Reliability assessment of a stochastic node-failure network with multiple sinks under tolerable error rate

Internet service providers and enterprise customers are increasingly focusing on the tolerable error rate during transmission. This has high significance especially in service-level agreements. This study concentrates on a stochastic node-failure network (SNFN) with multiple sinks, where each component (arc or node) has several possible capacities and a transmission error rate. Measuring the performance level for an SNFN is an important issue in the system-design and system diagnosis phases. This study evaluates the system reliability of a network, or the probability that the demand is fulfilled under the tolerable error rate. In terms of minimal paths, we propose an efficient algorithm to find all minimal capacity vectors, and then compute the system reliability in terms of all minimal capacity vectors using the recursive sum of disjoint products algorithm. Finally, we test the proposed algorithm for a benchmark network and two practical networks.