Research On The Integration Of Communication,Sensing And Computing For 5G MmWave Connected Automated Vehicles

In recent years,automatic driving technology has developed rapidly and combined with various emerging technologies such as 5G mobile communication and Internet of things.Considering that a single vehicle has limited ability to obtain road condition information through radar sensing.Using 5G vehicle networking communication,perceptual information can be exchanged between vehicles,so that each vehicle can get global road condition information and make more accurate judgment on automatic driving behavior.Vehicles equipped with millimeter wave on-board radar sensor and 5G millimeter wave vehicle networking communication equipment to support automatic driving applications are defined as 5G mm Wave Connected Automated Vehicles,hereinafter referred to as Connected Automated Vehicles(CAVs).Traditionally,different frequency bands have been allocated for radar sensing and wireless communication.In the CAVs system,the two functions have the same purpose,and independent allocation is not conducive to the realization of the optimal performance of the system.Therefore,research on the integration of communication,sensing and computing for 5G mm Wave Connected Automated Vehicles is introduced,which aims to joint communication and sensing,and flexibly allocate the resources occupied by communication and sensing in order to achieve the optimal system performance.The main work of this paper is as follows:(1)In this paper,a CAVs system based on time resource segmentation is proposed.Radar sensing and wireless communication share equipment and spectrum resources and divide them in time.Two time allocation modes are proposed and verified to be reasonable.Taking mutual information as the performance index,the optimization problem of adjusting time strategy to maximize radar sensing performance in the scenario of CAVs is proposed.(2)For this nonconvex optimization problem,the maximum performance obtained by exhaustive search is impossible to achieve due to its high complexity,so the time allocation algorithm based on potential game is used to solve it.The effectiveness of the algorithm is verified by simulation.For example,under the 6-period allocation mode of10 vehicle scenario,the performance of the algorithm is improved by 22.75% compared with other optimal algorithms,and the more vehicles and the more complex the time allocation mode is,the greater the performance improvement of the algorithm is.And the comparison proves the rationality of the 6-period allocation mode.Although the algorithm has the best performance,the calculation time is seconds,which is not applicable to the CAVs system with low delay requirements.(3)Taking the solution obtained by the potential game algorithm as the training data,the convolutional neural network(CNN)model is trained.Using the trained CNN model,the solution performance can reach 84.87% of that of the potential game algorithm,and the suboptimal solution is obtained with very low computational complexity.The calculation time of the CNN model is less than milliseconds,which is practical for CAVs system.In this paper,the integration of communication,sensing and computing for 5G mm Wave Connected Automated Vehicles is studied.The time allocation problem of optimizing system performance is solved by using potential game method,and the sub optimal solution is obtained by using convolutional neural network with very low complexity,which has the realizability under the CAVs system.