An Overview of Network Information Flow Problem

The effect of network coding began widely realized when a network with its optimized throughput can be achieved only if linear coding was introduced to its intermediate nodes was presented by R.Ahlswede et al. in [1]. A series of advantages of network coding over routing have been discovered during afterwards research by R.Koetter, M.Médard et al. This talk tries to provide an overview of network information flow problem. 1 Basics of network coding Network coding has been proved to be an effective technology in solving network information flow problem, which is derived from traditional multicommodity flow problems and have recently absorbed some ideas from information theory and coding theory. Our information network are directed acyclic graphs G = (V, E). The vertex V contains three disjoint subsets: source nodes S, target nodes T and intermediate nodes I. For each source node s, the messages it transmits through intermediate nodes to target nodes through edge set E are drawn from a fixed finite alphabet A with size |A| > 1. For each target node t, the messages it requires is a subset of messages from source nodes. The intermediate nodes can not only duplicate and forward messages they receive from in-edges, but also use mathematical functions to compute these messages before forwarding them(the power of node’s computability is unlimited).