Deep Reinforcement Learning based Optimal Routing with Software-Defined Networking

An effective method for network congestion avoidance is to optimize the routing. We propose an optimal routing controlling that combines Deep Reinforcement learning(DRL) based policy optimization and Software Defined Network(SDN) based flow control. It uses DRL to learn the optimal network routing policy in different traffic patterns. SDN controller provides network traffic information for DRL agent to make decision, and implements flow routing control according to the decision. The results of simulation experiments show that our DRL-based route optimization has good performances in congestion avoidance.