SDN环境下基于QLearning算法的业务划分路由选路机制

随着当前网络的多元化发展，用户对于网络的需求也不断增加，网络的承载性受到极大挑战。在软件定义网络环境下，提出一种以业务属性为依据、基于QLearning算法的路由选路机制，设计了链路发现、链路分类、强化学习训练及Q值表下发4个模块，根据网络的实施情况对不同属性的业务流分配不同的路径以保障用户路由的 QoS。实验结果表明，所提算法可使数据流的总体分组丢失率低于 5%，对于部分属性的业务流可接近0分组丢失，同时对于时延也有大幅降低。

QLearning based business differentiating routing mechanism in SDN architecture

With the diversified development of the current network,users’ demand grows as well which brings great challenge for its load ability.A QLearning based business differentiating routing mechanism in SDN architecture was proposed to guarantee users’ routing QoS.To do that,four modules (discovery of link,classification of link,intensive learning and Q-value table sending) were designed to assign different paths to data streams of different attributes.The experiment shows that the proposed algorithm can decrease the packet loss rate to less than 5% and for some of the data streams,this rate is almost 0,simultaneously,it also helps the delay problem.