Privacy-preserving Computation in Edge Service Scenario of Internet of Vehicles:A Review ofTechnical Basis and Research Progress

With the deep integration of intelligent vehicles,edge computing,and wireless communication technologies,the “vehicle-road-cloud” collaborative intelligent IoV edge service system is rapidly developing,optimizing traffic efficiency and driving safety through real-time data processing.However,the interaction and computation of massive vehicle perception data(such as location trajectories and driving behaviors) in an open edge network environment face privacy leakage risks such as eavesdropping attacks and inference attacks.Although the existing privacy protection schemes have gradually enhanced the effect of privacy protection,the characteristics of dynamic topology and resource constraints in the edge environment of the IoV create a conflict between the strength of privacy protection and service performance.Privacy-preserving computation,as an effective means of privacy protection,is of significant importance for safeguarding users’ personal rights and promoting the sustainable development of the IoV industry,and has become one of the key research areas for ensuring the services in the IoV.Initially,it outlines the edge service architecture of IoV and analyzes the potential privacy leakage risks within it.Subsequently,based on the different mechanisms of privacy-preserving computation technologies,it categorizes and discusses the privacy-preserving computation methods for IoV based on data transformation,secure multi-party computation,federated learning,and trusted execution environment technologies.On this basis,a systematic analysis and comparison of these privacy-preserving computation methods are conducted from four key evaluation dimensions:privacy leakage risk,data utility,overhead,and scalability,along with corresponding optimization strategies.Finally,the challenges faced by privacy-preserving computation technologies for IoV edge services and future research directions are discussed.