Distributed Compressive Spectrum Sensing Using Robust Power Estimation Techniques Under Non-Gaussian Noise

Compressive wideband spectrum sensing in a cooperative cognitive radio network requires each secondary user to reconstruct the original signal from the compressed measurement. The reconstruction algorithms are computationally complex and hence overloads the processing of each secondary user. To mitigate this, we reconstruct the signal using distributed hard thresholding (DiHaT) and distributed hard thresholding pursuit (DHTP) at each secondary user in a non-Gaussian environment. Further, we propose a robust power estimation technique using (i) diffusion maximum correntropy criterion and (ii) diffusion Huber cost function of the reconstructed signal to minimize the impact of non-Gaussian noise on detection. The simulations are carried out on a multi-carrier Universal-Filtered Multi-Carrier (UFMC) signal. The detection performance of distributed robust detectors using reconstructed signal is compared with its uncompressed counterpart.