Optimal information dispersal for reliable communication in computer networks

Abstract

In an (m, n) Information Dispersal Scheme (IDS), the sender node decomposes a message M of length L into n pieces S/sub i/, 1/spl les/i/spl les/n, each of length L/m, such that any m pieces collected by the receiver node over different paths suffice for reconstructing M. Because of variations of network traffic, the number n of available vertex-disjoint paths for the transmission from the sender node to the receiver node may vary in time. It is very difficult to determine the best n and m which gives the highest communication reliability, when given the maximum number of available disjoint paths and an upper bound for the information expansion rate (n/m). In this research, we discovered several interesting features of (m, n) IDSs which can help reduce the complexity for computing the highest communication reliability. From these findings, we propose a method for determining the optimal IDS.