Direct-Conversion Spectrum Sensor Impaired by Symmetric α-Stable and α-Sub-Gaussian Noises

Abstract

Spectrum sensing in underwater cognitive acoustic networks (UCANs) can be impaired by impulsive noise generated by snapping shrimps. In mathematical analysis or simulations, the amplitude variations of this noise are commonly modeled by the symmetric alpha-stable (SS) distribution. As an alternative, the alpha-sub-Gaussian (SG) distribution can model both temporal correlation and amplitude variations. This article assesses the performance of underwater spectrum sensing with a direct-conversion receiver (DCR) under impulsive noise modeled by the SS and SG distributions. Several recent test statistics are compared, demonstrating that they have different degrees of robustness against impulsive noise and that the DCR is significantly less sensitive to this noise, compared to the conventional receiver model that does not take into account the influences of hardware characteristics into the performance of spectrum sensing.

Performance of Detectors for Cooperative Spectrum Sensing Under Laplacian Noise

“…It is easy to notice in the literature that the SαS distribution is typically adopted to characterize the heavy-tailed behavior of non-Gaussian noise, specially impulsive noise, in the CSS context, as can be seen for instance in [10]- [12]. For instance, in [13], the performance of a centralized CSS with data fusion is assessed, considering both SαS and alpha-sub-Gaussian (αSG) impulsive noise models. The latter keeps the amplitudes for each random variable following the SαS distribution, but introduces a non-zero temporal correlation among a group of adjacent samples.…”