Before Dyk, Plager, and Stoll. Appeal from the Patent Trial and Appeal Board.
Summary: An IPR petitioner relying on a prior-art algorithm is not required to discuss all potential permutations in its petition, and does not raise “new arguments” on reply by disclosing a different example of the same algorithm.
Andrea sued Apple for infringement of a patent relating to reduction of background noise in audio signals. Apple thereafter filed two IPR petitions. The PTAB instituted both petitions and consolidated proceedings. In one of the IPR’s, Apple argued that an algorithm in one of its prior art references disclosed the required limitations of the asserted claims. In support, Apple’s expert provided an example using the prior-art algorithm which assumed there were no “sub-windows” (units relating to the amount of signal being analyzed). Andrea responded that Apple’s argument was flawed because that prior-art algorithm requires at least multiple sub-windows, and even in view of multiple sub-windows, the algorithm failed to disclose the claimed limitations. On reply, Apple submitted an expert declaration discussing an example of the prior-art algorithm with multiple sub-windows, which showed that even in the multiple sub-windows context, the algorithm still disclosed the claimed limitations. The Board declined to consider Apple’s responsive arguments regarding multiple sub-windows, finding that the arguments raised a new theory of unpatentability for the first time in reply, and held the relevant claims were not invalid.
On Appeal, the Federal Circuit vacated the Board’s decision and remanded for consideration of the arguments raised by Apple in its reply. The Federal Circuit noted that Apple’s reply asserted the same legal position (that Andrea’s claims were obvious over the prior art), relying upon the same prior art algorithm, to arrive at the same result (that Apple’s prior art reference disclosed the required limitations). The court distinguished this from situations where an IPR petitioner relies on “previously unidentified portions of a prior-art reference to make a meaningfully distinct contention” or cites “new non-patent literature references.” The panel held that IPR petitioners relying on an algorithm cannot be required to discuss all potential permutations in their petition or risk waiving the opportunity to discuss other examples in reply.