摘要： 为提高遗传算法的效率以适于计算组播QoS路由，对算法的几个关键点作了深入研究。介绍了算法评价的两个指标：计算量和收敛率；提出了一种新的适应度函数；采用了比例选择与最优个体保存策略相结合的选择操作。经过大量的仿真实验，得出这种算法的最佳变异概率值应在0.4左右；当群体规模略小于每位基因座可选基因数的最大值时，算法的效率较高。

关键词: 组播, 服务质量, 遗传算法, Steiner树

Abstract: This paper deeply introduces several key points of the genetic algorithm in order to improve its efficiency in computing group broadcast QoS router. It introduces two indexes of algorithm evaluation: computation amount & convergence rate, puts forward a new adaptation algorithm, using select operation composed of proportion selection & the best individual preservation. Many simulation experiments have shown that the best variation probability of the algorithm should be around 0.4. The algorithm has better efficiency when the group scale is slightly less than the biggest gene amount for selecting of each gene base.

中图分类号: 

• TP18