Distributed coordination in network information theory

Abstract

Constructing a large-scale wireless data network is spectacularly expensive. It is therefore important to understand how to efficiently utilize the physical infrastructure and available battery power, which are expensive system resources. Unfortunately, we currently understand very little about efficient communication in a distributed environment. In distributed wireless networks, there appears to be an interesting and complex trade-off between trying to take advantage of independent noisy signals at different relays and closely coordinating relay transmissions to a receiver. Designing the right structure for efficient communication, by choice of source transmission codebook and relay terminal processing, is the important and difficult problem on which we focus. We use an information theoretic framework to study several very simple multiple terminal networks, focusing exclusively on single source, single destination networks where communication must take place through intermediate nodes. Our goal is to determine how much data we can get reliably from source to destination, placing no importance on delay or computational complexity. The core problem then involves distributed detection at the intermediate nodes and coordination in relaying information to the destination.