Research On Signal Processing Technology For Automotive Radar

Automotive radars can work 24 hours a day without being limited by nature conditions,such as darkness,wind and so on.The radar has made great contributions to the driving applications.Automotive radars perform more and more functions as the concept of driverless vehicle is put forward.Accordingly,the market requests greater performance for radar signal processing algorithm,especially for the higher measurement resolution.In this thesis,the parameter estimation algorithm with high resolution based on 77 GHz automotive MIMO LFMCW radar is studied.Considering the sparsity of targets in the range-velocity-angle space,this thesis proposes four algorithms with high resolution for the automotive radar based on compressed sensing,FFT,random sampling and interpolation technology.It is verified that the proposed algorithms show higher resolution than traditional algorithm through computer and AWR1843 platform.And the influence of noise and indicators is analyzed.In this thesis,following works were carried out in the aspects of algorithm research,model construction,simulation verification,experiment verification and so on.1.Firstly,the traditional parameter estimation algorithms for 77 GHz automotive MIMO LFMCW radar are studied.And the constraints for the resolution of range,velocity and angle are analyzed respectively.Then,the compressed sensing theory and reconstruction algorithm are introduced.2.In order to obtain high resolution of range and velocity for the automotive radar,compressed sensing theory is utilized based on the sparsity of targets.A novel idea based on compressed sensing for estimating parameters is proposed according to the form of the beat signal.And three high resolution models are constructed.CS-SS model offers high resolution of range or velocity.The data amount is huge if CS-SS model is used to improve resolution of range and velocity simultaneously.To address the problem,FFTCS-SS model and RS-CS-SS model are proposed.Simulations verify the effectiveness and robustness of the three algorithms.In addition,the performance is researched.3.The resolution of angle is limited by the small antenna aperture of the automotive radar.Based on the angle estimation model of the MIMO array,the method aiming at high angle resolution developed from compressed sensing theory is studied firstly.Whereas,the results of the algorithm are unsatisfactory.Consequently,a novel method named Interpolation-CS which extraordinarily improves the resolution of angle is proposed.Under certain conditions,simulations prove that the targets with an angle interval of one degree can even be distinguished.4.The hardware platform based on AWR1843 radar sensor is built.In outdoor scenario,real-time measurement experiments and raw data based verification tests were performed.Results verify that the proposed algotithms can provide high resolution of range,velocity and angel for the automotive radar.