ABSTRACT
Cognitive radio (CR) has been suggested as the solution to spectrum scarcity due to the fixed allocation employed worldwide by regulatory bodies.A secondary usercan opportunistically access the licensed frequency bands without causing harmful interference to the licensed user. In order to avoid interference to a primary user signal,the CR has to be aware about the spectrum usage inthe geographic area in which it wants to operate. The process of spectrumsensing is a fundamental task for obtaining this awareness and the result of this process determines the successful operation of cognitive radio. Energy detection is one of the methods of spectrum sensing with the lowest computational complexity but with low performance at low signal to noise ratio. Exploring energy detection has led to the application of many techniques one of which is the use of time-frequency analysis. This method employs distribution techniques for analyzing the energy spectral density of an observed signal with a view to setting a sensing threshold. However, the distribution techniques that were used in literature suffered from the problem of cross-terms which affect the analysis of the resulting distribution thereby leading to poor sensing performance at low signal-to-noise ratio. Smoothed pseudo Wigner-Ville distribution (SPWVD) of the time-frequency analysis has been employed in this work to reduce the effect of cross-terms and a better sensing threshold was gotten validated through comparison with the existing work which employed pseudo Wigner-Ville Distribution (PWVD) with an average reduction of 2.7% and 3% for additive white Gaussian noise (AWGN) channel, 4.1% and 4.7% for Rician channel, 6.4% and 8% for Rayleigh channel in the probabilities of missed detection and false alarm respectively. These results showed that significant reduction was achieved using SPWVD to set threshold. This work was carried out using the MATLAB R2013b time-frequency tool box.