A GENETIC ALGORITHM FOR MULTICOMMODITY STOCHASTIC-FLOW NETWORKS SYSTEM WITH UNRELIABLE NODES

Abstract

Many real-life systems can be modelled as stochastic-flow networks, where resource flows traverse through unreliable nodes and arcs. This paper addresses the problem of optimizing resource flow allocation and control strategies in multicommodity stochastic-flow networks under budget constraints. To evaluate the reliability of such networks, we develop a genetic algorithm (GA) that identifies every vector with limited capacity meeting the specified demand and budget constraints. The system reliability is then resolute based on these vectors, ensuring the maximization of the probability that sink nodes' demands are satisfied as resource flows transmit from source nodes. The proposed GA is designed to efficiently seek the optimal allocation strategy, balancing computational complexity with effectiveness. To demonstrate the efficacy of the algorithm, a numerical example is presented. Results show that the GA not only achieves high reliability but also exhibits better temporal efficiency, making it applicable to larger and more complex networks.