A genetic algorithm for joint optimization of capacity and flow assignment in packet switched networks

One of the design problems concerning packet switched computer communication networks is the joint optimization of capacity and flow assignment (CFA). In this problem the topology of the network and the traffic requirements between node pairs are given, and the objective is to find the flows on the different links and their capacities that minimize the total network cost and keep its average delay below a certain upper limit. This paper proposes an algorithm based on hybridization of genetic algorithms (GA) and the shortest path routing. The proposed algorithm has shown simplification in implementation and improvement in performance. Computational results indicated a 3.17% reduction in network cost when compared with a previous method.