Your browser is not supported. Please update it.

8 January 2015

All set tops chips need watermarks after Movie Labs mandate

It is over a decade since forensic watermarking was first deployed to protect video content, but the field has only really taken off following Movie Labs’ decision to mandate it for UHD content as of April 2014.

Momentum was already growing as a result of OTT proliferation and the corresponding rise in content redistribution as a piracy threat, for both on-demand and live material. But the Movie Labs’ mandate galvanized activity and now after the latest flurry of announcements we can say that all major set top UHD SoCs, notably those from Broadcom, ST Micro and Sigma Designs, will soon have embedded watermarking.

The mandate was driven by the much greater opportunity UHD will present to pirates for redistributing high quality content, either by bypassing security controls or even just streaming directly via a camcorder. For movie content, watermarking enables illegal streams to be tracked to the source and cut off hopefully before too many infringing copies have been distributed.

Meanwhile for live content, most notably premium sports, the challenge is even more acute, since illicit streams that may be derived “legitimately” by pirates taking out subscriptions using stolen credit cards, need to be taken down within minutes to avoid significant leakage, given that the content  value window only lasts for the duration of the event.

The main focus of recent announcements is on recorded content such as movies or TV series, rather than live sports, even if we are talking about linear broadcast. For example Verimatrix has announced a fresh collaboration with Broadcom to embed its Videomark forensic watermarking technology in the latter’s UHD SoCs. This will be part of a wider collaboration between the two firms to develop a complete revenue protection system based on watermarking and encryption.

As both companies admitted, the collaboration is to ensure compliance with MovieLabs’ UHD premium service requirements around watermark based content tracking. The key point of watermarking is that it enables individual streams and sources of content to be uniquely identified, in principle, in near real time. There are still debates over the precise watermarking technology used, with varying levels of robustness against tampering and also techniques for responding to attacks on the security system itself. But the underlying principle has been established, with forensic watermarking accepted as offering demonstrable protection against redistribution of content after it has been decrypted (although pirates themselves may choose to re-encrypt the content to protect their own “revenue”).

It is in fact nine years since Verimatrix and Broadcom first collaborated over forensic watermarking. In May 2006 Verimatrix declared that integration of Videomark with Broadcom’s then set top reference design platform BCM97398 had yielded the first ever robust and yet invisible system capable of tracking content to individual set tops. This was almost a year after Verimatrix had licensed patents from Digimarc, an Oregon based developer of security products, under a multi-year agreement.

The obvious question is why forensic watermarking did not take off at that stage. One reason is that it was then a solution looking for a problem. Watermarking then was pitched not at OTT, which barely existed, nor even particularly IPTV, which was in its infancy, but at traditional cable and satellite services, where the main piracy threat was smart card cloning. Even control word sharing was only just becoming a significant issue, enabled by growing availability of broadband Internet access.

The other main impediment back in 2006 and for several years afterwards was that pay TV operators were not geared up for the deployment of watermarking, largely because the infrastructure was not ready. Watermarking by itself does not make content secure, any more than an unbreakable encryption algorithm. It all depends on how it is implemented and integrated with the logistical and human processes involved. In the case of watermarking, there has to be an associated robust network forensics system enabling content to be searched, located and identified on the basis of the marks and this just did not exist in 2006.

Watermarking has an overhead and cost, requiring additional steps in the content workflow and greater encoding complexity, and using more bandwidth. Until very recently the overhead of watermarking has outweighed the perceived benefit, especially for live streaming, although all of a sudden there has been an explosion of interest in that. Just about every major pay TV operator wanting to distribute premium sports content online are now evaluating options for real time forensic watermarking.

But to date watermarking has been largely confined to catch-up and VoD applications where the cost to implement is less. Deploying watermarking as part of a non-live workflow, even within a linear service, is significantly simpler. Another issue that has arisen with non-live content is that even when pirates have been detected, they still often evade prosecution. This may be because they are in countries beyond normal jurisdiction or just because they are individually not sufficiently big to be worth pursuing.

However all this is changing, with signs of growing collaboration between relevant parties, including governments, peering companies, CDN providers and ISPs, exploiting the fact that ultimately streams, illicit or otherwise, have to traverse legitimate infrastructure to reach users and so can potentially be trapped. This too is helping to stimulate development of more advanced forensic watermarking technologies, as well as the associated networking and database systems, to unlock the full potential.

Against this background, Civolution, another of the major watermarking technology providers, has just announced integration of its technology in chipsets from both Sigma Designs and ST Micro.  Sigma Designs has integrated Civolution NexGuard forensic watermarking technology in its latest STV7701 and STV7703 Smart TV chipsets. These SoCs support 4K Ultra UHD resolutions for both HEVC (H.265) and VP9 codecs, as well as high dynamic range (HDR) technology.

At the same time, also at this week’s CES in Las Vegas, ST Micro has announced the integration of Civolution’s NexGuard forensic watermarking in its UHD Cannes and Monaco SoCs. All these parties of course also cite the MovieLabs guidelines as a prime motivator, highlighting how this has changed the focus of watermarking over the last year or two. One interesting wrinkle here is that Broadcom announced integration of Civolution watermarking in April 2013 at the NAB Show, but for automatic content recognition, not for combating piracy. Watermarking is being deployed for content recognition in targeted advertising and second screen synchronization, but now the biggest emphasis is back where it started, on forensics.