Research On Multi-frame Tracking Before Detection Of Array Signals And Fast Implementation On GPU

Array radar completes the estimation of the direction of angle by calculating the phase difference of the target echo in the snapshot data composed of multiple arrays,which has important application value in the military and civilian fields.Array signal multi-frame tracking before detection algorithm builds array measurement model based on array radar multi-antenna echo data.By introducing the concept of multi-frame measurement data and target motion model,multi-frame energy accumulation is carried out to complete the excavation of weak targets.It can be used in the fields of radar detection,microphone array sound source positioning and communication.Taking the vehicle-mounted millimeter-wave array radar as the starting point,this paper studies the parameter estimation method of the millimeter-wave array radar,the conventional millimeter-wave signal processing flow,the construction of the array measurement model,the multi-frame accumulation of the array measurement,and the GPU parallel implementation of the algorithm.The specific work content is as follows:1.Using Texas Instruments' vehicle-mounted millimeter-wave array radar AWR1642 as the platform,the system structure of the millimeter-wave array radar is studied in detail,the principles of LFMCW radar ranging,speed and angle measurement are deduced,and the signal processing flow of the conventional millimeter-wave array radar is introduced.2.Aiming at the problem of applying multi-frame tracking-before-detection algorithm to array signal processing,starting from Bayesian estimation theory,the conventional multi-frame tracking-before-detection algorithm is derived,and the combination of array signal processing and multi-frame detection tracking algorithm is analyzed.Difficulty,an array signal measurement model that can effectively accumulate energy is designed,and a detection and positioning algorithm for effectively discovering weak targets through a multi-frame accumulation method is proposed,which solves the problem of difficult detection of weak targets in the array signal,and The effectiveness of the algorithm is verified in simulation experiments and measured data processing.3.In view of the difficulty in engineering implementation of the tracking algorithm before multi-frame detection of array signals,GPU was selected as the signal processing platform,and the GPU physical structure and programming model were studied in depth,combined with signal processing requirements and GPU parallel architecture to accelerate the implementation of radar signal processing algorithms,And developed and tested a set of offline data processing software to accelerate the processing of offline data.The above proposed algorithm is compared with the traditional array signal processing method in simulation and actual measurement.The experimental results verify the effectiveness and stability of the proposed algorithm.After many tests,the developed offline processing software for measured data can stably output the correct detection and positioning results,achieving the purpose of accelerating data processing in engineering.