A Routing Algorithm for High-Speed Mobile Ad Hoc Network Based on Deep Q Network

In this paper, a high-speed Mobile Ad Hoc Network (MANET) routing algorithm based on Deep Q Network (DQN) is proposed. The routing decision problem is modeled as a Markov Decision Process (MDP). The current node id with packets and the distance from the destination node are taken as the input state. The reward function is designed considering the neighbor nodes' number, the remaining capacity of the link and the relative speed between the current node and the next node. DQN is used to determine the next hop for data transmission. Simulation results show that compared with AODV, the proposed algorithm reduces the communication delay by 75% and increases the data arrival rate by 15%. Compared with Q-Routing, the communication delay and data arrival rate of this algorithm are more stable in the network where nodes move at high speed. The algorithm can adjust the routing decision in time when network topology changes and improve the stability of the network.