摘要
提出基于多目标遗传算法的QoS组播路由优化算法,在遗传进化过程中分别使用三种方法:随机权重方法随机生成权重,使算法具有可变搜索方向,沿Pareto前沿面均匀采样,增加算法成功率;Pareto排序方法合理分配适应值,使Pa-reto解具有相同的适应值,并能调整选择压力;Pareto竞争方法通过适应值共享维持种群多样性,提高遗传算法的性能。实验仿真在不同网络规模下研究算法的遗传进化过程、成功率、收敛速度和可扩展性,并与相关算法进行比较与分析,证明本文提出的算法是可行的、有效的。
Multi-objective optimization algorithms of QoS multicast routing are proposed, and three methods are used in the process of evolution： the random weight approach creates weights randomly, makes the algorithm search in alterable directions, samplings unffomdy along Pareto-optimal Front, and increases the success rate. The Pareto ranking approach assigns the fitnesses in a reasonable way, makes the Pareto solution have same fitnesses, and can adjust the pressure of selection. The Pareto tournament approach maintains the diversity of the population by fitness sharing, and improves the performance of the genetic algorithm. The process of genetic evolution, success rate, speed of convergence and scalability of the algorithm are studied under different network size by simulations and compares with associated algoritimas. A great deal of simulations result show that the algoritimas proposed are feasible and effective.
出处
《计算机与现代化》
2010年第2期47-51,共5页
Computer and Modernization
