Impact of information on network performance - An information-theoretic perspective

Abstract

Available network information is an important factor in determining network performance. In this paper, we study the basic limits on the amount of network information that should be transmitted in the network to achieve a given level of network performance. From the perspective of information theory, network information is an information source, and the lower bound on network information is the minimum code letters required to encode the source. We propose a general information-theoretic framework, which can be applied to any network, to study the effect of network information on the performance of any network protocol. We also analyze the tradeoff between network performance improvement and network information collection overhead. To illustrate our approach, we use the framework to determine the lower bound on the traffic information for a simple scheduling protocol in wireless networks. The results in this paper may be used to analyze and evaluate network protocols and guide future designs.