Rf Signal Sensing in A Wide Band Spectrum with Subnyquist Sampling for Cognitive Radio

Current trend in Cognitive Radio (CR) technology demands that the CR system not only detects the occupancy of the spectrum to identify any available spectrum opportunity, but should achieve the same dynamically in real time, in order to efficiently exploit the opportunity while leaving the PUs undisturbed. This seemingly contradictory requirement, demands for the CR to have the capability to achieve reliable and quick sensing of the data over the entire spectrum, which the conventional methodologies operating at Nyquist rates cannot meet. Fortunately, the knowledge that the signal spectrum is generally sparse, compressed sensing techniques paves way for meeting the demands. In this paper, subNyquist sampling technique is employed to reduce the computational complexities, still retaining a high rate of probability of detection. This is achieved by computing the correlation matrix of a finite number of noisy samples, and Simplified MUSIC-like (SML) algorithm is adapted to identify the vacant and active cells of the wideband spectrum. The performance of this methodology is evaluated by computing the detection probabilities as a function of number of samples and the SNR of the randomly input signal. Simulation results show that the proposed sensing algorithm is reliable even at lower sampling rates and is robust against noisy (low SNR) environment.