软件定义车辆网络中自适应负载均衡路由算法

Adaptive load balancing routing algorithm in software defined vehicle networks

城市交通车辆密度高,为解决车辆通信过程中,数据包转发时中继节点负载分配不均衡、限制车联网中吞吐量等性能问题,本文提出在基于软件定义的移动自组网络架构中引入强化路由,来自适应学习负载分配决策,根据邻居节点的带宽状态学习负载分配收益;通过强化学习构建状态-策略表,使节点在不同状态下进行带宽分配决策,最终实现SDN数据层内的车辆相互协调,寻找最优路径。仿真结果表明,该算法可实现网络负载的均衡分配。与传统的路由算法相比,当车辆数为300辆时,该算法的丢包率可低至20%以下,端到端时延低于4 s,网络能量消耗更加均衡。

Urban traffic is characterized by high vehicle density,which poses challenges for vehicular communication such as unbalanced load distribution among relay nodes during data packet forwarding and limited throughput within the vehicular networks.To address these performance issues,this paper proposes the introduction of reinforced routing within a Software Defined Mobile Ad hoc Network(SDN-MANET)architecture to adaptively learn load distribution decisions.This is achieved by learning the load distribution benefits based on the bandwidth status of neighboring nodes.Through reinforcement learning,a state-policy table is constructed,allowing nodes to make bandwidth distribution decisions under varying states,ultimately enabling vehicles within the SDN data layer to coordinate with each other to find the optimal path.Simulation results indicate that the algorithm can achieve balanced network load distribution;compared to traditional routing algorithms,with a vehicular count of 300,the proposed algorithm can reduce the packet loss rate to below 20%,achieve end-to-end latency of less than 4 seconds,and result in more balanced network energy consumption.