Research On Privacy Protection For Intelligent High-speed Rail Scenario

The overall layout of the new infrastructure has accelerated the transformation of China’s railways to digitalization and intelligence,with the development of technologies such as 5G,edge computing,and the Internet of Things,the amount of the data generated by the high-speed railway network has increased exponentially.The development of artificial intelligence can provide advanced technical means for data processing and analysis to meet the needs of diversified intelligent high-speed rail services.However,there are many new challenges to be faced while analyzing and processing the data.For example,the centralized architecture leads to large data communication overhead,transmission delay and privacy security risks.Traditional and complex privacy protection mechanisms have only focused on privacy issues and greatly weakened data availability,and with the rapid growth of data,these complex privacy protection methods will cause great delay in processing massive encrypted data.Secondly,in edge scenarios,due to the limited computing and communication resources of onboard terminals and the dynamic mobility of vehicles,it is difficult to effectively process data so as to reduce the data availability while protecting data privacy.Therefore,it is particularly important to provide a comprehensive and efficient privacy protection mechanism for high-speed railway scenarios.The main research contents of this thesis are as follows:(1)Aiming at the security threats and data leakage risks caused by the centralized computing mode,a distributed privacy protection architecture for intelligent high-speed rail is proposed.Typical application scenarios and security requirements of intelligent high-speed rail privacy protection are analyzed,based on the security requirements and the deployment concept of "public and dedicated network integration",a hierarchical privacy protection architecture is proposed.Security analysis and performance evaluation are conducted and experimental results verify the effectiveness and scalability of the proposed architecture.Finally,the challenges of data security and privacy protection,data availability,and resource limitation in high-speed railway scenarios are discussed,which is the foundation of next research.(2)Aiming at the data security and privacy protection of high-speed rail,an asynchronous update privacy protection mechanism based on obsolescence degree is proposed.Based on the distributed architecture,the impact of the uncertainty of the dynamic network topology and access channel caused by the high-speed mobility of the train on the performance is mainly considered.In order to realize data privacy protection and ensure highly reliable communication computation in dynamic timevarying networks,an asynchronous update privacy protection mechanism based on staleness is proposed jointly considering the heterogeneity and mobility of vehicles.The original problem is transformed into a dynamic double obsolescence correction problem,and local differential privacy is adopted to achieve dual protection of data security and privacy.Theoretical analysis and simulation results show that the proposed algorithm has better privacy protection effect and faster model convergence speed,and also achieves good performance under the non-independent identical distribution of data.(3)In the privacy protection scenario of intelligent high-speed rail based on federated learning,a joint optimization algorithm of client selection and resource management based on mobility awareness is proposed to solve the problem of link connection unavailability and resource limitation caused by the dynamic time-varying environment caused by the high-speed mobility of trains.First,quantify latency,energy consumption,mobility,and reputation models,then construct a joint optimization problem of vehicle selection and resource management by considering the impact of mobility on bit error rate and dataset quality.Finally,split the original NP-hard problem,dynamically optimizing the trade-off between maximizing the number of vehicle choices and minimizing the total energy consumption.The simulation results show that the proposed algorithm can effectively achieve the trade-off between vehicle selection and energy consumption compared with other algorithms while ensuring high reliability of communication.