An information theoretic view of network management

Abstract

We present an information theoretic framework for network management for recovery from non-ergodic link failures. Building on recent work in the field of network coding, we describe the input-output relations of network nodes in terms of network codes. This very general concept of network behavior as a code provides a fundamental way to quantify essential management information as that needed to switch among different codes (behaviors) for different failure scenarios. We give bounds on the network management information needed for link failure recovery in various network connection problems, in terms of basic parameters such as the number of source processes and the number of links in a minimum source-receiver cut. This is the first paper to our knowledge that looks at network management for general connections.