Analysis of Noise Uncertainty and Frequency Selectivity Effects in Wideband Multimode Spectrum Sensing

The present work is devoted to the comprehensive analysis of the detrimental effects of noise uncertainty in wide-band multimode subband based spectrum sensing over frequency selective channels. Unlike existing analyses that are mostly limited to single-band detection, an analytical model is firstly formulated for the case of wide-band multimode subband based spectrum sensing, considering both noise uncertainty and frequency selectivity. This model is also extended for the case of detecting a reappearing primary user (PU), as well as for the sensing scenarios where the frequency range of a PU is unknown. Novel closed-form expressions are derived for the corresponding probabilities of false alarm and probabilities of detection. The derived expressions are subsequently employed in quantifying the effects of noise uncertainty and frequency selectivity. A subband based scheme for the sensing of multi-channel primary systems is presented and analyzed. The proposed method is found to exhibit significantly reduced sensing time and greatly improved robustness against noise uncertainty compared to the basic wideband energy detection method.