Research On VANET Routing Algorithm Based On Reinforcement Learning

The concept of Vehicular Ad Hoc Network(VANET)was published at the ITU-T Conference on Automotive Communication Standardization in 2003,where Ad Hoc network technology was formally applied to vehicle driving.VANET is a key part of the intelligent transportation system.Its primary purpose is to ensure the safety of drivers and passengers by providing real-time road warning and public safety information to the driver.VANET can provide different application services,such as data download,access to the Internet,online video voice,to provide convenience for drivers and passengers.Routing algorithm is the most critical component of VANET,to a large extent affected the efficiency of VANET.This paper analyzes the current situation of VANET routing algorithm at home and abroad,and proposes routing algorithm for VANET environment.The research contents are as follows:(1)Aiming at the problem that the routing algorithm has low performance due to the vehicle mobility and traffic density in VANET,We use the electronic map to incorporate the intersection of the urban environment into the routing decision and adopt the stable "intersection to the intersection" form as the routing path.Based on the Q-learning method,the routing algorithm of Q-learning and electronic map is proposed,which solve the problem that the packet delivery rate is low due to the frequent disconnection of the link and the change of the traffic density of the road section.(2)In view of the slow convergence rate of QLAODV,this paper improves QLAODV based on heuristic Q-learning.By introducing the heuristic function and combining the delay information between nodes to guide the forwarding action of the nodes and speeding up the learning convergence rate,a routing algorithm based connectivity and heuristically accelerated Q-Learning is proposed.The simulation results show that the routing algorithm based on heuristic Q has higher delivery rate and lower delay than QLAODV.(3)We analyze the different application services and corresponding transmission messages contained in VANET,and classify different types of transmission messages and assign different priorities to them.A routing algorithm based priority and heuristically accelerated Q-Learning is proposed.In the case of VANET's network resources(such as bandwidth,cache,etc.),the network resources are preferentially assigned to messages with high service quality requirements.The simulation results show that the routing algorithm based priority and heuristically accelerated Q-Learning has better load balancing ability in different traffic cases and is suitable for high-load network environment.