Research On Ant Colony Algorithm For Solving QoS Routing Optimization

At Present , Networks have many applications with QoS requirements. QoS routing is a research hotspot in today's network technology field. Research shows that as a NP-complete problem, Multiple Constrained QoS Routing Problem can not be solved easily with conventional methods.With the features of Stability, positive feedback, distributed computing, being constructive and easy to combine with other algorithms, Ant colony algorithm(ACA) is a new artificial intelligence algorithms. In addition, Ant colony algorithm with distributed computing, self-organizing characteristics matchs well the characteristics of network routing optimization problems such as calculation of the distribution, non-static random dynamic and the updating of Asynchronous network status. Therefore, it is a good choice to to solve the network QoS routing optimization problem with ant colony algorithm.This paper first expounds systematically and completely the basic concepts and principles of QoS Routing. Then it describes in detail the basic principles of ant colony algorithm, the algorithm model, algorithm flow, etc. The paper also makes a thorough analysis of the algorithm parameters and compares the advantages and disadvantages of the algorithm.As to the drawbacks of ACA, such as its slow convergece in the beginning,long convergence time later and its being prone to fall into local optimization, three improved idea and simulated experiments has been proposed in this paper. Frist of all ,the search mechanism for two-way division of labor was introduced. Path mutation strategy and applying ACA twice was implemented. Secondly,it improve the node selection strategy and update pheromone dynamically base on the objective function value. Thirdly,pheromone has always been controlled within a certain range base on the principles of the Max-Min ant colony algorithm. Local search method was also applied to improve the efficiency of the search. Simulations are made on some examples for the improved ACA in the QoS routing. Results of those simulations demonstrate that the improved ACA is effective and feasible.