基于双层递归神经网络模型求解最优组播路由

A Modified MTLRNN-Based Approach for Optimal Multicast Route in VPON

基于覆盖网络的组播作为一种新的IP网络组播解决方案已得到广泛关注。提出了一种利用改进的双层递归神经网络模型求解VPON网络环境下的QoS(服务质量)最优组播路由的方案。该方案在选择路由时综合考虑链路的可用带宽及节点的剩余处理能力,并运用一种基于改进的双层递归神经网络模型——MTLRNN进行求解,与其它启发式组播路由算法相比,该方案在满足应用的QoS要求的前提下,使全网的负载分配更加均衡,同时在解的有效性及接纳的组播应用会话数方面都有比较大的改善。

Overlay muhicast approach has become a promising multicast solution. In this paper a multicast routing scheme named MTLRNN for VPON is proposed. In this scheme two parameters-available bandwidth for overlay links and residual process power for overlay node are taken into account in selecting the route. Based on the traditional Hopfield NN a modified NN model-MTLRNN is proposed to solve the constrained multicast problem. Compared with other heuristic algorithms , this approach can effectively balance the network load among the physical resource and adopt more muhicast sessions in addition to guaranteeing the QoS requirements . Also the quality of solutions is analyzed.