This paper introduces a multiband spectrum sensing method that can accomplish the sensing task using sampling rates considerably lower than the ones demanded by the uniformsampling- based techniques. It utilises nonuniform sampling in conjunction with an appropriate spectral analysis tool. The approach is referred to as Spectral Analysis of Randomized Sampling (SARS), namely for the Random Sampling on Grid (RSG) scheme. The statistical characteristics of the adopted periodogram-type estimator are presented and the effects of the cyclostationary nature of the processed communication signals on SARS are addressed. Reliability guidelines that ensure the credibility of the sensing procedure amid a sought system performance are derived. Unlike a number of previously reported nonuniform sampling schemes e.g. Total Random Sampling (TRS), the RSG provides safeguards making it more suitable for implementation in hardware. Numerical examples testify the presented analytical results.
展开▼