A Q- Learning based Routing Optimization Model in a Software Defined Network
Yupeng Wang, Xin Zhou, Xinyue Fan · 2021
Aiming at the routing optimization problem of software defined network (SDN). A Q-learning based routing algorithm is proposed, which generates routing and forwarding rules to guarantee user QoS. A novel reward function adopted in the Q-learning model is proposed in this paper for the SDN with non-identical available bandwidth among routers and heterogeneous QoS packets. The proposed routing algorithm based on Q-learning considering both transmission bandwidth and delay, solves the routing optimization problem through the experience on the interactions among transmission environment, actions and rewards. By using the proposed algorithm, a routing strategy namely the Q table is obtained, and the data will be forwarded according to the Q table for transmission delay optimization. Through the simulation experiments, the proposed routing optimization algorithm based on Q-learning outperforms the conventional routing algorithm in the aspect of transmission delay.