Spectral variation constrained power spectral density estimation for wideband spectrum sensing

In this paper, a novel power spectrum density (PSD) estimation approach is proposed for accurate and efficient wideband spectrum sensing in Cognitive Radio (CR) systems. Based on the observed signal from a wideband receiver, the goal of determining the fluctuation-free signal PSD is formulated as a constrained Bayesian estimation problem, subject to spectral variation constraints between neighboring spectral frequencies. The extracted signal PSD obtained using the proposed approach can then used in the energy detection process to make informed decisions with regards to the identification of free spectrum resources for opportunistic access by the CR. Experimental results using Monte Carlo simulations and real terrestrial digital TV (DTV) signal acquisitions show that the proposed approach allows for accurate PSD computation using wideband receivers under unknown noise and fluctuation conditions. Therefore, there is great potential for integrating the proposed method into existing energy detection methods for more accurate and efficient wideband spectrum sensing in CR systems under unknown noise and channel conditions.