| With the rapid development of intelligent transportation system,millimeter wave(MMW)radar technology has many characteristics such as low cost,high detection accuracy,strong resistance to severe weather interference,and all-weather uninterrupted monitoring.Therefore,MMW radar technology is widely used in many scenarios such as autonomous driving and intelligent high-speed construction.However,the small size of MMW radar also brings the challenge of improving the angle resolution.If the angle resolution of MMW radar is too low,the MMW radar will not be able to correctly distinguish the vehicles or pedestrians with small spacing in front,which will cause fatal impact in the actual vehicle or high-speed detection application scenarios.Super-resolution algorithm can break through the Rayleigh limit and obtain a high resolution,but its calculation complexity is too high to be realized in practical engineering.Based on the research background above,the main research contents of this paper are as follows:1.On the basis of studying the working process of MMW radar and the principle of parameter estimation,aiming at the problem of low resolution of traditional angle measurement schemes,this paper focuses on the analysis of three subspace super-resolution algorithms of MUSIC,Root-MUSIC and ESPRIT algorithms.The simulation results show that the performance of Root-MUSIC and ESPRIT is similar and better than MUSIC algorithm under the condition of single snapshot.2.The way to improve the performance of the super-resolution algorithm in low complexity is studied.An angle measurement scheme combining spatial angle measurement with beam space super-resolution algorithm is proposed.The scheme can avoid the grating lobe interference in the beam space spectrum,and retain the super-resolution performance of the subspace algorithm.Moreover,the reduction of computational amount caused by dimension reduction and local search is much larger than the increase of computational amount caused by the introduction of spatial angle measurement,which effectively reduces the amount of computation.The simulation results show that the detection ability and estimation accuracy of this scheme are better than the basic MUSIC algorithm.3.The iterative cyclic angle estimation scheme is studied,and a fast super-resolution algorithm based on FFT and iterative interpolation is proposed.The algorithm does not need to predict the number of sources,and the number of iterations of the refined estimation is reduced by about 50% under the convergence strategy based on the residual change rate.The algorithm simulation is carried out based on the specific parameters of MMW radar.The simulation results show that the proposed algorithm has better estimation performance than RELAX and Root-MUSIC.Under the same SNR,the angle interval that the algorithm can distinguish is smaller.Under the condition of double super-resolution,the signal-to-noise ratio needed by the algorithm to achieve the detection probability of more than 90% is lower,which is only 8.12 d B.4.The 24 GHz roadside MMW radar is used for field testing,and the data collection is carried out for four scenarios of large truck,small car,two-car front and rear,and two-car parallel.The feasibility of the algorithm in practical application scenarios is verified by the measured data.In the multi-vehicle scene with interfering targets,the super-resolution angle measurement scheme proposed in this paper can remove interfering targets,reduce false alarms,and obtain better estimation results. |
PDF Full Text Request