Implementation Of Signal Processing Software Based On GPU For Passive Radar

Passive radar has a wide range of application because of its ability to resist interference, good concealment and anti-stealth. But compared to traditional radar, the signal processing of passive radar is more complex and the processing data is bigger. GPU has developed rapidly in recent years, and its peak processing capacity is 10 times of the same level CPU. Especially the development of CUDA architecture, GPU programming is becoming simpler and more purpose, and not needing to use the image API. So in many areas it has been widely used. In order to meet the requirements of real-time system, this paper presents a CPU+GPU+PCIE based architecture to finish the main signal processing algorithm in the passive radar. At first the paper describes the development of the passive radar at home and abroad, and introduces the basic principle of GPU in signal processing.Then the paper introduces the main algorithm of the passive radar including correlation, Keystone, MTD, CFAR and gains the conclusion that the architecture of CPU+GPU+PCIE is better than that of DSP+FPGA. Then the paper describes the CUDA programming framework and focuses on the implementation of the passive radar in GPU. At last the paper gives a specific implement of signal processing for passive radar including the overall structure of the system, the realization of the data transmission and the display and control function. Then the paper describes two main test methods including the laboratory test of radar simulator and the actual experiment in the field. The construction of the experiment platform and the result of the experiment are shown at last.