Federal Circuit Addresses Indefiniteness of Disclosed Algorithms in Means-Plus-Function Claims

By Shumaker & Sieffert, P.A.

In Alfred E. Mann Foundation for Scientific Research, Advanced Bionics, LLC v. Cochlear Corporation, NKA Cochlear Americas, Cochlear Ltd.[1], the Court of Appeals for the Federal Circuit addressed various issues resulting from infringement litigation with respect to U.S. Patent No. 5,609,616 (‘616 patent) and U.S. Patent No. 5,938,691 (‘691 patent) assigned to Alfred E. Mann Foundation (hereinafter, “Cross-Appellant”).  This article focuses on holdings of indefiniteness of claim 6 of the ‘691 patent and definiteness of claim 1 of the ‘616 patent.

“The patents are directed to an ear implant with telemetry functionality for testing purposes, and generally describe a two-part system comprising an external wearable system with a wearable processor (WP) and headpiece, and an internal implantable cochlear stimulator (ICS).”[2]  In relevant part, claim 6 of the ‘691 patent states:

A cochlea stimulation system, comprising:

[a] audio signal receiving means;

[b] an externally wearable signal processor (WP) for receiving and processing the audio signals received by the audio signal receiving means and including means for generating data indicative of the audio signal; … ”[3] ​

Both sides agreed that “means for generating data indicative of the audio signal” is a mean-plus-function feature, and that the corresponding structure is a microprocessor.[4]  At issue, however, was whether the disclosure provides sufficient structure to render claim 6 definite.

“To satisfy the definiteness requirement, a means-plus-function claim requires sufficient disclosure of the underlying structure.”[5]  When the structure is a general purpose computer or microprocessor, the underlying structure is defined by the algorithm, and an “algorithm[s] in the specification need only disclose adequate defining structure to render the bounds of the claim understandable to one of ordinary skill in the art.”[6] 

For the means for generating data indicative of the audio signal, Cross-Appellants argued that the “microprocessor implements a logarithmic conversion algorithm to generate data indicative of an audio signal.”[7]  The Cross-Appellants further argued that “the algorithm performed by the microprocessor has two steps: first, the microprocessor receives digital data from the A/D converter 28, and second, the microprocessor uses a logarithmic conversion function to format the data.”[8]

One issue was a dispute over whether the specification truly describes where the logarithmic conversion function takes place, and another issue was a dispute over whether the specification describes how the logarithmic conversion function could be implemented.[9]  The issue of whether the specification states with sufficient specificity that the microprocessor implements the logarithmic conversion is less interesting than the second issue of whether the algorithm of the logarithmic conversion function needs to be described or was described sufficiently.

The Federal Circuit stated that claim 6 is indefinite because “the logarithmic conversion may be implemented through various unspecified algorithms.”[10]  The Federal Circuit further stated that “As the testimony reflects, the ‘691 patent does not disclose an algorithm, or even a small set of algorithms for performing the claimed logarithmic conversion function.[11]  The Federal Circuit noted that “[a]lthough Cross-Appellants argue that a person of ordinary skill in the art would know of potential logarithmic conversion functions to implement…this does not create structure in the patent where there was none to begin with.”[12]  In essence, the Federal Circuit held that even though algorithms to implement logarithmic conversions are known, because one was not described, there is no corresponding structure and claim 6 of the ‘691 patent is indefinite.

The Federal Circuit then turned to claim 1 of the ‘616 patent, which recites in relevant part an “external processor means coupled to the transmitting means of the external head-piece/transmitter for receiving and processing the status-indicating signals to derive information therefrom regarding the operation of the implanted stimulator and its plurality of tissue stimulating electrodes.”  Again, there was no dispute that this is a means-plus-function limitation, and that the microprocessor is the corresponding structure.

At the district court, Cross-Appellants argued that the patent discloses a two-step algorithm, where first, the microprocessor accepts signals representative of voltage, and second, the microprocessor applies Ohm’s law to convert the voltage into an impedance value.[13]  The district court rejected these arguments because the patent does not explicitly identify Ohm’s law and there are multiple ways of calculating impedance.[14]

On this point, the Federal Circuit disagreed with the district court, noting that the specification for the ‘616 patent does describe that impedance is determined based on voltage and current.[15]  However, the Federal Circuit also relied on the fact that “[b]oth parties’ experts testified that a person of ordinary skill would know to apply Ohm’s law to voltage and current to yield impedance values.”[16]  The Federal Circuit on this basis found claim 1 of the ‘616 to be definite.

It is curious why the Federal Circuit, with respect to claim 1 of the ‘616 patent, would even note that Ohm’s law is well known to one of ordinary skill in the art for finding claim 1 definite.  Unquestionably, Ohm’s law is extremely well known, as a basic principle of electricity.  However, the degree to which an algorithm is well known seems to be a factor that the courts, correctly or incorrectly, do not evaluate.  Indeed, the Federal Circuit in the same decision notes that although a person of ordinary skill in the art would know of potential logarithmic conversion functions to implement, this does not create structure in the patent where there was none to begin with.[17]  Notably, in making such an assertion, the Federal Circuit makes no finding as to whether potential logarithmic conversion functions are extremely well known or not, but with respect to Ohm’s law, the fact that it is well known is a factor in determining that claim 1 of the ‘616 patent is definite.

On first read, it appears that the analysis with respect to claim 6 of the ‘691 patent (claim found indefinite because no structure) and claim 1 of the ‘616 patent (claim found definite because sufficient structure) cannot be squared.  Patent practitioners may be left questioning whether an algorithm is sufficiently well known to merit analysis similar to claim 1 of the ‘616 patent or is not sufficiently well known such that an analysis similar to claim 6 of the ‘691 patent is applied. 

However, the decisive point may not be that Ohm’s law is well known but, rather, that there is really only one way to determine impedance while various ways exist to implement a logarithmic conversion algorithm.  For example, the Federal Circuit, quoting testimony from the district court cases, states that “[Impedance] is always calculated based on the ratio of voltage to current. One of ordinary skill in the art would readily understand from the disclosure in the ‘616 patent that this [sic] the algorithm is implemented.”[18]  The fact that there is one way to determine impedance may have been the reason why Ohm’s law being extremely well known was a factor in the finding of definiteness of claim 1 of the ‘616 patent.  Contrariwise, the fact there are multiple ways to implement logarithmic conversion may have been a factor in the finding of indefiniteness of claim 6 of the ‘691 patent.

In writing the specification, patent practitioners should be mindful to describe all operational steps even if well known.  If it is likely that if there is only one well known way to perform a function, then full disclosure of that algorithm may not be needed.  However, to minimize the risk of indefiniteness, patent practitioners should consider including the algorithm.  The penalty of indefiniteness is too great. When there are various possible algorithms, even if well known, patent practitioners should consider including at least one example algorithm.  This way, there will be at least some corresponding structure as the courts appear hesitant to “create structure in the patent where there was none to begin with.”[19]

In addition to reading the majority opinion, patent practitioners should take time to read the dissent, which argues that claim 6 of the ‘691 patent is definite.  To summarize, the dissent notes that there is no dispute that logarithmic conversion algorithms are well known, and disagrees with the majority that the claim is indefinite because there are multiple logarithm algorithms and none are described in the specification.[20]

The dissent makes the point that “[p]recedent does not require that well-known formulas must be stated in the specification, when they are known in the relevant art,”[21] that “[a] known procedure is not rendered indefinite when there is more than one known way of carrying it out,”[22] and that “[m]y colleagues’ holding that it was necessary to state which of the two or three known logarithmic conversion routines was used, on pain of invalidity, is unsupported by mathematics, reason, or precedent.”[23]  The dissent states that “[t]he court errs in finding [the claims] invalid of indefiniteness, upon new and ill-defined requirements for patent specifications and pitfalls with no benefit to anyone,”[24] and concludes the indefiniteness analysis with: “As computer-implemented technology continues to provide new public benefits, consistency of judicial view is essential to stability of the law and progress of the technology.”[25]

The law of indefiniteness appears to remain in flux.  Patent practitioners should keep up with best practices as there may be pitfalls to come.