基于人工鱼群算法的多播树演化寻优Evolvingmulticasttreebased.PDF

第33 卷第9 期 通 信 学 报 Vol.33 No. 9 2012 年9 月 Journa l on Communications September 2012 基于人工鱼群算法的多播树演化寻优 马炫，刘庆 （西安理工大学自动化与信息工程学院，陕西西安710048 ） 摘 要：提出了一种双域模型人工鱼群算法。算法采用前驱节点指向的编码方法形成多播树表示人工鱼，将搜索 空间分为可行域和非可行域。分别赋予可行域和非可行域的人工鱼不同的游动目标，设计行为算子自适应地执行 4 种人工鱼行为。数值实验结果表明，提出的算法可以有效利用非可行个体，具有较好的求解时延约束最小代价 多播树的性能。 关键词：多播树；时延约束；人工鱼群算法；双域模型 中图分类号：TP 18 ：A 文章编号文献标识码：1000-436X(2012)09-0001-07 Evolving multicast tree based artificial fish swarm algorithm MA Xuan, LIU Qing (School of Automation and Information Engineering, Xi’an University of Technology, Xi’an 710048，China) Abstract: An artificial fish swarm algorithm with two regions model was proposed . The algorithm used a multicast tree formed by the predecessor orientation method to represent an artificial fish, and divided the search space into feasible re- gion and infeasible region. The artificial fishes in the feasible region and the infeasible region were given different mov- ing objectives, respectively. The behavior operator was designed to adaptively implement four behaviors of artificial fish. Experimental results show that the proposed algorithm can effectively make use of the infeasible individuals and per- forms better in searching least-cost multicast tree with delay constraint. Key words: multicast tree; delay constraint; artificial fish swarm algorithm; two regions model 播树，使求解质量和求解速度得到大幅提高，详细 1 引言 算法可参考上述文献。 多播(multicast)是一种由源节点同时向多个目