Homepage

Research Area

Non-asymptotic (Finite Block Length) Analysis of Communication Systems

Non-asymptotic or finite block length analysis catches people's attention in information theory. Asymptotic anlaysis is to study the limitation of performance of channel and source coding when the block length approaches infinite, while this area focuses on the scenario when block length is not so large. Several interesting results are obtained, which are infeasible in asymptotic analysis. For example, with error probability more than 0.5, channel coding rate can exceed channel capacity. And the best random channel codes may not be generated by capacity-achieving input distribution.

Interacticve Information Theory

The fact that interaction is the essence of communication is often understated, depreciated, or even ignored in the research of Information Theory (the theory for desing of reliable communication system). In practice, however, two-way communications are commonly seen, in which transmission in one direction is the main concern, while limited transmission in the other direction is allowed. A typical example is TCP-IP Protocol in data networks, where information receivers are allowed to respond to transmitters without generating too much traffic in networks. A natural question raises here: what and how much benefit can be gained if we take advantage of this limited, and therefore often ignored, resource?

Image and Video Compression

With people’s ever-growing demand for exchanging information, bandwidth becomes more and more valuable resource. The situation is getting even worse when it comes to wireless communication. On one hand, due to the nature of wireless environment, the bandwidth is very limited; on the other hand, the portability of communication tools emerges as one of prior concerns. The ongoing smart-phone wave shows the inevitable trend to implement the multimedia (mainly video, audio and image) applications on small, portable and battery-limited wireless devices. And one of ultimate questions is: how is possible to transmit the huge amount of multimedia data (especially video data) with critical bandwidth and power budget?