Abstract
This paper describes how SVG is being used for the forthcoming mobile digital television service in the USA. It also explains MobiTV's interest in, advocacy for, and activities around the usage of SVG in Mobile DTV.
Table of Contents
The U.S. standard for digital broadcast television is ATSC, which is really a standards body. The ATSC group has established a new standard, A/153, to address the mobile space named ATSC-M/H (mobile/handheld). Mobile DTV is the new "consumer friendly" way to refer to ATSC-M/H. It's meaning is more obvious, and it is easier to remember.
It has been extremely difficult to receive the existing terrestrial ATSC broadcasts while moving. The mobile stream contains massive amounts of error protection and correction. In addition, it is not reasonable to expect handheld devices such as mobile phones and personal media players to decode and scale down HD video to the small screen sizes. The mobile broadcast provides a widescreen 416x240 resolution. Furthermore, this new broadcast signal is backwards compatible with all existing ATSC tuners.
To be competitive with the current and future mobile entertainment landscape, mobile TV must expand beyond the mobile playback of video and audio. Part 5 of the A/153 1.0 standard specifies an Application Framework, which will provide a new set of tools for Web-like personalization and interaction. Interactive television will allow for the ATSC-M/H viewer to engage in real-time with the television program. Real-time mobile commerce will generate new and novel revenue opportunities. Interactive advertising will increase CPM. Real-time audience measurement data can be provided via the application framework. Surveys, usage and other data go directly to the broadcaster.
The Open Mobile Alliance (OMA), a standards body that develops open standards for the mobile phone industry, has developed a Rich Media Environment (RME) specification. The ATSC has adopted this standard for its application framework.
The OMA-RME specification adds a few things to the the 3GPP DIMS standard. Namely, it adds URI support, screen orientation events, and keyboard identifiers for the "D" pad and soft keys. Most importantly, it adds the sceneCommandGroup element to make multiple 3GPP DIMS scene commands into a single, well-formed XML document.
The 3GPP DIMS standard is based heavily upon the SVG Tiny 1.2 standard. It adds much in the area of packaging and delivering. In particular, it specifies the packetized delivery over RTP. Additionally, it defines a doScript element for immediate script execution.
The 3GPP DIMS standard is also heavily based upon the MPEG-4 LASeR standard (ISO/IEC 14496-20). LASeR stands for Lightweight Application Scene Representation. Most notably, LASeR provides the scene update and state management commands. The scene update commands are markup for declarative DOM manipulation. The state management is like cookies or HTML5 localStorage.
LASeR commands include:
NewScene: SVG document
RefreshScene: random access and error recovery
Insert
Delete
Replace: update with overwrite
Add: update with addition/concatenation
SendEvent
Save, Restore, Clean: state
The first to bring live television to the mobile platform, MobiTV continues to shape the mobile media landscape by consistently introducing innovations to the market. As such, MobiTV is always looking for new ways to improve the mobile media experience, one of which is Mobile DTV.
Our principle is that Mobile DTV will help accelerate the overall use of mobile video, which in turn benefits MobiTV. We also believe a hybrid approach integrating Mobile DTV into a combined broadcast and unicast delivery model will provide the best possible user experience, while addressing a potential business model for Mobile DTV.
The technology benefits of Mobile DTV cannot be ignored, including excellent image and sound quality.
MobiTV has the technology to combine these aspects of Mobile DTV with the best principles of unicast delivery, including interactivity, personalization, audience measurement, dynamic ad insertion, and premium on-demand content with DVR-like capabilities.
As the leader in mobile media delivery, providing more content to more subscribers than anyone in the industry, MobiTV is well positioned to address an application framework for Mobile DTV with the necessary technology platform and managed services it requires.
MobiTV, along with others, recommended to the ATSC the encapsulation of UDP/IP to facilitate the integration of existing made-for-mobile (i.e. 3GPP) encoders. Furthermore, RTP has less bandwidth overhead than the alternative MPEG-2 Transport Stream. In addition, RTP is the streaming transport required by the application framework we were recommending.
We, alongside Nokia, recommended the usage of SVG Tiny 1.2, in particular OMA-RME, for the application framework. Partial implementations of SVG Tiny 1.2 have already been proven in the mobile device space, it is comprehensive and expressive without being too large, and it based on familiar web and IT standards.
Unfortunately, the the SMIL video element adopted by SVG Tiny 1.2 presented challenges. The broadcast audio/video is continuously live independent of the scene. The application framework is optional in the broadcast and user agent; it does not make sense for the optional component to define the lifetime of the core components (audio and video). Poorly authored scenes or user agent implementations must not interfere with the audio and video user experience. It creates a different tuning model for the RME vs. non-RME channels making things more complicated and impacting channel change time. Finally, it makes development of the user agent very difficult especially on resource-constrained mobile devices thereby hindering adoption.
Consequently, we recommended a layered model for the presentation of SVG Tiny distinct from the broadcast video. In this model the video window and the SVG canvas are the same size and origin. The SVG canvas is a layer with a transparent background above the video. The user agent may optionally support alpha-blended compositing of the canvas layer with the video layer; however, it must at least be able to perform non-blended compositing after converting the alpha channel to a bitmask.

In the image above, the size of the SVG drawing is the same as the video size: 416x240. Only the graphics in the top, right corner come from SVG. The graphics in the lower, left corner are of a logo that was composited with the video prior to encoding and transmission. The positioning of the visible SVG graphics are controlled within the SVG and not any external means. The position, size, and source of the video are not controlled by the SVG.
MobiTV recommended the addition of some new interfaces available through ECMAScript to facilitate the application use cases we developed.
Interface ATSCUsage
Interface ATSCUsage contains the following functions:
getATSCUserName
getATSCUserEmail
getATSCUserAge
getATSCUserLocation returns a ATSCCoordinates reference
Obviously, the user must volunteer this information by manually entering the information somewhere in the device platform or user agent configuration. If the user agent retrieves the information from the platform, it should inform the user. Some user agents may want to present a dialog to allow or deny the information upon each request.
Interface ATSCCoordinates
Interface ATSCCoordinates contains the following attributes as floating point values:
latitude
longitude
altitude
Interface ATSCGlobal
Lastly, the ATSCGlobal interfaces offers a tune() function. A broadcaster may offer multiple programs, and this allows a broadcaster to tune to a different program. The user agent will not allow the RME/SVG, which might be provided by a third party, to tune away from the broadcaster.
We also developed and demonstrated several business models and simple use cases with working applications:
show information
emergency information
local weather and traffic
e-commerce (e.g. ticket purchase)
audience measurement
polling/voting
advertisement
interactive
synchronized
targeted
age-appropriate
location-based
As of September, 2009, MobiTV has not announced any concrete plans or agreements to provide a service or product for ATSC-M/H.
For the National Association of Broadcasters (NAB) Show in 2008, we developed the world's first Mobile DTV player that supports SVG Tiny 1.2 for the interactive stream. In addition, we developed the world's first OME-RME interactive delivery server. The server also handles specific web service requests to support the interactive applications on display:
Voting and show information: Dancing With the Stars
Local traffic and weather
Show promotions
Advertisment
Audience measurement: channel change logging
For the NAB Show in 2009, we converted our player from last year into a NPAPI web browser plugin so that various partners could embed it into their player applications. More significantly, we proposed and demonstrated MixTV: a revenue-sharing business model that combines free, off-the-air mobile DTV with subscription-based on-demand content delivered over a mobile wireless data network. Furthermore, the on-demand content may come from a 3-7 day window of previous local broadcasts recorded at the broadcast facility, uploaded to the MobiTV managed streaming service, and offered by the mobile network operator for playback over their 3G or 4G data network. Additionally, MixTV demonstrated a SVG Tiny 1.2-based enhanced program guide that aggregates these otherwise disparate offerings. Finally, another demonstration took advantage of the scripting capability of the SVG player to send viewing information to a Nielsen web service for audience measurement.