A sub-nyquist wideband spectrum sensing scheme using aliased spectral edge detection and database matching

The problem of sparse wideband spectrum sensing (WBSS) of multiple primary users (PUs) is considered in this paper. Assuming all PUs have a similar brick-type spectrum, we can employ sub-Nyquist sampling for the wideband signal, while still being able to identify the activities of all PUs. First, we estimate the Welch power spectral density (PSD) of the aliased signal, and then apply the multi-scale wavelet product and CA-CFAR detector to capture the rising and falling spectral edges of the aliased wideband spectrum. Then the edge feature set is compared to all possible cases of a pre-computed database in terms of a well-defined matching score, and all the PU channels can be sensed at the same time. Finally, the feasibility of the proposed sub-Nyquist WBSS scheme is evaluated by computer simulation as well as instruments.