Research On Passive Radar Signal Processing And Target Positioning Technology Based On CPU And GP

Passive radar refers to all radar systems that do not radiate electromagnetic waves by themselves and obtain target information by intercepting electromagnetic waves reflected by the target.The distributed multi-node passive radar system has better anti-jamming capabilities,low-altitude detection and anti-stealth capabilities than a single station by deploying multiple nodes.In a passive radar system,on the one hand,the target echo is easily interfered by direct waves and multipath clutter,and target detection faces many difficulties.On the other hand,modern radar technology requires the processing platform to have real-time processing capabilities.The existing processing equipment such as FPGA,DSP,etc.are restricted by hardware platform and portability.This dissertation starts from passive radar signal processing,studies related issues such as direct wave suppression,signal accumulation,target detection and positioning,and completes the parallel processing of passive radar signal processing based on the CPU and GPU heterogeneous architecture and target positioning calculations.The main research work is as follows:1)Researched on the conventional algorithms of passive radar signal processing,including beam forming,direct wave suppression,range-doppler processing and constant false alarm rate detection.The design plan of distributed multi-node passive radar signal processing is given,and the ECA algorithm and its improved algorithm in time domain clutter suppression are analyzed.The advantages and disadvantages of the ECA-B and ECA-S algorithms,the amount of calculation and the applicable occasions are discussed.The feasibility of the signal processing scheme is verified by simulation,and theoretical support is provided for the followup work.2)The multi-station target positioning algorithm is studied,mainly for AOA and TDOA.The positioning principle and positioning accuracy are respectively derived,and solutions to the possible positioning ambiguity problem of solving the time difference positioning equation based on the Chan algorithm are proposed.Combined with the simulation results,the relationship between the positioning accuracy and the angle measurement error,the time difference measurement error,the station site measurement error,the station layout method and the station spacing is analyzed,and the positioning scheme used by the system is given.3)Based on the heterogeneous architecture of CPU and GPU,the passive radar signal processing and target positioning are realized,and the model and scheme design of the heterogeneous architecture are given.For multi-node signal processing,a method of asynchronously processing data transmission and data preprocessing of each node using CUDA stream is proposed,which improves the degree of concurrency of the module.By changing the calculation method of the matrix in the ECA algorithm,the purpose of reducing the amount of calculation is achieved.Use batch processing to complete the matched filter processing of multi-node signals,which improves the parallelism of the module.The time difference positioning module solved by the Chan algorithm is implemented based on the CPU,and the multi-threading is used to solve the problem of continuous processing of continuously arriving radar data frames.After many tests and verifications,the signal processing and target positioning based on the heterogeneous architecture of CPU and GPU can meet the requirements of real-time and accuracy.