citing
Ex. 1006 ¶¶ 746–750; Ex. 1008, 12:42–59, Fig. 4A). Patent Owner does not
dispute Petitioner’s contentions regarding these limitations. See generally
PO Resp.
IPR2020-00337
Patent 6,771,646 B1
61
Petitioner’s assertions are supported by the cited evidence and are
persuasive. We are persuaded that Petitioner has shown by a preponderance
of the evidence that these undisputed limitations are taught by the asserted
prior art and adopt the contentions set forth in the Petition and in
Dr. Weissman’s Declaration as mapped to these limitations as our own
findings. NuVasive, 841 F.3d at 974.
5. Conclusion Regarding Ground 1
For the foregoing reasons, we conclude that Petitioner has shown by a
preponderance of the evidence that the subject matter of claims 1, 2, 7, 16,
and 18 is unpatentable for obviousness over Riddle in view of Ferdinand and
Wakeman. As to claim 3, however, we find that Petitioner has not provided
sufficient and persuasive evidence and reasoning that the combination of
Riddle, Ferdinand, and Wakeman teaches or suggests the associative cache
limitation, and we, therefore, conclude that Petitioner has not shown that the
subject matter of claim 3 would have been obvious on this ground.
F. Obviousness over Riddle, Ferdinand, Wakeman, and Yu, or over
Riddle, Ferdinand, Wakeman, and RFC1945
Petitioner contends that claims 1–3, 7, 16, and 18 also would have
been obvious over the combination of Riddle, Ferdinand, and Wakeman, in
further view of Yu or RFC1945. See Pet. 80–87 (Riddle in view of
Ferdinand, Wakeman, and Yu); id. at 87–95 (Riddle in view of Ferdinand,
Wakeman, and RFC1945). Having determined that Petitioner establishes by
a preponderance of the evidence that the subject matter of claims 1, 2, 7, 16,
and 18 would have been obvious over the combination of Riddle, Ferdinand,
and Wakeman, we need not address Petitioner’s additional grounds
challenging these claims. See SAS Inst. Inc. v. Iancu, 138 S. Ct. 1348, 1359
(2018) (holding that a petitioner “is entitled to a final written decision
IPR2020-00337
Patent 6,771,646 B1
62
addressing all of the claims it has challenged”); Boston Sci. Scimed, Inc. v.
Cook Grp. Inc., 809 F. App’x 984, 990 (Fed. Cir. 2020) (nonprecedential)
(“We agree that the Board need not address [alternative grounds] that are not
necessary to the resolution of the proceeding.”).
Further, with respect to claim 3, which we conclude that Petitioner did
not show to be unpatentable over the combination of Riddle, Ferdinand, and
Wakeman (see supra § III.E.3), Petitioner has not alleged that Yu or
RFC1945 teaches or suggests the recited associative cache subsystem. See
generally Pet.; Reply. Accordingly, we likewise conclude that Petitioner has
not shown claim 3 to be unpatentable by a preponderance of the evidence
over the combinations of Riddle, Ferdinand, and Wakeman with either Yu or
RFC1945.
IV. CONCLUSION13
Based on the evidence presented with the Petition, the evidence
introduced during the trial, and the parties’ respective arguments, Petitioner
has shown by a preponderance of the evidence that each of claims 1, 2, 7,
16, and 18 of the ’646 patent is unpatentable, but Petitioner has not shown
that claim 3 is unpatentable.

13 Should Patent Owner wish to pursue amendment of the challenged claims
in a reissue or reexamination proceeding subsequent to the issuance of this
decision, we draw Patent Owner’s attention to the April 2019 Notice
Regarding Options for Amendments by Patent Owner Through Reissue or
Reexamination During a Pending AIA Trial Proceeding. 84 Fed. Reg.
16,654 (Apr. 22, 2019). If Patent Owner chooses to file a reissue application
or a request for reexamination of the challenged patent, we remind Patent
Owner of its continuing obligation to notify the Board of any such related
matters in updated mandatory notices. See 37 C.F.R. § 42.8(a)(3), (b)(2).
IPR2020-00337
Patent 6,771,646 B1
63
V. ORDER
Accordingly, it is
ORDERED that claims 1, 2, 7, 16, and 18 of U.S. Patent
No. 6,771,646 B1 are held unpatentable under 35 U.S.C. § 103 as obvious;
FURTHER ORDERED that claim 3 of U.S. Patent No. 6,771,646 B1
is not held unpatentable; and
FURTHER ORDERED that, because this is a Final Written Decision,
parties to the proceeding seeking judicial review of this decision must
comply with the notice and service requirements of 37 C.F.R. § 90.2.
In summary:

14 As explained above, we do not reach this ground except with respect to
claim 3. See supra § III.F.
15 As explained above, we do not reach this ground except with respect to
claim 3. See supra § III.F.
Claim(s)

35
U.S.C. §
Reference(s)/Basis Claims
Shown
Unpatentable
Claims
Not shown
Unpatentable
1–3, 7, 16, 18 103(a) Riddle, Ferdinand,
Wakeman
1, 2, 7, 16, 18 3
1–3, 7, 16, 18 103(a) Riddle, Ferdinand,
Wakeman, Yu14
3
1–3, 7, 16, 18 103(a) Riddle, Ferdinand,
Wakeman,
RFC194515
3
Overall
Outcome
1, 2, 7, 16, 18 3
IPR2020-00337
Patent 6,771,646 B1
64
FOR PETITIONER:
Joseph F. Edell
Adam A. Allgood
FISCH SIGLER LLP
joe.edell.ipr@fischllp.com
adam.allgood@fischllp.com

Scott A. McKeown
James R. Batchelder
Mark D. Rowland
Andrew Radsch
ROPES & GRAY LLP
scott.mckeown@ropesgray.com
james.batchelder@ropesgray.com
mark.rowland@ropesgray.com
andrew.radsch@ropesgray.com

FOR PATENT OWNER:
R. Allan Bullwinkel
Michael Heim
HEIM PAYNE & CHORUSH, LLP
abullwinkel@hpcllp.com
mheim@hpcllp.com