Application of compressed sensing in wideband cognitive radios when sparsity is unknown

We present a novel solution for wideband spectrum sensing when the sparsity of the sensed signal or the occupation rate of the frequency band of interest is unknown. Moreover, the proposed method offers the first solution to sense the frequency band of interest when almost fully occupied. Generally, our method employs compressed sensing to acquire the received signal at sub-Nyquist rates. Then, the sum (or the average) of the frequency spectra per each frequency subband is estimated using least squares optimization where the sparsity is assumed as the total number of frequency subbands. Moreover, a detection criterion is formulated. A reduction in the design complexity of the sampling stage as well as the computational complexity of the recovery stage is achieved compared to compressive wideband spectrum sensing methods including the sequential compressed sensing method.