An Effective Adaptive Threshold Based Compressive Spectrum Sensing in Cognitive Radio Networks

Abstract

Spectrum sensing is playing a vital role in Cognitive Radio networks. Wideband spectrum sensing increases the speed of sensing but which in turn requires higher sampling rate and also increases the complexity of hardware and also power consumption. Compression based sensing reduces the sampling rate by using Sub-Nyquist sampling but the compression and the reconstruction problem exists. In compression based spectrum sensing, noise uncertainty is one of the major performance degradation factor. To reduce this degradation, compressive measurements based sensing with adaptive threshold is proposed. In this technique compressed signal is sensed without any reconstruction of the signal. When the nodes are mobile in the low SNR region, the noise uncertainty degrades the performance of spectrum sensing. To conquer this problem, noise variance is estimated using parametric estimation technique and the threshold is varied adaptively. In the low SNR region, this proposed technique reduces the effect of noise and improves the spectrum sensing performance.