Electrical Engineering
      and Computer Sciences

Electrical Engineering and Computer Sciences

COLLEGE OF ENGINEERING

UC Berkeley

   

2009 Research Summary

Wireless Multimedia Communication over Wide Area and Local Area Networks

View Current Project Information

Avideh Zakhor and Michael N Krishnan

Hewlett-Packard

In this project, we develop a spectrum of techniques for delivery of multimedia information over wireless medium, particularly wireless local area networks (WLANs). WLANs are becoming ubiquitous in many cities such as San Francisco and Philadelphia providing full WiFi coverage thanks to providers such as Google and Earthlink. Increasingly, the data that is being transmitted over WLANs are video, audio, graphics, and multimedia, rather than just text. Popularity of websites such as YouTube and social networking sites such as My-Space are a testimonial to the popularity of multimedia in general, and video in particular among users. Yet, today's WLAN protocols and standards have been primarily designed to handle data applications such as web browsing. Even though the 802.11e standard has been designed to handle QoS, recent studies with voice indicate that it does not provide sufficient delay and jitter performance in situations where voice and data co-exist on a WLAN. The problems for video are more severe in that it occupies a significantly larger percentage of the available bandwidth on today's WLANs such as 802.11b, a, or g. In this project, we will develop a suite of techniques to facilitate video communication over WLANs. In doing so, we exploit cross-layer techniques, integrating physical, MAC, transport, and application layers in a seamless way in order to ensure the highest possible video quality with the fewest re-buffers and freezes. We plan to design link adaptation techniques that are video-specific and that can distinguish between the two most dominant sources of packet loss in WLANs, namely fading and collision. In addition, we plan to investigate partial recovery techniques in conjunction with error location/detection codes by selectively forwarding corrupted packets to the application layer so that the application layer can utilize erroneous packets in reconstruction of the video signal.