摘要
针对多约束QoS多播路由的NP-Complete特性,提出一种可控的多播树分解与合并策略,使多播树的生成在兼顾低费用的同时具有多样性,有效克服多播路由优化的局部极值问题。基于该策略设计蚁群算法,分解蚂蚁种群为与多播目标点相对应的蚂蚁子群,引入基于“死点”惩罚和多播树奖惩的信息素更新机制,提高了算法的收敛速度。仿真实验表明,该方法能有效地解决QoS多播路由问题,且随着网络规模的增大保持了良好的性能。
In view of the fact that the multiple constrained QoS multicast routing is a NP-Complete problem, a controllable strategy for decomposing and composing multicast tree is introduced. Different trees of little cost will be gained under the strategy, so that the problem of local best can be solved effectively. An ant colony algorithm, which divides the ant colony into ant assembly according to the ends of multicast, is designed on the basis of the strategy. The pheromone updating mechanism is improved with the ＂dead-point＂ punishment and the encouragement and punishment to the multicast tree strategy, so that the algorithm can converge fast. Simulation results demonstrate that the method can solve the QoS multicast routing problem effectively, and perform well as the size of network increases.
出处
《国防科技大学学报》
EI
CAS
CSCD
北大核心
2007年第2期117-122,共6页
Journal of National University of Defense Technology
基金
国家自然科学基金重点资助项目(60234030)
关键词
多播路由
多播树
蚁群算法
multicast muting
multicast tree
ant colony algorithm
