Abstract
Wideband primary user (PU) signal identification approaches based on enhanced forward consecutive mean excision (FCME) algorithms are proposed for cognitive MB-OFDM UWB systems. Each single subband is resolved into a series of contiguous frequency bins via the FFT engine. The identification of PU signals can thus be performed in frequency domain with our proposed methods, namely the localization algorithm based on double thresholding with prior one-dimensional median filtering (MF+LAD) and the FCME algorithm with binary integration (FCME+BI). In both strategies, two thresholds are employed in the detection decision making stage to perform the frequency bin cluster based PU signal identification. It is verified through numerical experiments that the two strategies can significantly improve wideband PU signals' bandwidth and signal-to-noise ratio (SNR) estimation accuracies.