Dynamic Routing and Wavelength Assignment（RWA）is important in optical networks

which have a direct bearing on their QoS.This paper presents an improved Ant Colony Optimization（ACO）algorithm for this issue

which enhances the optimum pheromones of the ant colonies and converges them in accordance with the optimal solution in these colonies and increases the guiding role of the global information for them so as to achieve still better results.Simulation results indicate that compared with the conventional ACO algorithms

this improved one significantly reduces the blocking rate and improves the network QoS.