A simulation Method for Network Performability Estimation using Heuristically-computed Pathsets and Cutsets

Abstract

Consider a set of terminal nodes K that belong to a network whose nodes are connected by links that fail independently with known probabilities. We introduce a method for estimating any performability measure that depends on the hop distance between terminal nodes. It generalises previously introduced Monte Carlo methods for estimation of the K-reliability of networks with variance reduction compared to crude Monte Carlo. They are based on using sets of edges named d-pathsets and d-cutsets for reducing the variance of the estimator. These sets of edges, considered as a priori known in previous literature, heaviliy affect the attained performance; we hereby introduce and compare a family of heuristics for their selection. Numerical examples are presented, showing the significant efficiency improvements that can be obtained by chaining the edge set selection heuristics to the proposed Monte Carlo sampling plan.