基于双种群遗传算法的MPLS流量工程研究  

The Research Based on Dual Population Genetic Algorithm for Traffic Engineering in MPLS Networks

作　　者：黄镇建[1] 蔡群英[2] 

机构地区：[1]韩山师范学院物理与电子工程系,广东潮州521041 [2]韩山师范学院数学与信息技术系,广东潮州521041

出　　处：《计算机仿真》2010年第3期202-205,共4页Computer Simulation

摘　　要：针对因特网流量的快速增加和对服务质量要求的不断提高,为避免网络数据流过大造成拥塞问题,利用MPLS流量工程的路由算法成为业界研究的重点。介绍了MPLS流量工程的工作机制,并提出了数学模型。MPLS流量工程的关键问题是LSP的分布优化问题,针对这一问题的Np-hard特性,提出一种基于双种群策略的遗传算法求解方法,算法采用自然数编码来提高搜索效率。算法中,两个子种群周期性地交流最优染色体,进一步提高了性能。仿真实验表明,提出的算法能有效提高搜索速度,增加获得最优解的概率。Rapid growth of Internet traffic and increasing requirements for service quality have made traffic engineering an essential consideration in the operation of large internet backbone network. This paper introduces the traffic engineering working - mechanism and math model. The key problem of the MPLS TE is the optimized distribution of LSPs. As to the problem＇s character, this paper introduces a genetic algorithm with two populations to solve it, and it uses the natural number coding method to improve search efficiency. In this algorithm, two suhpopulations are evolved independtly and exchange best chromosome in a periodic way. Simulation shows that the algorithm can effectively speed up the searching process and boost the chance of achieving the best results.

关 键 词：流量工程 遗传算法 双种群 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]