Spectrum Sensing in Cognitive Radio Networks using Deep learning

In order to identify and make use of unused, underused frequency bands, cognitive radio (CR) makes extensive use of spectrum sensing (SS). Successful implementations of CR rely heavily on accurate spectrum sensing. As a result, this paper presents a novel multi-stage detector for use in high-fidelity signal and spectrum sensing equipment. As a first step is estimate the SNR of the sampled signal. Here, Convolutional Neural Network (CNN) architecture to estimate the SNR. Next choose a detection method that best fits predicted SNR of incoming signal. There are two ranges of SNR, one at low frequencies and one at high frequencies. In cases of high SNR, here "Energy Detector (ED) & Singular Value based Detector (SVD)," but in cases of low SNR, here employ modified Non-negative matrix factorization (MNMF). Here, use the Levy updated SLNO (LU-SLNO) method to choose the best possible weights for CNN. In addition, the combined results from many detectors provide definitive verdict on spectrum's tenancy and signal's presence.