GNSS Interference Detection Using Statistical Analysis in the Time-Frequency Domain

This paper presents a precorrelation interference detection method based on statistical analysis in the time-frequency (TF) domain for global navigation satellite system signals. In particular, the short-time Fourier transform (STFT) is considered as the TF tool due to its linear property and low computational complexity. A goodness-of-fit (GoF) test is applied to each frequency slice in the spectrogram of the received signal, which approximately follows a chi-square distribution in the absence of interference. The expected probability density function (PDF) of the observed TF-domain samples can be computed based on an interference-free signal or the noise power estimate. Two versions of the proposed technique are presented: one based on the canonical STFT with the maximum overlap size, and the other based on the block-wise STFT using nonoverlapped samples. The canonical STFT-based method shows better detection capability at the expense of degraded false alarm performance caused by the PDF distortion in the canonical STFT samples. The block-wise STFT-based method alleviates the false alarm issue but slightly weakens the detection capability. Simulations show that the proposed canonical and block-wise STFT-based methods improve the detection performance for both narrow- and wideband interference in low jammer-to-noise ratio environments when compared with the existing GoF test applied to the time-domain samples.