为了求解覆盖网络中最优组播服务节点MSN组合问题,提出一种基于混合实数编码的遗传算法(HRCGA)。采用实数编码方式对MSN组合进行编码,利用K-medoids聚类距离算子进行节点相似性度量,并根据MSN的实际特点,构造含有惩罚因子的适应度函数,限制不良个体参与进化。同时,针对标准遗传算法SGA局部搜索能力弱的特点,HRGCA引入个体进化控制策略.理论分析和仿真结果表明,该算法有效克服了传统K-medoids算法易陷入局部极小值、对初始中心选值敏感的问题。通过与传统SGA的仿真结果对比,HRCGA进一步提高了MSN组合全局寻优能力。
In order to select optimal multicast service node (MSN) of layered overlay network, a hybrid real-coded genetic algorithm (HRCGA) is proposed. The fitness function with penalty operator is created to prevent bad chromosomes participating evolution. Also, an evolving control strategy is proposed to enhance the local search ability. Results of computer simulation show that HRCGA not only avoids converging to local minimum value, but also is robust to initialization. And the global search ability is enhanced compared with SGA.