Research And Implementation On Radar Signal Processing Algorithm Based On Zynq

In recent years,the frequency modulated continuous wave radar technology used in military,meteorological and geological exploration has frequently appeared in many civil fields.The automobile industry is the most concerned field.Frequency modulated continuous wave radar has the advantages of small size and strong weather adaptability,often used as a sensor for advanced driving assistance system.It is susceptible to time and requires efficient signal processing algorithms and high-performance computing platforms.This thesis studies radar frequency modulated continuous wave technology,focusing on the Constant False Alarm Rate(CFAR)processing algorithm.In order to reduce the time-consuming of signal processing,Zynq So C is used to design and implement the algorithm.This thesis is divided into three parts :(1)This thesis introduces the system architecture of frequency modulated continuous wave radar.The multi-cycle beat signal model is obtained according to the radar transmitting and echo signals and then expounds on the range and velocity measurement principle based on the Range Doppler Map.These classical constant false alarm algorithms are often used in single-dimensional radar echo data: cell average CFAR,greatest of CFAR,smallest of CFAR,and ordered statistical CFAR.In terms of background noise distribution,the number of targets,and the adjacent degree of targets,their detection performance is different,so this thesis conducts simulation experiments on these algorithms.By analyzing the simulation results,it is found in this thesis that the cell average CFAR shows the best performance in uniform background noise,and the ordered statistical CFAR algorithm shows the best performance in multitarget scenarios and target mask suppression.(2)Since the target detection is carried out on the Range Doppler map,this thesis studies the two-dimensional CFAR algorithms based on the classical CFAR algorithms: OSCA CFAR and FOSCA CFAR.FOSCA CFAR uses an efficient two-dimensional reference window,so its computational complexity is lower than OSCA CFAR.This thesis changes the Range and Doppler dimension reference window size based on several two-dimensional reference windows for constant false alarm detection to reduce computational complexity further.After analyzing the test results,it is found in this thesis that changing the number of reference cells in the Doppler dimension of the reference window is almost difficult to affect the detection performance of OSCA CFAR.However,this change can significantly simplify the computation.In this thesis,a two-dimensional reference window is proposed to improve the computational efficiency and avoid the interference of the cell under test.The reference window analyzes the data in the actual road,proving that the reference window is feasible to a certain extent.(3)Aiming at the problem of low data throughput and large delay in the traditional FPGA +MCU heterogeneous architecture,this thesis selects a new processor: the Zynq So C processor,and proposes a signal processing system architecture according to its characteristics.Since the radar front end has a long time interval in multiple sampling periods,one-dimensional FFT is inserted in this interval.This thesis enables the sampling and onedimensional FFT to be completed simultaneously and reduces the processing time of the whole radar algorithm.This thesis also designs a two-dimensional CFAR detector based on the OSCA CFAR algorithm and the improved two-dimensional CFAR reference window,and its function is verified.After analysis,the signal processing system architecture greatly reduces the whole signal processing time.