An efficient approach to find reliable topology of stochastic flow networks under cost constraint

Abstract

Computer and communication networks are often stochastic capacitated networks due to partial or complete failure of network components such as links and nodes. Therefore, in practice, network components have different associated costs based on their bandwidth and failure probabilities. Different topologies which are composed of different components of the network may provide various reliability values for a given cost constraint. Finding a topology of a stochastic flow network that can pass a set of demand from multiple sources to multiple destinations with maximum reliability under given cost constraint is an important problem. This work presents an efficient approach to find such topologies in two phases: First, it finds the upper boundary topologies which are subset of topologies that satisfy the cost constraint and are more reliable amongst the others. Second, it evaluates the reliability of upper boundary topologies and chooses the one that provides maximum reliability. Reliability evaluation is computationally hard problem. Proposed method minimizes number of reliability evaluations of topologies through upper boundary topologies. Further, it presents two efficient methods for obtaining the upper boundary flow vectors which are required for reliability evaluation of the upper boundary topologies. We implemented the proposed methods in MATLAB and analyzed their performance using the benchmark networks available in the literature.