PITA Fiscal Year 2008 Projects - Information and Systems Technology

Distributed Information Processing For Wireless Ad-Hoc and Sensory Networks - Phase Three: Joint Source/Channel/Network Coding

This project, carrying a previous PITA project to further depth and width, focuses on developing and optimizing cutting-edge coding technologies to improve the efficiency and reliability of multi-terminal communication, a pivotal element in a modern economy and in national security. The exiting layered approaches, although extremely successful in the Internet, are becoming much less efficient and even fragile when applied on wireless ad-hoc and sensory networks which lack static topologies and centralized control and which are much more resource-limited in bandwidth and power provision. Targeting wireless and ad-hoc setup, the proposed coding strategies differ from the conventional ones in that they are distributed, adaptive, and designed through cross-layer and integrative optimization, rather than being centralized, static, and separately designed.

A complete cross-layer approach is extremely difficult even to formulate. We set realistic goals and target integrative coding that combines source coding (i.e. data compression) at the application layer, channel coding (i.e. error control) at the physical layer, network coding (i.e. packet routing) at the network layer, and/or multiple-access control at the MAC layer. In Phase I and Phase II of this research, we developed a few low-complexity but high-performance distributed source coding techniques and network coding strategies. These interesting results not only led to important technical publications, but also caught the attention of several companies, who are now seriously considering implementing them in real systems. The proposed work builds upon the successful results developed previously, and targeting integrating and jointly optimizing them to meet the specifications and requirements of different systems, especially short networks with very few collaborating users, highly mobile networks, networks with limited bandwidth provision, and networks with delay constraint. The tools used in this research are highly mathematical, and include linear algebra, graph theory, convex optimization, combinatorics, and finite geometry.