In a meeting room on the ninth floor of the Connexis South Tower in the Fusionopolis building, Singapore, Jo Yew Tham browses a tailor-made online movie catalogue and picks out a Hollywood action film. He plays the high-definition video on a large wall-mounted projector screen, and then effortlessly streams the same video onto a laptop computer and a mobile phone without even a stutter in the motion picture. “You may not realize it, but you are now witnessing a very new multimedia experience,” he says.
Tham is group leader and principal investigator of several video processing projects undertaken by the Signal Processing Department and Interactive Social Tele-Experience Programme of the A*STAR Institute for Infocomm Research. The technologies behind his new multimedia experience are what Tham calls the A*STAR Scalable Multimedia Platform (A*SMP)—a comprehensive middleware platform for online video content providers that offers scalable video encoding, data storage, robust streaming and intelligent client playback on different devices in a simple and efficient manner.
Unlike conventional video streaming, which requires different versions of files to be stored and streamed for different client devices, the A*SMP uses a single multilayer file. The server delivers different layers of video depending on the capabilities of the client system and current network conditions.
“Right from the beginning, we have adopted a systemic research and development approach to transform A*STAR’s various research inventions into concerted licensable technologies that can impact our daily lives in the networked multimedia world, rather than doing silo-based research that often results in standalone components,” says Tham, who has already used his experience and entrepreneurial drive to manage a handful of commercial start-ups in Silicon Valley and Singapore.
One file, one system for all devices
Tham’s project teams have developed an unprecedented system for A*SMP that allows a video source to be encoded just once for playback on a range of different client devices, from smartphones and car infotainment systems to PCs and high-definition televisions.
The system is based on the latest international standards for video coding, file format and content streaming. The scalable streaming server and intelligent client automatically compute and assemble the best combination of scalable video layers for delivery and playback in real time.
Tham’s approach differs radically from current technologies, which demand the source video to be encoded multiple times to generate the multiple versions of compressed video streams required by different audio–visual devices. The conventional approach results in a more complicated content generation, storage and distribution process, and requires considerable duplication of data storage at the server—typically more than ten files for each video source.
“With A*SMP, different users can independently request the appropriate video layers from a single scalable video file on the server,” Tham explains.
A variety of sub-systems are under development by Tham’s team to assist in the smooth transmission of data streams. “For example,” says Kwong Huang Goh, a principal investigator on the projects, “we are inventing advanced ‘cross-layer video scrambling’ methodologies to generate packet-based streams that are robust with respect to data loss during network transmission. Another important sub-system involves a technique that quickly masks the effects of lost data to provide the best possible video playback quality at all times.”
Jo Yew Tham and members of his team demonstrate the A*SMP system at the 2009 Communicasia technology exhibition.
Coverage of live events
Tham’s team is currently working with a Singapore-based company to implement A*SMP as a platform for launching an unprecedented video-on-demand service in Singapore in 2011, by which time 60% of the city-state’s offices and households are scheduled to be networked with optical fiber. The company is also looking to bring the service to other Southeast Asian countries in the near future. “We are actively working on an interactive multi-camera sub-system that will allow the coverage of live events such as sports games. Scalable video is still very new, so we are in a good position to become the world’s leading research group on next-generation online multimedia services,” says Tham.
About the Institute for Infocomm Research
The A*STAR Institute for Infocomm Research (I²R) was established in 2002 with the mission of becoming a globally preferred source of innovations in interactive secured information, content and services. I2R conducts research and development in information technology, wireless and optical communications, interactive and digital media, signal processing and computing.
