AMAZON.COM, INC. v. ZITOVAULT, LLC , No. 17-2147 (Fed. Cir. 2018)

Annotate this Case
Download PDF
NOTE: This disposition is nonprecedential. United States Court of Appeals for the Federal Circuit ______________________ AMAZON.COM, INC., AMAZON.COM, LLC, AMAZON WEB SERVICES, INC., BAZAARVOICE, INC., GEARBOX SOFTWARE, LLC, Appellants v. ZITOVAULT, LLC, Appellee ______________________ 2017-2147 ______________________ Appeal from the United States Patent and Trademark Office, Patent Trial and Appeal Board in Nos. IPR201600021, IPR2016-01025. ______________________ Decided: November 16, 2018 ______________________ DAN L. BAGATELL, Perkins Coie LLP, Hanover, NH, argued for appellants. Also represented by GRANT EDWARD KINSEL, Los Angeles, CA; CHRISTINA JORDAN MCCULLOUGH, JONATHAN R. PUTMAN, Seattle, WA. JUSTIN NEMUNAITIS, Caldwell Cassady & Curry, Dallas, TX, argued for appellee. Also represented by AMAZON.COM, INC. v. ZITOVAULT, LLC 2 JASON DODD CASSADY; MICHAEL RAYMOND CASEY, Oblon, McClelland, Maier and Neustadt, LLP, Alexandria, VA. ______________________ Before PROST, Chief Judge, O’MALLEY and STOLL, Circuit Judges. Opinion for the court filed by Circuit Judge STOLL. Dissenting Opinion filed by Chief Judge PROST. STOLL, Circuit Judge., Inc.,, LLC, Amazon Web Services, Inc., Bazaarvoice, Inc., and Gearbox Software, LLC, (collectively, “Amazon”), appeal from a final written decision of the Patent Trial and Appeal Board in which the Board held that Amazon failed to prove ZitoVault, LLC’s U.S. Patent No. 6,484,257 unpatentable. The Board did not err in its claim construction, and it correctly held Amazon to its burden of proof. Because it did not err in finding Amazon failed to carry that burden and because it did not violate Amazon’s procedural due process rights, we affirm. BACKGROUND ZitoVault’s ’257 patent seeks to improve computer systems’ handling of encrypted communications. See ’257 patent col. 3 l. 65–col. 4 l. 1. Rather than using a single “main” server to decrypt every communication, the disclosed system also enlists the computers receiving the communications as decryption agents, thereby avoiding bottlenecks, as shown in Figure 2. AMAZON.COM, INC. v. ZITOVAULT, LLC 3 Id. at Fig. 2; see also id. at col. 7 ll. 21–34. Representative claims 1 and 6 describe the claimed invention: 1. A system for conducting a plurality of cryptographic sessions over a distributed network of computers, employing a distributed automaton running on the network comprising M agents for servicing N number of simultaneous cryptographic sessions wherein bandwidth and number of sessions are scalable by the M agents and latency is potentially reducible to zero comprising: a main server; one or more clients communicating over the distributed network with said main server and agents; 4 AMAZON.COM, INC. v. ZITOVAULT, LLC M agents communicating with the main server for enlisting additional agents to support incremental cryptographic sessions with the clients to maintain system performance at a desired level; and for encrypting and decrypting communication traffic as it arrives from the clients via the main server, the agents comprising a single-tomany connection (1 client, M agents) with respect to the clients, such that portions of the bandwidth are equally divided among the M agents for processing, and the agents combine the processing power of all computers connected to the system to service encryption and decryption and enable bandwidth to be scalable by the M agents and to reduce latency substantially to zero. *** 6. A method for implementing a scaleable software crypto system between a main server and one or more agent servers communicating with one or more clients such that performance of the crypto system is increased to meet any demand comprising providing a secure communication between the main server, agent server, and one or more clients such that communication between the main server and agent server enlists additional agent servers to support incremental secure sessions in response to maintaining performance at a desired level. Id. at claims 1, 6 (emphases added to highlight disputed claim term). After ZitoVault sued Amazon for infringement, Amazon petitioned for inter partes review of the ’257 patent. Amazon raised three grounds of unpatentability, each AMAZON.COM, INC. v. ZITOVAULT, LLC 5 based on U.S. Patent No. 6,065,046 (“Feinberg”), and each instituted by the Board. Over the course of the IPR, the parties’ dispute crystalized around the issue of whether Feinberg discloses the claimed “sessions.” Amazon relied on Feinberg for every claim limitation reciting “sessions.” But Amazon did not delineate exactly where Feinberg describes the claimed sessions and did not explain what constitutes a session in Feinberg’s system. Amazon also did not propose a construction of “sessions,” but its expert testified that a “session generally refers to one or more communications exchanged between two entities over some period of time.” J.A. 540. At the institution stage, the Board accepted Amazon’s contention that Feinberg discloses “sessions.” Citing a telecommunications dictionary, it preliminarily construed “sessions” as “a set of transmitters and receivers, and the data streams that flow between them.” J.A. 180. It found that “based on that construction, the mere exchange of data (e.g., encrypted code modules), as disclosed in Feinberg, falls within the scope of the claimed sessions.” Id.; see also J.A. 185–86 (“[W]e adopt a broader construction of the term ‘session’ that encompasses simply the exchange of [data] packets.”). In its Patent Owner Response, ZitoVault maintained that Feinberg lacked the claimed sessions. It offered expert testimony that a “session” “must refer to a connection with a defined beginning and end” so that the server can determine which incoming data belongs to which session. J.A. 219–20, 1184–86. ZitoVault further contended that Amazon’s petition was defective because it “fail[ed] to specifically identify what it contends is the ‘session’ in Feinberg or how that session is initiated, maintained, or terminated.” J.A. 235. ZitoVault separately urged the Board to find that a reference must disclose “negotiating the initiation of a stream with a 6 AMAZON.COM, INC. v. ZITOVAULT, LLC defined beginning and end,” to disclose a “data stream,” as required by the Board’s preliminary construction of “sessions.” J.A. 237–38. At oral argument before the Board, Amazon conceded that a “session” would have a beginning and an end, J.A. 317, 319–20, 327, but it maintained that a reference need not disclose “protocol level” details regarding data stream initiation and termination to meet the claims, J.A. 319–20. In response to questions from the Board, Amazon specifically invited the panel to require a beginning and an end as part of its construction of “sessions,” J.A. 326–27, asserting that “there can’t be any meaningful doubt” that Feinberg discloses a session with a “recognizable beginning and end,” J.A. 381. In its final written decision, the Board narrowed its preliminary construction of “sessions.” It construed the term as “a set of transmitters and receivers, and the data streams that flow between them wherein each data stream flowing between the transmitters and receivers has a recognizable beginning of the data stream transmission and a recognizable end of the data stream transmission.” J.A. 15 (emphasis added). It explained that it added the italicized portion of the construction to clarify “that there must be delineation between multiple sessions to allow one to distinguish multiple sessions from one another.” J.A. 14. The Board noted that both parties had agreed to its final construction. J.A. 22. And it emphasized that its “modified interpretation” did not limit a “session” “to any particular technique or protocol for recognizing the beginning and end of a session exchanged between a transmitter and a receiver,” rejecting that aspect of ZitoVault’s argument. J.A. 14. Applying the modified construction, the Board held that Amazon failed to prove that Feinberg disclosed “sessions.” It stated that Amazon “failed to identify what constitutes a ‘session’ in Feinberg,” J.A. 22, and that AMAZON.COM, INC. v. ZITOVAULT, LLC 7 Amazon’s “excessively generic” analysis of limitations reciting “sessions” left “uncertain what Petitioner regards as a session in Feinberg, much less how it begins and how it ends,” id.; see also J.A. 20–26. The Board therefore held that Amazon had not shown invalidity, and it entered a decision for ZitoVault. Amazon appeals, challenging the Board’s construction of “sessions” and its rejection of Amazon’s anticipation and obviousness grounds, and asserting procedural due process violations. We have jurisdiction under 28 U.S.C. § 1295(a)(4)(A). DISCUSSION This court reviews the Board’s legal determinations de novo and the Board’s factual determinations for substantial evidence. Belden Inc. v. Berk-Tek LLC, 805 F.3d 1064, 1073 (Fed. Cir. 2015). “A finding is supported by substantial evidence if a reasonable mind might accept the evidence to support the finding.” Polaris Indus., Inc. v. Arctic Cat, Inc., 882 F.3d 1056, 1064 (Fed. Cir. 2018). I Amazon first challenges the Board’s construction of “sessions.” We review the Board’s ultimate claim constructions de novo, see In re Man Mach. Interface Techs. LLC, 822 F.3d 1282, 1285 (Fed. Cir. 2016), and we review any subsidiary factual findings involving extrinsic evidence for substantial evidence, see Teva Pharm. USA, Inc. v. Sandoz, Inc., 135 S.Ct. 831, 841 (2015). In IPR, claims receive their broadest reasonable interpretation. 1 See Per recent regulation, the Board will apply the Phillips claim construction standard to petitions filed on or after November 13, 2018. See Changes to the Claim Construction Standard for Interpreting Claims in Trial Proceedings Before the Patent Trial and Appeal Board, 83 Fed. Reg. 51340 (Oct. 11, 2018) (to be codified at 1 8 AMAZON.COM, INC. v. ZITOVAULT, LLC 37 C.F.R. § 42.100(b). We have emphasized that under the BRI standard, the Board’s construction must be reasonable, that is, consistent with the record evidence and the understanding of one skilled in the art. See Microsoft Corp. v. Proxyconn, Inc., 789 F.3d 1292, 1298 (Fed. Cir. 2015), overruled on other grounds by Aqua Prods., Inc. v. Matal, 872 F.3d 1290 (Fed. Cir. 2017) (en banc). Because the Board’s construction comports with that standard, we affirm it. The Board construed “sessions” as “a set of transmitters and receivers, and the data streams that flow between them wherein each data stream flowing between the transmitters and receivers has a recognizable beginning of the data stream transmission and a recognizable end of the data stream transmission.” The required “recognizable” beginning and end of the data stream transmission, the Board explained, permits the system to distinguish one session from another. J.A. 12, 14. The plain claim language supports requiring a “recognizable” beginning and end. Independent claims 1, 7, and 10 each recite conducting a “plurality” of “sessions,” claim 6 requires multiple “sessions,” and claim 7 recites providing “simultaneous” sessions “among agents, server and clients,” all of which suggest that the system recognizes distinct sessions. See, e.g., ’257 patent at claims 1, 6, 7, 10. The specification similarly supports the understanding that the system must delineate between sessions. It describes that the server “determines if it can accept a new session” after considering the bandwidth used by existing sessions, id. at col. 7 ll. 25–36 (emphasis added), and it explains that the server may transfer or redirect particular sessions as needed, id. at col. 7 l. 62–col. 8 l. 9. The prosecution history contains no contrary disclosure. 37 C.F.R. pt. 42). Because Amazon filed its petition before November 13, 2018, we apply the BRI standard. AMAZON.COM, INC. v. ZITOVAULT, LLC 9 Extrinsic evidence further supports the Board’s construction. Both parties’ experts agreed that a session has a beginning and end. Amazon’s expert, Dr. Aviel Rubin explained that “[s]essions have start and end times,” J.A. 1072 at 140:19–20, while ZitoVault’s expert, Dr. Jonathan Katz, testified that a “session has a discernable beginning and end” and explained that this temporal aspect allows the system to distinguish among sessions, J.A. 1186. A telecommunications dictionary cited by the Board similarly emphasizes the temporal aspect of a “session,” defining the term as “[a] set of transmitters and receivers, and the data streams that flow between them. In other words, an active communication, measured from beginning to end, between devices or applications over a network.” J.A. 1254 (emphasis added). Moreover, both parties agreed to the Board’s construction. Amazon informed the Board that: If adding, you know, a beginning or an end to the session would provide the Board with comfort, the Panel with comfort, that we can distinguish between when one session starts or one session ends then, you know, I don’t have any objection to that given that that is what Dr. Rubin [Amazon’s expert] said a session was. See J.A. 326 (emphases added); see also J.A. 320 (“Now, if all we’re proposing is that there needs to be a beginning and an end to a session . . . I think that is clearly – that’s something that we had proposed or at least suggested through Dr. Rubin’s declaration.”); Oral Arg. at 7:10–15, 17-2147.mp3 (“We agreed to ‘recognizable’ . . . .”). The Board adopted the construction invited by Amazon, and that construction accords with both the intrinsic and the extrinsic evidence. We decline to disturb it on appeal. See Key Pharm. v. Hercon Labs. Corp., 161 F.3d 709, 715 (Fed. Cir. 1998) (“The impropriety of asserting a position 10 AMAZON.COM, INC. v. ZITOVAULT, LLC which the trial court adopts and then complaining about it on appeal should be obvious . . . . Indeed, we could appropriately refuse to entertain [party’s] appeal on the issue of claim construction.”). Amazon contends that even if the Board did not err in its literal construction of “sessions,” it erred by implicitly requiring disclosure of the protocols by which a session begins and ends. See Appellants’ Br. 25, 32 (asserting “the Board required the prior art to disclose a particular technique for recognizing the beginning and end of a session” and that the Board “did not simply require a ‘recognizable’ beginning and end”). We disagree. Contrary to Amazon’s argument, the Board expressly stated that its “interpretation is not limited to any particular technique or protocol for recognizing the beginning and end of a session” and that “[a] wide variety of techniques for such beginning and ending determinations are within the scope . . . of ‘session.’” J.A. 14 (emphasis added). We reject Amazon’s assertion that the Board implicitly adopted a construction that it expressly rejected in its final written decision. II Amazon also objects to the Board’s ultimate rejection of its anticipation and obviousness challenges, claiming that Feinberg discloses the “sessions” limitations under the Board’s construction. See, e.g., Appellants’ Br. 25 (“The Board specified that its modified interpretation was ‘not limited to any particular technique . . . .’ Had that been the Board’s final word, Feinberg’s disclosure would still have satisfied the ‘sessions’ limitations.”); Reply Br. 3–6; Oral Arg. at 8:51–9:47. Without resolving the question of what Feinberg discloses, the Board found that Amazon failed to offer sufficient proof that Feinberg discloses “sessions.” We agree. In an IPR, the petitioner bears the burden of proof. See In re Magnum Oil Tools Int’l, Ltd., 829 F.3d 1364, AMAZON.COM, INC. v. ZITOVAULT, LLC 11 1375–77 (Fed. Cir. 2016). “[M]ere conclusory statements” cannot suffice; the petitioner “must instead articulate specific reasoning, based on evidence of record” to support its grounds of unpatentability. Id. at 1380; see also Harmonic Inc. v. Avid Tech., Inc., 815 F.3d 1356, 1363 (Fed. Cir. 2016) (“In an IPR, the petitioner has the burden from the onset to show with particularity why the patent it challenges is unpatentable.”). In its petition, Amazon never explained how or when Feinberg distinguishes between sessions—Amazon never even defined a “session” in Feinberg’s system. Even in its reply, after ZitoVault placed the “sessions” terms at issue, Amazon argued only that Feinberg discloses “sessions,” without explaining whether any of Feinberg’s “sessions” begin and end, despite admitting in the same paper that a “session” would begin and end. J.A. 279–82; see also J.A. 540 (asserting that a “session generally refers to one or more communications exchanged between two entities over some period of time”). At oral argument, following the Board’s close questioning on this issue, Amazon offered only attorney argument that Feinberg “clearly” discloses sessions having a beginning and an end. J.A. 381. To the extent Amazon intended to present an inherency argument, these assertions fall short. See Oral Arg. at 5:35–6:35. We agree with the Board that Amazon’s failure to explain “what [it] regards as a session in Feinberg” and “whether a request in Feinberg begins a session, or something else in Feinberg begins a session,” dooms its petition. 2 J.A. 22. At argument, Amazon asserted that “a session begins when there is a request for service and then when the file is transferred it’s done.” Oral Arg. at 9:38–44. We express no opinion on this theory or on whether Feinberg discloses “sessions” to one of ordinary skill. Like the 2 12 AMAZON.COM, INC. v. ZITOVAULT, LLC III Finally, Amazon argues that at oral argument before the Board, ZitoVault argued for the first time that “a ‘session’ required a pre-conversation between communication endpoints that established specific parameters for the conversation.” Appellants’ Br. 3, 7, 59–61. Amazon asserts that in relying on this new argument without providing it with an opportunity to respond, the Board violated its due process rights. Parties before the Board must receive adequate notice of the issues the Board will decide and an opportunity to be heard on those issues. See Genzyme Therapeutic Prod. Ltd. v. Biomarin Pharm. Inc., 825 F.3d 1360, 1367 (Fed. Cir. 2016); see also 5 U.S.C. § 554(c). We have therefore rejected evidence and arguments supplied for the first time at oral argument. See Dell Inc. v. Acceleron, LLC, 818 F.3d 1293, 1301 (Fed. Cir. 2016) (“[T]he Board denied [party] its procedural rights by relying in its decision on a factual assertion introduced into the proceeding only at oral argument, after [party] could meaningfully respond.”). Here, however, ZitoVault advocated for the inclusion of a “pre-conversation” requirement in its patent owner response. See J.A. 237–38 (“Petitioners must demonstrate that Feinberg discloses negotiating the initiation of a stream with a defined beginning and end.” (emphasis added)). Indeed, Amazon specifically addressed this argument in its reply brief and at oral argument. J.A. 279–81, 319–20; see also Reply Br. 21–22 (conceding that “Appellants suspected ZitoVault was positioning itself to inject unwarranted additional limitations . . . . Appellants attempted to warn the Board as Board, we “decline to speculate in that regard” on this record. J.A. 22. AMAZON.COM, INC. v. ZITOVAULT, LLC 13 much . . . .”). Having had—and taken—at least two opportunities to address ZitoVault’s argument, Amazon cannot credibly argue that it was taken unaware. Moreover, as discussed, the Board rejected ZitoVault’s request to require protocol-level disclosures. See J.A. 11–15. We therefore reject Amazon’s argument that the Board denied it procedural due process. CONCLUSION We have considered the parties’ remaining arguments and find them unpersuasive. The Board’s final written decision is therefore affirmed. AFFIRMED COSTS Costs to Appellee. NOTE: This disposition is nonprecedential. United States Court of Appeals for the Federal Circuit ______________________ AMAZON.COM, INC., AMAZON.COM, LLC, AMAZON WEB SERVICES, INC., BAZAARVOICE, INC., GEARBOX SOFTWARE, LLC, Appellants v. ZITOVAULT, LLC, Appellee ______________________ 2017-2147 ______________________ Appeal from the United States Patent and Trademark Office, Patent Trial and Appeal Board in Nos. IPR201600021, IPR2016-01025. ______________________ PROST, Chief Judge, dissenting. Because I believe that the Board’s analysis of the term “sessions” was flawed, I disagree with the majority. First, Amazon never agreed that “sessions” must include protocol-level instructions for beginning and ending a data stream. Second, the intrinsic evidence does not justify reading in this limitation. I therefore respectfully dissent. In my view, the claim construction should be reversed and the Board’s analysis of anticipation and AMAZON.COM, INC. v. ZITOVAULT, LLC 2 obviousness, which relies on that construction, should be vacated. I As an initial matter, I agree with the majority that the Board’s original construction was problematic. As the Board eventually recognized, its preliminary construction was overbroad. J.A. 12. Under that construction, “sessions” encompassed any electronic movement of data—requiring nothing more than a data stream between two endpoints. See J.A. 11–12. Consequently, in a given data stream, a session could be made up of multiple packets. But it was also possible that even a single packet in that stream could constitute a session. 1 Due to the breadth of the Board’s preliminary construction, there was no way to tell which data belonged to which session. In turn, the Board’s chief concern with its preliminary construction was that it did not “allow multiple sessions to be distinguished from one another.” J.A. 12. In light of the claim language, that concern was well grounded. Particularly, claims 1, 6, 7, and 10 all refer to a plurality of “sessions.” Claim 7 refers to “simultaneous” sessions. On this basis, the Board concluded that “there must be sufficient delineation . . . that separates a ‘session’ between one client and one server and a different ‘session’ when one of the client and server are replaced, and that separates multiple sessions from each other even between the same pair of client and server.” J.A. 12–13. To illustrate the problem, the Board explored the following hypothetical: imagine a sender sends a message totaling one gigabyte. See J.A. 315–16, 327. In view of the original construction, because “every byte is an information sent over a connection,” it is unclear whether in this example there would be one session of a thousand bytes, or a thousand sessions of one byte each. J.A. 316. 1 AMAZON.COM, INC. v. ZITOVAULT, LLC 3 II Though the Board’s reason for revising its construction was sound, its implementation was flawed. In an attempt to clarify that each session must be separately identifiable, the Board improperly read in a limitation from the specification. Specifically, the Board imposed the requirement that a session have a “recognizable beginning” and a “recognizable end” of the data stream transmission. J.A. 14. At ZitoVault’s suggestion, the Board then went further. It construed the term to require specific disclosure of a “technique,” such as a protocol, to “determine[]” the begin/end point of a communication between endpoints. Id. That is, based largely on attorney argument at the oral hearing, the Board imported an additional requirement that each session have an instruction with protocollevel detail about how to initiate and terminate a data stream. 2 See J.A. 22. However, the patent does not require that all sessions have a defined instruction for beginning or ending. A system certainly could end according to a given protocol, by transmitting a termination instruction that the endpoints recognize. But a session may not have a prescribed The Board did not limit “sessions” to a specific type of protocol (e.g., Transmission Control Protocol (“TCP”)). See J.A. 14 (clarifying revised construction is not limited to “any particular technique or protocol” for “recognizing the beginning and end of a session exchanged between a transmitter and a receiver”). Rather, it concluded that a “wide variety of techniques for such beginning and ending determinations are within the scope of our interpretation of ‘session,’ including SSL and IPSec protocols disclosed in the exemplary embodiments of the ’257 patent.” Id. 2 4 AMAZON.COM, INC. v. ZITOVAULT, LLC “end.” Instead, it might terminate “if a sender stopped sending.” J.A. 950 at 18:21–22 (Rubin Dep.). The patent is agnostic about whether a session has a defined begin/end instruction in advance of the communication. Nothing in the ’257 patent excludes the possibility that a session may end simply because no more data is transmitted. Nonetheless, ZitoVault offers two main arguments to justify importing this limitation. First, ZitoVault claims Amazon agreed to this limitation during oral argument before the Board. Second, it claims the intrinsic evidence supports the limitation. The majority accepts these arguments. I cannot. I address both arguments below. A ZitoVault first argues that Amazon agreed to the Board’s narrow revised construction. Though Amazon agreed with the simple premise that all communications have an identifiable beginning and end, it consistently resisted ZitoVault’s attempts to read “sessions” as requiring details or pre-defined instructions of how and when a communication begins and ends. In its reply brief before the Board, Amazon explained: “Of course a session begins and ends at some point—that is axiomatic. But that is not the same as requiring disclosure of the details of an initiation and termination process.” J.A. 282 (emphasis added). During the oral hearing before the Board, Amazon agreed that the preliminary construction could be revised to clarify that a session has “a beginning or an end” if that “would provide the Board with comfort . . . that we can distinguish between when one session starts or one session ends.” J.A. 326. In other words, Amazon agreed that the construction could include an express acknowledgment that all sessions have a discernable “beginning” and “end.” However, at no point did it concede that the term AMAZON.COM, INC. v. ZITOVAULT, LLC 5 requires pre-defined instructions or protocols specifying when the session begins or ends. Instead, Amazon emphasized that “it is axiomatic, as we said in the papers, that a session has a beginning and an end. I don’t know that that needs to be part of the definition and it certainly shouldn’t be used as a leverage point.” Id. Consistent with that position, Amazon then explained that its expert, Dr. Rubin, opined that a session “generally refers to one or more communications exchanged between two entities over some period of time.” J.A. 327–28 (quoting J.A. 450 (Rubin Decl.)). Amazon clarified that this opinion was consistent with the Board’s original construction. See J.A. 328 (“I continue to think that is effectively what the Panel has adopted . . . .”). The Board responded: “That would seem to satisfy [ZitoVault’s] assertion it has to be well defined. That would seem to define the session. It is just that we don’t need further details as to the protocol.” Id. (emphasis added). Amazon replied: “I think that’s right. I mean, clearly Dr. Rubin believes that you would have a session over time, it is generally measured over time . . . .” Id. As shown above, Amazon simply conceded that all sessions or communications naturally involve exchange of information over some period of time. That period of time inherently has a beginning and an end. That beginning and end can be recognized or identified. Despite ZitoVault’s selective quotes from the oral hearing, however, Amazon consistently opposed the idea that “sessions” required a defined instruction governing how a data stream begins and ends. As such, I conclude that Amazon never acquiesced to this narrow construction. B ZitoVault’s second argument that the intrinsic evidence supports this construction is without merit. Neither the claims nor the specification require a technique 6 AMAZON.COM, INC. v. ZITOVAULT, LLC for beginning and ending a session. Unsurprisingly, the Board cited no clear intrinsic evidence to support such a requirement. See generally J.A. 11–15. At bottom, the Board’s construction improperly reads in a limitation from the specification. The focus of the claimed invention is not some improvement on protocols for how sessions begin or end. Instead, the ’257 patent focuses on how to “scale” a network involving multiple sessions. ’257 patent col. 4 ll. 45– 51 (“[A]n aspect of the invention provides a distributed software solution for encryption/decryption which is infinitely scaleable in the number of simultaneous sessions capable of being processed by a server . . . .”); id. at col. 11 ll. 8–11 (“This is a new and non-obvious application of distributed algorithm technology as applied to encryption and decryption and greatly enhances scalability.”). Unsurprisingly, then, the claim language does not focus on the need for specific techniques governing when a session begins or ends. The only claim to even recite “terminating” a communication is dependent claim 9. Claim 9, which is not at issue, adds the required step of “terminating encrypted session communication upon successfully transferring a session from main server to one or more agents.” No other claim mentions terminating a session. As the majority notes, the specification does provide a single example of how a session might be established and terminated. This example, however, is provided “[i]n accordance with an aspect of [the] invention.” Id. at col. 6 ll. 2–48 (discussing how the preferred embodiment in Figure 4 establishes a session); id. at col. 8 ll. 23–27 (discussing how the same embodiment terminates a session). Dependent claim 9 appears to claim the preferred embodiment of Figure 4. As no other claim even discusses “terminating,” I decline ZitoVault’s invitation to AMAZON.COM, INC. v. ZITOVAULT, LLC 7 read the preferred embodiment of Figure 4 into the rest of the claims via the term “sessions.” III As discussed above, the claims require “sessions,” which in turn requires that sessions can be delineated from one another. Consistent with the intrinsic evidence, the term “sessions” should be construed as: “Sets of data exchanged between endpoints, wherein each set is exchanged over a period of time and is uniquely identifiable.” The level of detail required by the claims is simply whether a data set exists that is identifiable or distinguishable from other sets in a data stream. There is no specific restriction on how the set of data is identified. For the reasons above, I believe the Board’s construction adopted by the majority was flawed. As a result, the Board’s rulings on anticipation and obviousness should be vacated, and the case should be remanded for further proceedings consistent with the proper construction of the term “sessions.” Therefore, I respectfully dissent.