How Selecting The Wrong Prior Art References Will Doom An IPR
The Federal Circuit recently affirmed a Patent Trial and Appeal Board (“PTAB”) inter partes review (“IPR”) decision in Palo Alto Networks, Inc. v. Finjan, Inc., No. 2017-2059, holding that the PTAB did not err in concluding that a person of ordinary skill would not have combined certain prior art identified by Palo Alto Networks, Inc. (“PAN”) in a way that would teach a claim limitation common to the challenged claims. The Federal Circuit’s opinion underscores that the prior art references selected for assertion by an IPR petitioner are the key to any IPR petition’s ultimate success.
As related in the opinion, on September 30, 2015, PAN petitioned the PTAB for an IPR of claims 1, 2–7, 9, 12–16, 19–23, 29 and 35 (“challenged claims”) of U.S. Patent No. 8,225,408 (the “’408 patent”) issued to Finjan, Inc. (“Finjan”). The ‘408 patent relates to “methods and systems for detecting malware in data streamed from a network onto a computer,” by building “a parse tree data structure” from the “incoming content” that is evaluated for “specific patterns of content that indicate malware.” The following limitation is common to the challenged claims:
dynamically building, by the computer while said receiving receives the incoming stream, a parse tree whose nodes represent tokens and patterns in accordance with the parser rules
After institution, the PTAB adopted PAN’s unopposed construction for “dynamically building” based on the plain claim language, to mean “requires that a time period for dynamically building overlap with a time period during which the incoming stream is being received.” PAN had alleged that the challenged claims were invalid on two grounds: (1)that the claims would have been obvious over U.S. Patent No. 7,636,945 issued to Chandani (“Chandnani”) and U.S. Patent No. 5,860,011 issued to Kolawa (“Kolawa”); and (2) that the claims would have been obvious over Chandnani, Kolawa, and U.S. Patent No. 7,284,274 issued to Walls (“Walls”).
The Chandnani patent was the focus of the PTAB’s decision. Chandnani teaches a method of “detecting malware in a data stream, including determining the programming language of the data stream and detecting viral code.” Chandani also “teaches tokenizing the incoming data stream by breaking it into smaller pieces known as tokens.” Chandani summarizes its tokenizing procedure as follows:
To tokenize the data stream, a script language used in the data stream is determined using the language check data. The data stream is analyzed using the language check data to select the language definition data to use for the detection process. Next, the selected language definition data and the data stream are supplied to the lexical analyzer. The data stream is lexically analyzed again, this time using the language definition data, to generate a stream of tokens. As mentioned above, each generated token corresponds to a specific language construct, and may be a corresponding unique number or character.
The PTAB determined that the dispositive issue was whether PAN demonstrated that the prior art teaches or suggests “dynamically building” a parse tree “while” receiving an incoming stream of program code as required by its construction of “dynamically building.” The PTAB found that Chandnani did not teach the “dynamically building” limitation of the ’408 patent because it does not demand or even imply that the data stream is being received while being tokenized.
On appeal, PAN challenged the PTAB’s interpretation of the Chandnani reference. The Federal Circuit held that the PTAB’s finding was “supported by substantial evidence because the reference itself, by using the word ‘again,’ indicates that the data stream is lexically analyzed more than once and not simultaneously.” The Federal Circuit also found that the PTAB did not err by relying on Finjan’s expert, Dr. Medvidovic, who opined that:
simply because Chandnani’s tokenizer operates on a data stream does not demand or even imply that the data stream is being received while being tokenized…
Chandnani would still temporarily store the entire data stream in memory at least between the first and second lexical analyses.
PAN also argued that the PTAB erred by “analogizing Walls to Chandnani and by not meaningfully reviewing Walls as a separate reference that discloses the dynamically building limitation.” In response, the Federal Circuit concluded that the PTAB “did not fail to meaningfully consider the teachings of Walls” because the PTAB “considered the disclosure of Walls and the expert testimony regarding Walls from both parties” and “performed its own review of Walls” before reaching its decision.
PAN further argued that the PTAB “erred by not considering particular cross-examination testimony from Finjan’s expert witnesses when analyzing whether the prior art taught ‘dynamically building’.” In response, the Federal Circuit noted that the PTAB “acknowledged Dr. Medvidovic’s cross-examination testimony and explained its reasoning for why Chandnani does not disclose the required claim limitation.” In reaching its decision, the Federal Circuit relied on its prior decision in PGS Geophysical AS v. Iancu, where it held that “we may not supply a reasoned basis for the agency’s action that the agency itself has not given, [but] we will uphold a decision of less than ideal clarity if the agency’s path may reasonably be discerned.”
TAKEAWAY: This case presents a further example of the Federal Circuit’s deference to the PTAB’s evaluation of substantive evidence in IPR proceedings and to the PTAB’s decision-making processes around such evaluations. Further, practitioners who anticipate challenging a patent via an inter partes review should exercise great care in ensuring that identified prior art in their petition is consistent with their claim constructions.