| With the rapid popularization of intelligent vehicles,the gradual maturity of drone technology,and the continuous enrichment of road network services,the construction pace of intelligent transportation is also accelerating.Utilizing UAV assisted IoV as a new architecture to optimize road communication and network service quality is currently one of the hot research directions.In order to deal with the huge amount of data generated in the road,the introduction of edge computing technology is regarded as an appropriate solution,but the ensuing network security problems such as DDoS attacks also threaten the network security of the Internet of Vehicles.Aiming at the characteristics of high dynamic network topology and low tolerance of network delay in UAV assisted IoV,a mobile target defense technology for vehicle networking proxy switching based on road communication constraints is proposed.It is combined with a trusted scoring mechanism based on user behavior to achieve precise identification and blocking of malicious users,thus achieving the goal of defense.In addition,to alleviate the network pressure near the DDoS attacked area,this thesis proposes a UAV trajectory planning strategy based on reinforcement learning to make decisions on the flight direction of the UAV,thereby assisting the proxy switch strategy to screen malicious users in the network and improve network service quality.On this basis,for the situation where multiple attackers launch DDoS attacks simultaneously in the vehicle networking technology,this thesis further proposes a UAV cluster deployment strategy based on multi-agent reinforcement learning,which uses a centralized multi-agent proximal policy optimization algorithm to make decisions on the deployment of the UAV cluster and improve the overall network quality.This thesis uses actual data to simulate the vehicle networking scenario and conducts comparative analysis with other existing algorithms.The experimental results show that the proposed algorithm and strategy are highly feasible and applicable. |
