| 摘要: |
| 针对软件定义网络(Software Defined Network，SDN)的负载均衡问题，为使网络的资源分配更加合理，防止网络拥塞，设计了一种基于Q-学习的负载均衡(Q-learning Load Balance,QLLB)算法，可根据网络环境自行作出决策，避免网络拥塞，实现网络资源的合理分配。与最短路径算法Dijkstra、蚁群算法进行的性能对比结果表明，QLLB算法有效实现了负载均衡，使得各个链路的带宽利用率更加平均，吞吐量分别提升了约8%和2%，可有效提升网络性能。 |
| 关键词: 软件定义网络 强化学习 Q-学习 负载均衡 |
| DOI： |
|
| 基金项目: |
|
| Q-learning load balancing algorithm in software-defined network |
| WANG Weifa,ZHANG Daming,LIU Kunqian,KE Feng,FENG Suili |
| (1.School of Electronic and Information Engineering,South China University of Technology,Guangzhou 510641，China;
2.The 7th Research Institute of China Electronics Technology Group Corporation,Guangzhou 510310,China) |
| Abstract: |
| For the load balancing problem of software defined network(SDN),in order to make the resource allocation more reasonable and prevent network congestion,a Q-learning load balance(QLLB) algorithm is designed,which can make decisions according to the network environment,avoid network congestion and realize the reasonable allocation of network resources.The experimental results show that，compared with Dijkstra and ant colony algorithm，QLLB algorithm effectively realizes load balancing,the bandwidth utilization of each link is more average,and the throughput is increased by about 8% and 2% respectively,which improves the network performance efficiently. |
| Key words: software defined network(SDN) reinforcement learning Q-learning load balancing |
《电讯技术》编辑部 