Fast Compressive Wideband Spectrum Sensing Based on Dictionary Linear Combination

Wideband spectrum sensing is applied to detect the unused spectrum hole for cognitive radios. Given the sparseness of the spectrum, the compressive sensing can be applied to the coarse sensing of spectrum. In the spectrum sensing, there are several spectrum bands in a frequency range with the same length. In order to sense this kind of spectrum with high speed and low computation complexity, this paper proposed a novel approach based on linear basis combination and basis pursuit algorithm. The main thoughts of the approach are as follows: before sensing, we carry on a linear combination computation for the power spectrum in each subband to obtain the corresponding power level, and use the same way to set a new sensing dictionary. Then, based on the traditional BP algorithm, the new model is formulated to recover each subband power level spectrum. The newly proposed method is different from the traditional one, for it uses linear basis combination to greatly reduce the dimension of the dictionary. Thus the computation complexity was reduced, and fast spectrum sense can be obtained. Both theoretical analysis and numerical results demonstrate the advantages of the novel one.