Research On Gpu High-efficiency Implementation Method Of Digital Receiving And Processing Of Complex Waveform Radar Signals

With the increasing complexity of the electromagnetic environment,the data scale of radar electronic reconnaissance digital signal processing has become larger and larger,and with the gradual development of intelligent electronic reconnaissance technology research,traditional radar electronic reconnaissance digital signal processing systems cannot meet the demand.GPU is widely used in many fields due to its good computing performance and has achieved good acceleration effects.This thesis studies the efficient implementation of a complex signal radar signal detection and processing system based on GPU processing architecture,and uses GPU high-parallel computing to improve the processing performance of the reconnaissance system.The work and contributions of the thesis are as follows:1.From the three aspects of GPU hardware architecture,memory system,and CUDA software architecture,the differences between GPU and DSP in terms of computing power,memory size and reading speed,software programming features and library functions are compared,and the advantages and disadvantages of FPGA+DSP architecture and CPU+GPU architecture are summarized,a radar signal digital detection and processing system based on the CPU+GPU architecture is designed on the basis of the processing flow of the FPGA+DSP architecture.2.The signal processing algorithms of the radar signal digital detection and processing system channelization,signal detection,modulation type recognition,parameter estimation and signal sorting modules are summarized,and the parallel processing method of the channelization module based on channel matching is designed.The process of signal detection module based on time-frequency joint is given,the algorithm implementation of modulation type identification,parameter estimation of complex waveforms and multi-pulse signal sorting is completed.3.The GPU efficient implementation method of radar signal digital reconnaissance and processing system is studied,including channelization,signal detection,modulation type recognition,parameter estimation and signal sorting modules,etc.;through the system functional simulation test,the execution efficiency of each module of the system is analyzed,comparing the effects of several optimization strategies of CUDA parallel programs,and verifying that the radar signal processing system based on the CPU+GPU architecture has good acceleration effects and radar signal processing capabilities.