Saturday, June 15, 2013

German VP8 infringement cases show Google's inability to cut through the codec patent thicket

I don't think any responsible standardization body will be able anytime soon to declare VP8 a "royalty-free" codec. Everyone in the industry knows -- though a few deny against better knowledge -- that there's a whole patent thicket surrounding video compression techniques, and that all codecs implement more or less the same fundamental concepts. Only a fool could believe Google when it originally claimed that there were no issues concerning third-party rights. Google's license deal with 11 MPEG LA contributors comes with terms that Google certainly wouldn't have accepted if it had not been seriously concerned about the risk of litigation brough by those right holders. And with Nokia refusing to make its patents available for VP8 on royalty-free or even FRAND terms (Nokia reserves the right to seek injunctions, or to charge as much as it wants, no matter how much above a FRAND rate it might be), there's at least one company that is actually suing over VP8.

Nokia brought its latest patent assertion against VP8 as part of an ITC complaint and its first two infringement actions over VP8 in Mannheim, Germany, and this post is mostly about the trial held on Friday over the second one of these cases. Also, since my last post on VP8 I've had the opportunity to read the full text of the order to reopen the proceedings in the first case. As I wrote two weeks ago, that order wouldn't have come down without at least a finding of a likely infringement. It relates to the possibility of staying the case pending a parallel nullity proceeding, and German courts don't reach the question of a stay if they can outright dismiss a case for lack of infringement. The order says one can conclude that devices implementing VP8 infringe EP1206881 if the asserted claim (claim 46) is interpreted "broadly", in which case there is room for invalidity contentions (which, as the order doesn't say explicitly because it's clear to patent litigators, wouldn't be relevant to a narrower claim construction). Therefore, the court now wants to take a closer look at a particular invalidity theory. While the court could still reach a different conclusion on infringement, it would have dismissed the case immediately (to conserve court and party resources) if Google and HTC had convinced the court that Nokia's claim construction is overbroad. Absent a relatively unlikely change of mind by the court, the likely outcomes are a stay or an injunction. A stay would mean that an injunction will issue if the asserted patent claim survives the nullity proceedings in its granted form, or in a form that still warrants an infringement finding. This means that Google won't have legal certainty concerning the German part of EP'881 (courts in other European countries will not be bound by the related German decisions) for years to come.

Yesterday's patent-in-suit, EP1186177 on a "method and associated device for filtering digital video images" is less likely to be deemed infringed when the Mannheim court announces a decision on August 2, but Nokia can and presumably will appeal a dismissal, and despite the inclination indicated by the court at yesterday's trial, EP'177 is indicative of VP8's patent issues because it's within striking distance of VP8: the whole case hinges on only one claim element. All other claim limitations are considered satisfied by HTC's VP8-implementing devices. That fact reduces to absurdity all those claims that VP8 is so very original that no third party patented anything that might read on it.

While video compression techniques involve advanced mathematical operations, it's actually quite easy to explain the only claim limitation that may (and, apart from the possibiltiy of an appeal, probably will) help Google and HTC avoid an infringement finding:

Video codecs -- and this is true of H.264 as it is of VP8 -- divide an image into macroblocks (rectangular areas). Each macroblock is processed separately. End users just see the overall image and shouldn't notice the edges of macroblocks, which are only an internal thing. But they would notice those borders if certain mathematical operations resulted in artefacts (unwanted elements that appear in an image but don't exist in reality) that a codec failed to filter out. The patent relates to this filtering process. The codec looks at the overall image and identifies actual vertical and horizontal lines that were filmed (for example, a corner of a building). Only artefacts at the edges of the macroblocks are filtered out; real-world lines are preserved.

For each macroblock there are different compression methods. The patent mentions "intra coding", "copy coding", "motion-compensated prediction coding", and "not-coded coding". Google and HTC can't dispute that VP8 uses all of these. But their non-infringement argument, which the court views favorably, is that VP8 does not "determin[e] at least one parameter of the filtering operation based on the types of the first and second prediction encoding methods". They say VP8 analyzes the pixels of two neighboring macroblocks, but it does not make a determination by directly checking on which encoding methods were used for the two blocks in question. Nokia, however, says that the pixels result from processing of a macroblock using a certain encoding method, and that there are correlations between the encoding methods used and at least one parameter (they particularly stressed the number of pixel rows/columns subjected to filtering).

At German patent trials claim construction and infringement analysis usually aren't separated as systematically as in U.S. lawsuits, in which the court firstly provides interpretations of disputed terms (which are later put before a jury if a jury trial is requested, which is the case in most U.S. patent infringement cases). In a U.S. proceeding over the same patent (I couldn't find a granted U.S. equivalent of this particular one, but let's assume so for the sake of the argument), the term "determining at least one parameter [...] based on the types of the first and second prediction encoding methods" would have resulted in a claim construction dispute, with Google and HTC trying to defeat Nokia's infringement theory by inserting into the court's construction of the term a word like "directly", while Nokia would probably have argued that the plain and ordinary meaning is fine and has scope for its infringement theory, possibly also proposing an alternative like "depending on" (for "based on"), which is even broader. Additionally or alternatively, the defendant and the intervenor would have tried to have "the first and second prediction encoding methods" interpreted as "the types of the first and second prediction encoding methods".

At yesterday's German trial the question was not how to rephrase the term, but the analysis that had to be performed was the same. Both parties advanced reasonable arguments. Google's counsel pointed to items (a) and (d) in paragraph 24 of the patent specification. Item (a) is "the type of block on either side of the boundary", while (d) relates to "differences in pixel values", which are only influenced by the encoding method types. Google plausibly argues that the description of the patent therefore distinguishes between these two (and two other) criteria. So far, so good. It makes a lot of sense, but it doesn't prove beyond reasonable doubt that the disputed claim term relates to only item (a) and not item (d). Nokia pointed to other parts of the description which, such as paragraph 37, show that the patent discusses an analysis of pixel values.

In order to back up its "royalty-free" claims, Google needs legal certainty, but there are just too many patents in this field covering the same or at least very similar techniques as those used by VP8. There's clearly a lack of certainty if the outcome of a single infringement case depends on whether "based on" can or cannot be reasonably understood to have the meaning of "depending on". And as I wrote further above, the first finding of a likely infringement has already been made in connection with another Nokia patent.

Nokia and Google may also have some patent issues to sort out over Google Maps according to a TechCrunch article. At this stage I just wanted to mention this nonjudgmentally.

If you'd like to be updated on the smartphone patent disputes and other intellectual property matters I cover, please subscribe to my RSS feed (in the right-hand column) and/or follow me on Twitter @FOSSpatents and Google+.

Share with other professionals via LinkedIn: