| Speaker: | Dr. Anant Sahai, University of California-Berkeley |
|---|---|
| Abstract: | Channel and source coding are traditionally considered in a block-code setting with encoders and decoders designed for a given channel/source and a fixed block-length. Block-length plays a double role --- it is a proxy for implementation complexity as well as a stand-in for end-to-end delay. My goal is to isolate the role of delay itself, since it is what fundamentally allows the exploitation of the laws of large numbers to reduce errors.
In channel coding, this prompts a study of the error exponent with delay. While closely related to block-coding error exponents, it is fundamentally distinct. While feedback often does not improve fixed block-length reliability functions, it can significantly improve the reliability with respect to fixed delay! (Contrary to a "theorem" by Pinsker claiming otherwise.) A new bound, called the "focusing bound," allows us to calculate the limit of what is possible with feedback and sheds some light on how feedback can change the nature of the dominant error events. This bound is also asymptotically achievable for many channels with noiseless feedback and the achievable scheme points out the value of "flow control" in channel coding. Furthermore, this scheme works universally over sufficiently large delays and so the channel encoder does not need to know the application's target delay. This delay-universal (or anytime) sense of reliability turns out to be fundamental when considering interactive applications like control systems over noisy feedback channels. In source coding, we can use a focus on delay to "solve" the old problem of finding a variable-length but zero-error generalization to the Slepian-Wolf problem of noninteracting encoders. In general, it is impossible to achieve interesting points in the SW region if we require the decoder to be certain that it has decoded correctly. By only requiring that the decoder eventually decodes everything correctly, we can get all the interesting points in the SW region. Aspects of this talk reflect joint work with S. Mitter, T. Simsek, S. Draper, C. Chang, and others. |
| Biography: | Anant Sahai received the B.S. degree in EECS from the University of California at Berkeley in 1994, and the M.S. and Ph.D. degrees in EECS, both from MIT, in 1996 and 2001, respectively. In 2001, he developed adaptive signal processing algorithms for low SNR software radio at the startup Enuvis in South San Francisco. He joined the EECS department at Berkeley as an Assistant Professor in 2002. His current research interests are in control over noisy channels, the role of delay and feedback in information theory, and questions of spectrum sharing in wireless communication. |
| Presented On: | Monday, November 7, 2005 |
| Videotape: | Sahai.mov |
