Q-AODV Based on Reinforcement Learning Algorithm

Wireless ad hoc networks are very important in modern communication fields, in which routing protocols have been a hot research. AODV protocol has an important role in ad hoc networks. However, AODV uses flooding to find routes, and let the route from which receives the first RREP be the best routing, which means that the shortest time is the best. This mechanism is a reactive routing with expensive cost of time, and it does not have adaptive capacity to the network environment. In this paper, the solution of combining AODV with reinforcement learning is designed to make the improved AODV (which is called QAODV (Q-routing of AODV)) be adaptive to network environment. The node can reduce the frequency of finding routes by multi-path approach by QAODV. The simulation results show that latency and packet loss rate (PLR) of the improved routing protocol is significantly improved.