Design and Implementation of Time-varying Wideband Interference Suppression Algorithm via GPU

Time-varying wide-band interference (WBI) may severely deteriorate the target detection performance of radar systems. Highly overlapped with target echoes in time and frequency domain, WBIs could hardly be mitigated by 1-dimensional (1-D) signal processing methods. To effectively suppress time-varying WBI, an interference mitigation method based on time-frequency (TF) 2-D interference detection is realized on graphics processing unit (GPU) device, in which there are three stages involved. Firstly, 1-D radar echoes are transformed into the 2-D TF domain using short-time Fourier transform (STFT). Then, a constant false alarm rate (CFAR) detector is introduced for WBI identification in the TF domain. After that, the detected interference is mitigated by the TF filter. Finally, the time domain purified signals can be obtained by inverse STFT. Considering the high computational complexity of the TF interference mitigation method, a GPU device is used to accelerate data processing. The experimental results show that the TF interference mitigation method via GPU could not only suppresses WBIs effectively, but also provides much higher data processing efficiency.