Summary of the Second W3C Web and TV Workshop

HTTP adaptive streaming

Most players of the Web and TV world who need to stream video on the Web require some adaptive mechanism to quickly react to network fluctuations and ensure a smooth user experience while watching videos.

The panel on HTTP adaptive streaming showed convergence of interests towards the MPEG DASH specification as a key enabler for video streaming on broadband connections. There was a long debate on the patent policy under which this specification could eventually be released. W3C standards follow a royalty-free patent policy and the potential adoption of DASH for use as a core Web technology would require this solution to be licensed on similar terms.

There was strong support among workshop participants for the IG to get back to 3GPP and MPEG DASH members and assess whether MPEG DASH may be licensed on a royalty-free basis for integration within HTML. As 3GPP members held a meeting the week after the workshop, the Web and TV IG chairs prepared and sent a heads-up message to 3GPP members to express the interest of workshop participants in adopting DASH as a core Web technology and working on its integration within HTML, provided licensing can be made compatible with W3C patent policy.

Integration within HTML may require exposing new functionalities in the browser, such as QoS (Quality of Service) counters, control of the adaptive streaming process, and access to the streaming manifest. The Web and TV IG should prioritize and discuss these needs on a second step.

Content protection

Most video/audio content providers intend to implement some form of content protection to make copying difficult, if not impossible.

The discussion showed that content protection always involves some proprietary solution that cannot be easily integrated in an open stack of technologies and that should not be standardized at this stage.

That being said, parts of the mechanisms used to protect content may still be up for standardization. In particular, a common encryption algorithm would allow to separate the choice of the delivery platform (technical decision) from the choice of a DRM system (business decision).

The IG should evaluate this possibility, probably in relation with discussions on HTTP adaptive streaming. While a complete integration of DRMs in HTML is not deemed necessary, additional functionalities may need to be exposed within the browser to enable support for DRM content, e.g. extensions to the canPlayType() method to report on protected content support or secure device identification to increase protection efficiency.

Second-screen scenarios

Second or Multi-screen scenarios come with a variety of use cases, from using a second device as TV remote controller to shopping, gaming and other interactive scenarios where the TV set drives content on a second screen or gets driven by some second screen.

To work on the Web, these scenarios require some form of real-time synchronization between devices at the browser level. This synchronization is in scope for the Web Real-Time Communications working group, whose draft charter is under review by the W3C Membership as this report is written. Provided this working group gets chartered, many second-screen scenarios would be enabled as a direct consequence. Interested W3C Members are encouraged to review and show support for the draft charter.

For multi-screen scenarios to work smoothly, device discovery must be possible from within a Web browser to actually pair devices together. DLNA comes to mind for home networking scenarios. The Device APIs and Policy (DAP) Working Group is currently looking into re-chartering, and the possibility to work on some form of device discovery mechanism is being discussed on the public public-device-apis@w3.org mailing-list (see the public archives of the mailing-list) and may also prove to be of interest for workshop participants. The Web and TV IG should serve to centralize interest for this topic coming from the Web and TV community, follow these discussions, and react accordingly, e.g. by discussing the creation a working group dedicated to device discovery if this topic ends up not being addressed by the groups mentioned above.

On top of synchronization and device discovery, Web browsers may need to expose additional functionalities such as an API or events dedicated to remote controllers. The Web and TV IG will need to refine requirements, see how these needs fit with ongoing efforts in e.g. the Web applications working group, and agree on next steps.

Metadata / Semantic Web

The session on metadata was the occasion to review different aspects of the technology and its use in Web and TV environments.

The requirements for metadata exposed in the different presentations touched different complementary areas like what is needed to:

• discover services
• reference/identify and describe content (including profiles per application or media type)
• expose this information (manifests on web pages, data for widget consumption, reuse by middleware solutions, etc.) and convince content providers it is for their benefit (provide open access to metadata is vital for visibility)
• consume content across a variety of platforms and media (access and transactions)
• develop social interactivity and simplify user access

Several solutions/implementations have been mentioned, including Connected Media Experience, DVB and TV-Anytime, and Hikari-TV implementing the ITU-T specifications for IPTV.

Based on the dicussions held during the workshop, the Web and TV Interest Group should probably take on the following actions:

• Review the technologies currently used for hybrid web-TV solutions and do a gap analysis with the activities in W3C
• Coordinate metadata related developments including HTML5 activities on the use of the video tag and its attributes, RDFa in HTML, the Ontology for Media Resources developed by the Media Annotations Working Group, Media Fragments URI developed by the Media Fragments Working Group.
• Propose a vision on metadata and a roadmap for convergence between W3C developments and industrial developments or television and radio standardisation activities
• Collect and publicise best practices

Extensions to HTML5

A number of organizations external to W3C have worked on extending existing Web standards to meet requirements for connected TVs in hybrid broadcast/broadband scenarios. These organizations include HbbTV, DVB, and DTV/IPTV in Japan.

Some of these extensions may need to be taken up as core Web technologies, for instance support for trick modes (common video playback functions, e.g. pause, play, fast forward, fast rewind, slow forward, slow rewind, jump to previous or future frame), recording, and downloading content, or control of the active video channel in a broadcast environment.

Extensions are of various kinds, from the introduction of new markup attributes or new DOM events to new API functionalities or specific ontologies, or to mechanisms to synchronize video with other content.

The Web and TV IG should discuss with other organizations, and determine priorities as well as starting points for potential standardization efforts within W3C.

The HTML working group has an open issue (ISSUE-152) on support for multiple audio/video tracks within HTML5 that impacts many actors of the TV world. Since this issue is to be resolved by 22 April 2011, interested workshop participants are encouraged to raise their concerns directly on the HTML mailing-list if their point of view has not already been taken into account.

Accessibility

Most participants pointed out that, in many ways, accessibility in the TV industry has been much better than accessibility on the Web, partly thanks to regulations that companies must comply with to sell TV products.

Accessibility needs to be taken into account right at design time and not as an after thought. While sometimes considered by companies as a regulatory requirement, participants observed that adaptive accessibility can often be leveraged in non-accessible contexts, be it only to watch a movie with subtitles in a sound-sensitive environment.

In particular, accessibility was shown as important for the design of multimodal applications that work on various kinds of devices. Works on Accessible Rich Internet Applications (WAI-ARIA) and multimodal interactions were mentioned as candidate technologies to target heterogeneous environments.

The need to harmonize the Web Content Accessibility Guidelines 2.0, that serve as recommendations for making Web content more accessible, and various existing regulations from the TV world was also pointed out.

The Web and TV IG should seek to include experts in accessibility in discussions as soon as possible.

Profiling / Testing

In the most simple scenario, users pay for a consumer electronic device once and stick with it during several years. Firmware updates incur high costs on the manufacturer and/or operator's side, and are usually restricted to critical ones. This means that specifications implemented in consumer electronic devices need to be extremely stable. This stability contrasts with what often happens on the Web where software updates get distributed much more easily on an automated basis.

HTML5 introduces a number of features that are useful from the point of view of the TV world (e.g. support for video/audio) but HTML5 is not a W3C standard and cannot be considered as stable yet. The need to develop comprehensive test suites to ensure interoperability was stressed out. A few days after the workshop, the W3C confirmed the roadmap for HTML5, setting advancement to Last Call for the HTML5 specification for May 2011, and to final Recommendation by 2014, insisting on the testing effort as key for the success of the specification.

Similarly, consumer electronic devices are much more limited in terms of computing power. Material that is only a few years old is often obsolete from a performance's perspecive. Companies that want to agree on a common platform need to subset and/or profile existing specifications to meet their requirements both in terms of hardware constraints and time-to-market.

Discussion on whether profiles need to be specified at W3C is still open. W3C standards were compared to « toolkits » where standards are designed to be orthogonal and extensible on purpose. Profiles that do not subset specification but impose a particular set of specifications (usually coupled with specific extensions) may best be left to external organizations. On the contrary, subsets of specifications need to be discussed with the working groups that are responsible and own the copyright of the underlying specifications (e.g. the HTML Working Group or the CSS Working Group).