Discrete Wavelet Transform Based Hybrid Edge Detection Technique for Wideband Spectrum Sensing in CRNs

An accurate and efficient spectrum sensing technique is needed to find the spectrum white spaces in a wireless radio environment for cognitive radio networks. Recently, a couple of edge detection algorithms in order to detect the edges of sub-bands of the received signal's wideband spectrum using wavelet transform (WT) is reported, which has the capability to detect singularities of sub bands in wideband spectrum. The discrete WT (DWT) based technique exhibits a better detection performance in low noise variance, but the performance is degraded with the increase in noise variance. In this paper, we propose a novel DWT based hybrid edge detection algorithm, where we use conventional DWT technique in low noise variance scenario and DWT-Multiscale-Sum (DWTMS) technique in high noise variance scenario depending on the threshold value, which is decided by the standard deviation of DWT coefficients. The simulation studies indicate that the proposed algorithm gives enhanced performance compared to the conventional WT and WT multiscale based edge detection algorithms in terms of total error probability (P
_T
), detection probability (P
_d
), and false alarm probability (P
_f
) in both the high and low noise variance scenario.