Radar Simulation Modeling And Fast Implementation Using GPU

With the popularization of information warfare,the development of a new generation radar system requires high computational resources.The introduction of simulation technology can significantly improve the efficiency of radar system design demonstration and test evaluation,and greatly reduce the consumption of human,material,financial and other resources.However,with the improvement of radar system intelligence,the radar system design is faced with the problem of big data,which seriously affects the real-time performance of the simulation system,making it difficult to meet the high-precision requirements of the radar system in complex and changeable warfare.Graphics Processing Unit(GPU)provides a new research idea for solving the problem of poor real-time performance in the simulation process because of its powerful parallel computing capability.Based on the above background,this paper focuses on the research on the parallelization optimization technology of high-efficiency radar simulation system based on multiprocessor collaboration,which effectively improves the real-time performance of radar simulation system.The main completed work is as follows:1.Aiming at the problems of disordered radar signal simulation structure and poor scalability in complex electromagnetic environment,a standard and unified radar simulation system framework is designed based on the software modular design idea by analyzing the functions and combat tasks of the phased array radar system.And for the problem of poor real-time performance of radar simulation system,the feasibility analysis of algorithm parallelization in radar simulation system is carried out,and a radar signal based on the combination of Compute Unified Device Architecture(CUD A)and Array Fire graphics processing general acceleration library is proposed.The scheme of processing algorithm parallel programming hybrid model,and the evaluation index of computing efficiency under multi-processor platform is determined.2.In order to solve the problem of poor real-time performance of radar echo signal simulation and radar signal processing method,the parallel design technology under multiprocessor cooperation is studied,and a radar echo signal simulation and radar signal parallel processing method based on multi-processor cooperation is proposed.Firstly,in the simulation of radar echo signals,a target echo signal modeling method based on CUDA stream parallel,a clutter signal modeling method based on multi-processor cooperation,and a repeater jamming signal modeling method based on CUDA and ArrayFire are proposed.Then,in the parallel processing method of radar signal,the parallel acceleration method of Sidelobe Cancellation(SLC)and Sidelobe Blanking(SLB),the pulse compression and accumulation method based on ArrayFire,and constant false alarm detection optimization acceleration method based on CUDA stream are proposed.3.By controlling the variables that affect the computational efficiency,a comparative experiment of the computational efficiency of radar echo signal simulation and radar signal processing methods is designed in the Qt simulation software platform,which proves the feasibility of GPU parallel acceleration.Through the design of anti-jamming experiments,target search and tracking experiments based on multi-processor cooperation,the anti-side lobe interference ability,fast beam adjustment and gate tracking ability of the radar simulation system are verified.Synergistic reliability and real-time performance.And in target search and tracking experiments,the speedup of signal generation can be up to 50 times in a pure CPU environment.