It also signifies the rising importance of forensic watermarking at the network edge in the burgeoning era of video distribution over 5G networks.

ContentArmor belongs to a select group of developers and patent holders in the forensic watermarking field, as one of just seven members of the Digital Watermarking Alliance. The others are NexGuard, Verimatrix, Irdeto, Media Science International, MarkAny and Synamedia itself.

Before its takeover of ContentArmor however, Synamedia lacked the technology to apply watermarks effectively at the network edge. Its Streaming Piracy Disruption (SPD) has enjoyed some success as a managed service suite for detecting and disrupting streaming piracy across both broadcast and OTT services. But this operated mostly at the headend, injecting watermarks there. This technology had been integrated with edge platforms from CDN (Content Delivery Networks) such as Akamai’s for detecting and disrupting pirate streams in real time over those platforms and redirecting viewers to legitimate services. But Synamedia wanted to gain control of the technology to offer edge-based watermarking directly to video service providers.

“While head-end watermarking is now at the heart of any studio or service provider’s security requirements, we are primed for the future with our edge watermarking solution,” said Alain Durand, ContentArmor chief executive officer. “By combining our high-performance technology with Synamedia’s unparalleled intelligence-led security offerings, service providers can future-proof their revenues from the menace of piracy,” added Durand, who joins Synamedia’s security team.

Forensic watermarking has climbed rapidly up the video security agenda with the rise of streaming. The original piracy threats to pay TV came from cloning of smart cards to bypass conditional access controls around set top boxes, evolving into control word sharing to achieve the same goal without any physical device as internet access became more widespread.

Then as broadband capacity and quality increased, illicit stream redistribution became a major threat and this required some mechanism for tracing the source of those specific playouts or instances. Watermarking emerged as the only proven technology capable of achieving that by inserting some indelible and yet imperceptible marks into individual streams.

This can be done at the headend, network edge or even the client, with two broad alternative approaches. One is the AB segment approach, where content is encoded as two parallel streams of alternating segments. There are at least two versions of each video segment, generated by encoding varying watermarked pictures. These yield variants of the segment that typically differ in size as a result of contextual differential encoding of the video. Each user then receives a unique permutation of those segments, which can be orchestrated from the network edge. That provides the traceability, since each playout would incorporate a unique AB combination that would enable a device retransmitting it illicitly to be traced. That device might belong to a legitimate subscriber, so that it is only the event of unauthorized retransmission that becomes incriminating.

The alternative approach is bitstream watermarking, which has been adopted by ContentArmor among others. This entails insertion of the marks directly into each bitstream, which can be done at the network edge. This avoids having to transmit at least two separate versions of each stream across the CDN as with the AB segment method, with potentially huge saving in bandwidth as well as cached storage.

With the first implementations this had to be combined with encryption at the edge after the watermark had been inserted. There was a tradeoff between the computational cost adding bitstream watermarking in this process and the saving of 50% or more in storage compared to AB watermarking.

However, ContentArmor inserts the watermark payload directly in the compressed video with no need for decompression and subsequent repackaging. This reduces complexity and that computational overhead, while being easier to integrate into CDNs as a software-only approach.

Confidence in this approach has led Durand to set out six factors service providers should consider when selecting a watermarking package. Firstly comes fidelity, meaning that the embedded watermark must be invisible to the viewer. Secondly is robustness, requiring the embedded watermark to withstand various operations performed by pirates in their attempt to erase a mark. These include camcording and also screencasting.

Thirdly is security in the sense of surviving targeted technical attacks beyond those basic operations. Fourthly is throughput, that is the ability to encode and detect information as quickly as possible, which is particularly important when combating live sports streaming piracy where business damage can be incurred in minutes. Fifth is computational efficiency, without which an otherwise effective technology would be useless in practice. Finally, is ease of integration to minimize cost and time involved modifying systems already deployed.