光树分割和共享光路合并优化任多播阻塞率

Tree-Split and Shared Lightpath-Merged for Optimization Bandwidth Blocking Probability of Manycast

针对弹性光互联数据中心网络中任多播业务的频谱利用率低和带宽阻塞率高等问题,论文提出了一种基于光树分割和共享光路合并算法.该算法选择对光树最大长度和调制等级影响较小的树枝加入光树,使生成光树频谱代价最小.当光树传输所需频谱资源不足时,该算法将光树分割为可选较高调制等级的多子光树传输,提高频谱利用率.当网络中有足够资源时,设计一种基于链路共享度的将满足条件的子光树合并到光树上传输的机制以降低频谱消耗.仿真结果表明:论文所提算法具有较低的带宽阻塞率和较高的频谱利用率.

In elastic optical inter-datacenter networks,for addressing the problems of lower spectrum utilization and high bandwidth blocking probability,an algorithm based on tree-splitting and shared lightpath-merged for manycast is proposed. The newbranches having minimum influence on light-tree＇s maximal length and modulation level are selected to insert into the spanning light tree,making the cost of spectrum consumption minimal. For reducing bandwidth blocking probability,the light tree is split into several sub-light trees with higher modulation when bandwidth is insufficient for the manycast. When the adequate spectrum bandwidth is found,the scheme based on link-sharing degree is designed to merge the split sub-light trees on a light tree for reducing spectrum consumption. The simulation results showthe proposed algorithm can get the lowest bandwidth blocking probability and highest spectrum utilization.