摘要
设计了一种遗传算法,在使用较短通路情况下实现了链路和节点的负载均衡.在算法中采用了自然数编码、随机选取种群、简单的“双亲单子”交叉策略和固定的突变概率.以美国国家自然科学基金网络为例进行了仿真.对目标函数中的权值进行了讨论,分析了适应性函数值与遗传代数之间的关系,对种群选取范围与适应性函数之间的关系进行了验证.
A genetic algorithm for virtual topology design was proposed. Path length and load balancing of links and nodes were taken into account. The natural number coding, the random population selection, the simple crossover strategy of ＂two-parents-and-one-kid＂ and the fixed mutation probability were totally adopted. The algorithm was proved in virtual topology design of NSF（National Science Foundation）network. The effect of parameters of the object functions was discussed. The relation between generation and fitness was analyzed. And the effect of the selection range was referred to as well.
出处
《北京邮电大学学报》
EI
CAS
CSCD
北大核心
2006年第1期26-29,共4页
Journal of Beijing University of Posts and Telecommunications
关键词
遗传算法
逻辑拓扑
负载均衡
genetic algorithm
virtual topology
load balancing
