Research on Software Real-Time Spectrum Analysis Technology Based on GPU
摘要
Real-time Spectrum Analysis (RTSA) technology is currently the most powerful electromagnetic spectrum monitoring and analysis technology. Within the captured bandwidth, it can monitor all transient signals, dynamic signals, and RF pulses. The core technology of RTSA encompasses seamless windowing of IQ data, seamless fast Fourier transform (FFT), and probability density statistics of the spectrum. In this paper, GPUs are used instead of FPGAs as the computing core for RTSA. By leveraging shared memory optimization, the CUFFT library, and atomic operations, the GPU achieves RTSA of 250MB/s I/Q Data online. Through offline data comparison testing, the computational performance of the GPU has been demonstrated to be 22 times faster than that of the CPU, thereby meeting the softwareization requirements of RTSA technology.