Research On Physical Layer Security Algorithm Of Resource Allocation Based On Machine Learning In ICV

With the continuous development of unmanned driving,the security problem in the intelligent vehicle network has been paid more and more attention by the society.However,the traditional information security methods are not suitable for intelligent vehicle networking due to the high algorithm complexity and high computational requirements on equipment hardware.Moreover,the traditional information security methods have a high time delay.Obviously,they are not suitable for intelligent vehicle networking scenarios with very low time delay requirements.Physical layer security has the advantages of low complexity of algorithm and low requirement of computing power of equipment.Because of these advantages,more and more scholars attach importance to it.There are many useful technologies and methods in physical layer security,among which the resource allocation method is one of the physical layer security technologies.It mainly improves the security capacity by allocating radio resources and expanding the difference between legal channels and eavesdropping channels.The traditional physical layer security resource allocation algorithm is usually an optimization problem.However,due to the heterogeneity and high-speed variability of the Internet of Vehicles,this optimization problem is non-convex,and it is difficult to establish an accurate mathematical model,and it is difficult to obtain the analytical solution of the security capacity optimization problem.Aiming at the disadvantages of traditional physical layer security resource allocation algorithm,this paper designs a physical layer security resource allocation algorithm based on deep reinforcement learning.Aiming at the centralized resource allocation scenario,the base station is designed as an agent in this paper,and the appropriate state space,action space and reward function are designed.The security capacity of the system is optimized by optimizing the reward function.Base station according to oneself and vehicle detection channel in a fixed time slot upload information channel to get the channel state information,according to the channel state information,the base station for each legal vehicle allocation sub-band and transmitted power,and distributing the resource plan sent to various legal vehicles,legitimate vehicles according to the distribution of sub-band and the transmission power to communicate,according to the results of simulation experiments.This method improves the security capacity of the system.On the basis of centralized physical layer security resource allocation algorithm,a distributed resource allocation method based on multi-agent reinforcement learning is proposed to solve the problem of extra delay and cost of centralized resource allocation algorithm.The legitimate vehicle users are designed as agents,and the appropriate state space,action space and reward function are designed for the agents,and the cooperation of each vehicle is realized through the reward function to optimize the security capacity of the system.The vehicle obtains the channel state information,and chooses the sub-frequency band and transmission power to communicate autonomously according to the information.This method avoids the extra delay and overhead caused by communication in the centralized resource allocation method,improves the security capacity of the system,has the advantages of fast convergence,and makes it more close to the practical engineering application.