Robust Gaussian kernel based signal detection in the presence of non-Gaussian noise
Huadong Lai, Weichao Xu · 2020 7th International Conference on Information, Cybernetics, and Computational Social Systems (ICCSS) · 2020
This letter presents a robust Gaussian kernel based detector for the detection of random signal distorted by non-Gaussian noise. The proposed detection scheme does not need any prior information of the signal, noise and channel. The GKD detector is shown to be asymptotic energy detection (ED) as the width of Gaussian kernel becomes sufficiently large. The asymptotic null distribution of the GKD statistic is derived, enabling us to determine the detection threshold. Simulation results illustrate that in terms of receiver operating characteristic (ROC) curve and detection probability, the GKD detector not only performs comparably with ED in the presence of Gaussian noise, but also achieves better performance than the state-of-the-art detectors whether the noise is pure Gaussian or non-Gaussian.