Improving NSGA-II for Multi-Constrained QoS Routing

Abstract

With the development of 5G technology, the traffics which need multiple quality of service (QoS) constraints greatly increase. The existing QoS routing algorithms either support a single QoS constraint or support multiple constraints are mixed with fixed weights. The latter is essentially a single-constrained QoS algorithm. Few studies can provide the quality of service guaranteed under the true sense of multiple constraints, by which only one solution can be provided. In this paper, we propose NSGAII-MCR: a multi-constrained QoS routing algorithm based on non-dominated sorting genetic algorithm II (NSGA-II). In NSGAII-MCR, three representative constraints are selected: delay, bandwidth, and packet loss rate. The validity of the routing is guaranteed by a designed encoding and decoding scheme and a novel individual generation method. Meantime, the original NSGA-II crossover and mutation operators are modified. We further enhance the selection strategy to improve the diversity of solutions and the robustness of NSGAII-MCR. At last, multiple optional routes in the Pareto-optimal front are obtained. The experiment demonstrates that the proposed NSGAII-MCR can achieve high success rate by up to 95%. Compared with the benchmarking methods, NSGAII-MCR can support more constraints and finds more multi-constrained routes.