Sunday, June 30, 2013

Qualcomm may have to modify Snapdragon chip for Nokia to work around HTC graphics patent

On Friday the Mannheim Regional Court held a trial on HTC subsidiary S3 Graphics' patent infringement complaint against Nokia over the Qualcomm Snapdragon chip. As counsel for S3G confirmed when Judge Voss ("Voß" in German) asked about the possibility of a settlement, this lawsuit is "part of the wider dispute between HTC and Nokia". Between the U.S., UK and Germany, Nokia is asserting roughly 50 patents against HTC, and HTC (including S3 Graphics) has brought at least five German countersuits.

The longer I've been watching HTC's patent lawsuits, the more impressed I am with its handling of patent disputes. I admit that I underestimated HTC, as many others did, but I've come to recognize its defensive skills beyond stalling (a tactic at which HTC undoubtedly excels). And while S3 Graphics' first ITC complaint against Apple was dismissed in 2011 and a second one wasn't adjudged prior to the parties' settlement, this acquired patent portfolio may prove helpful to HTC in settlement negotiations with Nokia. Based on how the Friday trial went, I think HTC is more likely than not to win a German injunction against Nokia devices incorporating the Qualcomm Snapdragon chip, though the leverage this gives HTC will depend on whether the court agrees with its preferred or merely its alternative infringement theory. In the latter event, I believe leverage will be rather limited because Qualcomm would just have to remove a feature that Nokia doesn't use anyway, and the primary effect of a win for S3G would be that HTC could point to it whenever Nokia says HTC should cease its infringement of Nokia's patents.

The patent-in-suit is EP0797181 on "hardware assist for YUV data format conversion to software MPEG decoder". For information on the analog YUV format and the related YCbCr digital format, which is often referred to by the term YUV as well, let me refer you to Wikipedia (YUV, YCbCr) and provide my own condensed explanation:

While computers generally process and store images in a bitmap format (from left to right, from top to bottom), YUV is a format that separates for each pixel a luminance (brightness, Y) value from two chrominance (color, U and V) values. So you get one matrix of Y values (which would be enough to produce a black-and-white picture), one of U values, and one of V values. It's easy to convert a YUV image into a bitmap, but in connection with high-resolution video with a high frame rate, this transformation can put a drain on the central processing unit. The patent-in-suit covers the idea of assigning this task to a coprocessor. (Whether this idea should entitle someone to a patent is another question, a question that is not unique to this patent or company.)

Even though Nokia's litigators led by Bird & Bird's Boris Kreye (the same team is also defending Nokia against HTC's power management patent in four parallel actions) have developed various credible defenses, S3G (represented by Wildanger's Wolf Graf von Schwerin and Peter-Michael Weisse) appear to have persuaded the court of key parts of their infringement contentions.

I considered two of Nokia's defenses particularly important. One of them was discussed only toward the very end of the trial and relates to the claim element of a "predetermined address range", within which certain "video data addresses" must be. Nokia says that the accused technologies merely transfer information enabling the chip to determine the addresses, but that there's no separate communication or computation of a "predetermined address range". Without seeing the detailed infringement contentions I can't rule out that S3G's claim construction effectively renders the "predetermined address range" limitation meaningless. However. the court did not indicate that this issue might be outcome-determinative.

The more fundamental issue is the one that makes all the difference between S3G's primary and secondary infringement contentions. The specification of the patent, which is supposed to serve as its own dictionary, defines "component YUV format" as a planar format, meaning that there's one contiguous set of data for the Y, U and V values. But Nokia avers that its products only process something that can be described as pseudo-planar: instead of separate U and V planes, there's a single data block containing a pair of U and V values for each pixel. S3G's key argument in this regard is that the claim only requires "at least one second contiguous block of data of a second component" (as opposed to requiring two blocks) to be transmitted, and the only way in which this can make technical sense if there's a single block containing U and V data. But according to Nokia this still doesn't change the meaning of "YUV component data".

I saw some merit in both parties' claims. On the bottom line I would be inclined to agree with Nokia that the specification clearly defines "YUV component data" as a planar format and to attach more importance to this than to the "at least one ... block" argument. It appears to me that a significant part of the transformation the patent covers has occurred, or need not occur, if U and V data are already kept in a single block of paired values.

S3G has an alternative infringement theory for the event that Nokia's "not planar" defense succeeds: this is a hardware patent, and the hardware built into the accused Nokia phones can be configured to process fully planar data as well. There are registers, and the operating system (Windows Phone) could, though it doesn't at this stage, instruct the chip to handle planar data. S3G argues that a hardware patent claim like this is infringed if the capability is provided, even if not utilized in practice. The parties agree that hardware can't run without software. Still, S3G argues that shipping certain hardware functionality constitutes infringement -- and in this case, considering the structure of the claim-in-suit, I tend to agree with S3G.

If the court rules in S3G's favor based on its primary infringement theory (of pseudoplanar data falling within the scope of the claim), S3G's corporate patent, HTC, gets considerably more mileage out of this lawsuit than otherwise. If the court finds Nokia to infringe only to the extent that never-actually-utilized hardware functionality is shipped, it would be a minor tweak for Qualcomm to remove the "planar" option, such as by eliminating the related register or by having the chip ignore it (so as to always require data to be provided in a pseudoplanar format). There would be no performance impact. A special version of the Snapdragon chip for Nokia to use in Germany might affect Qualcomm's manufacturing costs as compared to a one-chip-fits-all scenario, but at the end of the day Qualcomm will want to ensure that its chips don't expose its customers to major IP liability issues if those can be avoided.

In March Nokia won its first German injunction against HTC over a power-saving patent, and HTC also worked around it, presumably by removing certain functionality without a non-infringing replacement delivering the same benefits. Here, the related functionality can always be removed, but if HTC's primary infringement theory succeeds, then there will be some impact on video processing performance. If a patent injunction issues, there will always be a workaround unless the patent-in-suit is a truly standard-essential patent. The key question is then whether the workaround is merely an internal thing or has effects (such as a degradation of battery life or video performance) that end users will notice.

A decision, which may or may not be a final ruling, has been scheduled for September 20, and if S3G wins, it will be interesting to see what Nokia says about a workaround.

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: