Evaluation of stochastic flow networks susceptible to demand requirements between multiple sources and multiple destinations

Abstract

Computer and communication applications such as big data analytics services, social networks, enterprise business and mobile applications often require to simultaneously pass data of different sizes between multiple source and multiple destinations in the network. In practice, the nodes in the communication networks are often connected through multiple intermediate links of different types with varying failure probabilities. Failure of intermediate links can adversely affect data transmission between source and destination nodes; thus, they can impact the quality of service. Therefore, reliability evaluation of such networks is of subtle importance in today’s service dependent world. This article presents an efficient method for reliability evaluation of stochastic flow networks that can pass various demands simultaneously from multiple source nodes to multiple destination nodes. The proposed method has three steps: First step obtains the combined minimal cut sets between the given set of source and destination nodes. Second step generates the set of simultaneous upper boundary flows for the varying demands using the combined minimal cut sets. Third step calculates the network unreliability by applying the Sum of Disjoint Product method on the upper boundary flows. The reliability of the network, i.e., the probability that the network can simultaneously pass the set of demands, is calculated as 1-Unreliability. The MATLAB simulation of the proposed method on bench mark networks show that the proposed approach takes less computational time than the existing method.