The Study Of Adaptive Routing Protocol In VDTN

In the context of Vehicular Ad-Hoc Networks(VANETs),the resources of nodes in the network are very limited,and the network topology changes rapidly.It is difficult to find a stable end-to-end communication path between nodes.The routing algorithm of Delay Tolerant Network(DTN)can tolerate the severe conditions such as huge delays and frequent connection interruptions,so using the routing algorithm in DTN is a feasible solution.Therefore,finding an efficient routing and forwarding algorithm is the main problem for VDTN(VANET/DTN).In this paper,three VDTN routing algorithms are proposed in the following aspects.(1)Node encounter rate is the rate at which a node meets other nodes.The higher node encounter rate represents the more active the node is.The more mobile and active nodes have a higher probability of encountering other nodes and are more suitable as relay nodes.The network environment changes rapidly so that the routing algorithm should have the ability to learn the network environment dynamically.Q-learning is a value-based reinforcement learning method,and the node activity-based routing algorithm proposed in this paper(Q-based multiple routing,QBMR)uses an optimal tree based message diffusion strategy,reinforcement learning of the network environment using the Q-learning algorithm,and a reward factor of Q-value based on the node encounter rate to select the appropriate relay nodes.Simulation results show that the algorithm has a higher delivery rate and lower network overhead than related algorithms.(2)In the simulation experiments of the QBMR algorithm,it was found that the QBMR algorithm performed poorly in certain network environments,such as low node speed and sparse number of nodes.This is due to the fact that in this type of network environment,in order to ensure efficient relay node selection,the QBMR algorithm message diffusion phase delivers message copies under more stringent conditions,resulting in a slower network start-up that This leads to the problem of poor final message delivery rate.To address this problem,this paper improves the message spreading strategy in the message spreading phase of the QBMR algorithm and proposes an improved routing and forwarding algorithm(Q-based hybrid multiple routing,QHMR)based on the QBMR algorithm.The simulation results show that QHMR has better performance in low-speed node environment and sparse node environment.(3)In VDTN,some nodes will refuse to take the responsibility of relaying messages because of insufficient resources,and such nodes are called selfish nodes.The reputation value is usually used to evaluate and mark the selfishness of nodes.The existence of selfish nodes in the network can greatly affect the performance of the network,so the routing and forwarding algorithm should consider avoiding selecting selfish nodes as relay nodes for message forwarding.The node reputation value based routing and forwarding algorithm(Q-learning multiple credit routing,QMCR)proposed in this paper as a routing algorithm without limiting the maximum number of message copies.It uses Q-learning algorithm for reinforcement learning of the network environment and determines the reward factor of Q-value based on the node reputation value to control the number of message copies per diffusion number of message copies at a time.To avoid network congestion caused by too many message copies in the network,the nodes in the QMCR algorithm use message caching optimization,which effectively alleviates this problem.Simulation results show that the algorithm maintains a high delivery rate while the network overhead is at a low level.In this paper,the routing algorithms and related technologies in VDTN are studied.Through the design and simulation verification of the above three algorithms,the results show that the routing problems in different network environments are better solved,and provide valuable references for the development of VDTN routing algorithms.