Optimal QoS Routing Based on Extended Simulated Annealing

Quality-of-service routing (QoSR) is to find an optimal path that satisfies multiple constraints simultaneously. As an NPC problem, it is a challenge for the next-generation networks. In this paper, we propose a novel heuristic SA MCOP to the general multi-constrained optimal path problem by extending simulated annealing into Dijkstra’s algorithm. The heuristic first translates multiple QoS weights into a single metric and then seeks to find a feasible path by simulated annealing. Once a feasible path is found, it optimizes the cost without losing the feasibility. Extensive simulations demonstrate that SA MCOP has the following advantages: (1) High performance in both success ratio and cost optimization. (2) High scalability regarding both network size and the number (k) of QoS constraints. (3) Insensitivity to the distribution of QoS constraints.