Limits of Information Flow in Networks of Two-Way Channels

点击翻译按钮获取中文摘要

Communication networks are critically important in thedissemination of information in today's technologicalsociety. Several academic disciplines have undertakenthe study of various aspects of their operation, butan integrated theory of networking has been elusive.The objective of network optimization theory is tomost effectively use network resources, but recentwork in network coding has demonstrated that a standardassumption of flow conservation from this theory hasimposed artificial restrictions on the workings ofcommunication networks. Network information theoryconsiders more general models of communication andoffers some bound on the limits of information transferin a communication system involving multiple sendersand receivers.This research fuses aspects of network optimizationtheory and network information theory in order to findand specify new limits of information transfer acrossa communication network. The investigators use thetwo-way channel from information theory as a generalmodel of the transmission medium between neighboringprocessors of a communication network and addressfundamental optimization problems pertaining to theanalysis and design of networks of two-way channels.The investigators seek optimal routing algorithms inmulticast settings and a broad classification ofsituations where network coding can or cannot improveupon routing. The investigators further aim tocategorize network scenarios in which a bidirectedcut set bound derived using information theoreticarguments is tight and to explore a network informationcounterpart to Kirchoff's law when the bidirectedcut set bound is not tight. The research alsoinvolves extending the notion of minimal cost flowas an initial approach to treating quality-of-serviceissues and devising networks with least cost orcapacity that meet requirements on information transfer.