Over the past several years, there has been an increased demand for fast, effective, and lower cost antipiracy solutions, specifically watermarking. This demand is largely driven by video service providers that are now delivering more premium content via the Internet and shifting their delivery strategies to target OTT/D2C services. Delivering content via the Internet can make it easier for pirates to take the illegal action of re-streaming a live video stream acquired from a copyrighted video service, since any computer can get, capture, distort, and forward the video stream. To combat this, advanced video service providers and content owners are investing in cost-effective methods to prevent and mitigate piracy with a fast response, cutting pirates’ access to the content and service.
A watermark-based anti-piracy solution requires the legal video service provider to embed a unique watermark in each video stream delivered to each customer. This unique pairing of watermark and customer enables the identification of the pirate in illegal platforms. While this approach has demonstrated its efficiency, it suffers from a weakness: embedding a unique watermark in every video stream is complex in a massive content delivery network (CDN). Today, the watermark is typically inserted prior to encoding and packaging. This approach requires two copies (A and B), which double the storage and computation costs. The CDN edge server then alternatively picks either A or B video segments to generate a unique sequence of A/B watermarks in each stream.
Unfortunately, the A/B watermark solution has three critical limitations: (1) it is resource hungry. The solution doubles the bandwidth in the CDN and doubles the memory in the edge servers. (2) it is slow. To generate a unique sequence of A/B watermarks for every end-user in a population of millions of clients, the CDN should deliver a dozen minutes of video content. (3) it is complex to implement, since it requires time synchronization between multiple entities (edge server, subscriber management server, video packagers) to guarantee the match between the subscriber identifier and the A/B sequence.
In this paper, we will discuss an alternative way to scale the delivery of unique watermarks. This approach leverages the concept of edge computing, where the CDN brings more intelligence and more capacities to the edge servers, beyond just being a cache function. This solution enables the edge server to modify the video content before delivering it, to embed the watermark on-the-fly. This approach is more efficient because it addresses both the compressed and encrypted domain, so the edge server does not have to decode and re-encode the video segments, nor to decrypt and re-encrypt – saving both time and costs.
Additionally, the solution requires only one version of the video content in the edge, which fixes the A/B resource consumption limitation. To embed the right watermark on the video content, the edge server follows instructions, which have been inserted into the video package. Thus, the pipeline is simpler to implement and fixes the A/B complexity limitation. Finally, it is possible to embed more information than just A or B in a video segment, so the pirate can be identified faster. This feature is a key solution to the A/B detection time limitation.
In this paper, Synamedia’s video network and content protection experts will share the significance of edge watermarking and the value of this favored and cost-effective approach.