Two Classes of Improved Quickest Spectrum Sensing Approaches in Cognitive Radio

In this work, we mainly discuss two classes of quickest spectrum sensing approaches. The first is the improved single-channel detection based on cumulative sum (CUSUM) and generalized likelihood ratio, applied with primary user’s signal unknown. The other is wideband spectrum sensing based on Gobar transform and Wigner-Ville transform, implementing the transformation of the parallel sensing in time–frequency domain from the serial sensing in time-domain. The quickest spectrum sensing framework is developed to investigate the detection delay of the two approaches. Performance is evaluated using theoretical analysis and numerical simulations. Simulation results show that the improved single-channel detection is close to the CUSUM in performance though with unknown parameters. Moreover, in the wideband spectrum sensing, the detection delay of the ameliorated method is obviously superior to that of STFT with the certain false alarm interval.