Simulation of Fast Threshold CFAR Processing Algorithm with CUDA

Signal level simulation is necessary to improve the fidelity of radar signal processing simulation, however the computation loads are too high for a CPU based platform. Compute Unified Device Architecture (CUDA) provides a new way to implement signal level simulation on a cost effective device. In order to compare the performances of signal level simulation based on Graphic Processor Unit (GPU) and CPU, we design and implement a typical module of radar signal processing system named with fast threshold Constant False Alarm Rate (CFAR) processing algorithm. Simulation results show that the operation rate based on GPU with CUDA is superior to that based on CPU.