M&E Journal: Implementation Challenges, Opportunities in Forensic Watermarking

Rajan Samtani, Graham Oakes — Media Science International

Digital content capture and distribution technologies have evolved at a rapid pace as evidenced by the recent push by CE manufacturers, OTT distributors and content owners to move from HD to 4K/UHD and HDR. Content creators and distributors use these new technologies to produce and distribute increasingly compelling and valuable content for consumers at much higher resolution and fidelity, i.e. more, as well as better, pixels.

Amid these improvements in the quality of the viewing experience, the business models in both movie and TV distribution are also under pressure to shrink the available “windows” of distribution as we make the transition from closed, managed service models for licensed content to open, over-the-top models, as well as internet and mobile distribution.

Unfortunately, digital content distributed at very high fidelity within shrinking windows and viewed on highly capable display devices creates an ideal environment for content pirates, who can quickly create very high quality copies using simple techniques that take advantage of the analog hole (i.e. using a camcorder to record content off a display device, or readily available digital-to-analog signal multiplexing techniques), thus greatly increasing the risk of unlawful reproduction and redistribution.

Based on video piracy statistics for SD and HD content available from multiple sources, it is clear that the threat of piracy from BitTorrent, on-demand cyberlockers, streaming sites and dedicated IPTV-oriented pirate services remains rampant and will only increase when content is widely available in UHD.

In anticipation of this threatening “perfect storm,” MovieLabs in April of 2014 published a new set of Enhanced Content Protection (ECP) guidelines–defining requirements and best practices to ensure the viability of these technical and creative advances through the addition of more stringent policies related to content protection methods required by licensees and distributors. These included hardened security in hardware and trusted software along with the use of forensic watermarking.

One of the major requirements in the MovieLabs ECP guidelines pertains to the use of “session-based” forensic watermarking to identify individual user information, and identify compromised devices that can be revoked and renewed, as required. The idea behind this requirement is that forensic watermarking technology is the only proven method robust enough to survive digital-toanalog transformations in order to trace the origin of an illegally redistributed stream or file back to the individual device from which it was captured.

However, the ECP guidelines do not go into any details about the specific requirements for the proper implementation of a forensic watermarking system. Industry players are working on defining and testing these requirements.

In addition to the requirements for the particular watermarking technology to be robust and imperceptible and to support UHD features for higher resolution, pixel depth, dynamic range and wider color gamut, building out the operational ecosystem and related policies for forensic watermarking that can lead to measurable piracy mitigation outcomes is a major challenge for all stakeholders. Several factors contribute to the challenge.

Complex ecosystem

In any watermarking implementation, several different companies in the value chain must interact and cooperate. There are entities that directly benefit from embedding, tracking and tracing the forensic information and that mandate its use in the ecosystem; others that have to implement the embedding and detection technology, and yet others that provide required anti-piracy remediation services.

Below is a synopsis of the required roles for various types of watermarking applications. Depending on the application, in some cases a single entity can play more than one role. Here are the roles and a simple definition attached to each role.

Mandator: This is the entity, usually the content owner, which issues the requirement to use watermarking and provides the “approvals” for required levels of robustness and perceptibility for a given technology.

Service provider: This is usually the distribution company that is contractually required to implement the watermarking system through integration with its suppliers of core watermark technology, middleware and infrastructure. In the case of pay-TV, this would be the MVPD.

Implementer: This is usually the entity that actually provides the watermark embedding technology or service. Often it is a middleware provider such as a conditional access system or digital rights management provider.

Dependent middleware: This is usually the entity that needs to interact with the watermarking technology during the embedding process, for example, an encoding or transcoding vendor.

Device manufacturer: An end-point device that must either react to an embedded watermark or at least “do no harm” to the watermark; for example, a set-top manufacturer or a SoC supplier.

Content delivery network: Usually a company that provides the delivery infrastructure. Monitoring service provider: This is the entity that supplies the crawling/monitoring function, and in some cases, the actual detector functionality to find the forensically watermarked files outside the authorized distribution domain.

Core technology supplier: This is the entity that supplies the core technology/ algorithms and basic functionality to implement the watermark embedder and detector.

The best way to think of these ecosystem interactions is to recognize that the “pressure” to implement watermarking originates with the mandator and subsequent business and technical responsibilities flow down to the other entities. The relationships and the obligations among the various parties are generally governed by the strictures of complex licensing agreements – and in some cases, augmented by industry groups that can guide some harmonization of best practices – which enumerate the requirements and responsibilities of the other players in the ecosystem.

Inconsistent policies

There is no “one-size-fits-all” watermarking solution that meets the needs of distributors and content providers for both VOD and live content (in broadcast as well as OTT distribution), because of where and how the embedding workflows need to be performed, such as server side/head-end, client-side/STB or player device, in the CDN etc.

In order for these systems to work, the following sets of collaborative principles need to be agreed and orchestrated within the ecosystem:

• With content available from a multitude of legal distribution sources, effective use of watermarking requires support and collaboration from many parts of the ecosystem, from chipsets to STB to MVPD to networks and core technology suppliers as well as monitoring providers. Therefore, content owner requirements for robustness and imperceptibility need to be conform to real world requirements.

• Due to the global nature of content piracy, piracy monitoring systems need to become more efficient and collaborative in order to create effective and sustainable enforcement strategies, instead of just relying on current DMCA whack-a-mole tactics.

• There are no standards in place for true session-based forensic tracking systems (aside from those used in digital cinema) and therefore enforcement efforts are impeded by the fact that pirated content can originate from multiple sources distributed by different operators, which may use multiple watermarking providers and keys. As a result, monitoring providers do not know which watermark may be in the pirated copy and therefore which detector to use, thereby making tracking across multiple operator environments hard to scale. There are some efforts underway to solve the business and technical challenges to with business processes and best practices, which can streamline the use of detection technologies at scale.

• Many jurisdictions have completely different privacy regimes which may hamper the development of harmonized policies to deter or delay the deployment of forensic watermarking on a global scale. Content owners need to deploy resources to educate and convince policy makers in different jurisdictions of the harm caused by inconsistency in piracy enforcement methods.

Misaligned incentives

In addition, watermarking imposes a relatively high cost burden on content distributors and multiplatform operators. The incentives to implement watermarking and pay for watermarking are misaligned between the parties that receive the direct benefits of watermarking (rights owners vs. distributors) and there is no consensus on whether and how the cost burdens could be shared by the various players in the ecosystem.

This issue is further exacerbated by the fact that, these days, piracy itself has become more of an OTT phenomenon, and a global one, at that. A pirate site providing access to current content is available instantly all across the world. Therefore, a single leak of content could be available everywhere. As a result, it is difficult for an operator to justify capital investment in implementing watermarking within its own footprint, unless they are assured that rights owners are requiring all operators to implement similarly stringent policies. It is important for content owners to be consistent in their contractual arrangements in different regions.

None of these concerns are addressed in the MovieLabs’ document. But if forensic watermarking is to scale and succeed in a true B2C context, with sustainable and supportable implementations that provide measurable ROI for distributors and make a dent in the losses due to piracy, there will need to be some level of collaboration, coordination, consistency and sharing of investment risk by the various entities in the ecosystem. If the industry can achieve this, we will definitely see a reduction in unauthorized, un-monetized redistribution of premium content.