Research On Super-Resolution Technology Of Vehicle-Mounted Millimeter Wave Radar

In recent years,with the rapid growth of economy and technology,the number of automobiles has increased rapidly and autonomous driving technology has become a hot topic and research area.And how to improve the angular resolution of automotive radar is still an urgent problem.There are two methods to improve the angular resolution of radar: virtual aperture method and super-resolution algorithm.Firstly,Multi-input Multioutput(MIMO)technology can achieve a larger virtual aperture without changing the physical aperture size,so as to improve the angular accuracy and angular resolution of radar;secondly,super-resolution algorithm can break the limitation of "Rayleigh limit".Secondly,the super-resolution algorithm can break the limitation of "Rayleigh limit" to further improve the angular resolution of radar without changing the size of radar array and achieve super-resolution.This thesis aims to improve the angular resolution of vehicle-mounted millimeter radar,and carries out theoretical and methodological research around the problem of super-resolution angular measurement of vehicle-mounted millimeter wave radar,the specific contributions of this thesis are as follows:1.The working principle of the vehicle-mounted millimeter wave MIMO radar,including the measurement principle of distance and velocity angle,is studied.The working methods of time division,frequency division and code division MIMO radar are studied,especially the advantages and disadvantages are compared,and the MIMO system used in this thesis is given.2.A high-resolution goniometry algorithm based on virtual aperture expansion is studied.Firstly,a virtual aperture reconstruction method based on frequency hopping signal is proposed,by transmitting frequency hopping signal in time,getting echo data of different frequency signals at the receiving end,and then performing phase compensation to form a larger virtual aperture with narrower beam width,and the simulation verification shows that using 10 frequency hopping signals,the obtained array beam width is 1/9 of the original one.secondly,a virtual aperture expansion method based on linear prediction is proposed.The simulations show that when the extrapolated array elements are equal to the real array elements,the extrapolated beamwidth is close to 1/2 of the original one,which can effectively improve the cross-resolution capability.3.Super-resolution goniometry algorithm based on MIMO radar is studied.Firstly,a classical super-resolution algorithm is introduced,followed by a sparse recovery superresolution algorithm based on singular value decomposition,and a grid refinement method is proposed to reduce the complexity of the search through a coarse-to-fine gridbased search.Simulation experiments show that the sparse recovery super-resolution algorithm can distinguish two targets with an angular separation of 0.05 degrees under an array of 12 transmits and 16 receives.4.The field experiments of the super-resolution algorithm for vehicle-mounted millimeter wave radar are carried out.Firstly,the experimental platform used in this thesis is introduced,and secondly,the scene setting method is given according to the real road environment.The effectiveness of the super-resolution goniometry algorithm based on sparse recovery is verified through the processing of real measurement data.The experimental results show that the algorithm can achieve lane level target discrimination within 50 meters after the number of virtual array elements is greater than 40,and it can still achieve lane level target discrimination within 50 meters when the number of snapshots is 5.