Time-frequency compressed spectrum sensing in cognitive radios

In this paper, we investigate the use of time-frequency analysis for improvement of spectrum sensing in cognitive radios and exploit compressed sensing (sampling) to reduce the extremely high sampling rate of signal in time-frequency plane. We suggest a non-parametric spectrum sensing technique similar to energy detection utilizing time-frequency analysis to generally compromise between accuracy and sensing time, though the computational cost is significantly increased. As the representation of signals on the time-frequency plane is intrinsically sparse, thus we use the compressed sensing to achieve a significant reduction in the number of measurements. We propose using of different time-frequency representation such as short time Fourier transform, wavelet, Wigner-Ville and pseudo Wigner-Ville distribution in conduction of compressed sampling technique. The simulation results evaluate the performance of the proposed time-frequency compressed detectors compared to other time-frequency and energy detectors using basis pursuit and Bayesian compressive sensing reconstruction algorithms for AWGN and also Rayleigh and Rician fading channels.