Optimization of Spectrum Sensing in Cognitive Radio Using Machine Learning Approach

Spectrum sensing (SS) is the most significant and challenging task in the CR technique which identifies the availability of the spectral holes for opportunistic access. Spectrum sensing technique should ensure the reliable detection of the primary user signal in various scenarios like fading channels, Noise uncertainty and Hidden node problems. Recently ML technique is widely used in wireless communication applications. In this paper, ML based spectrum sensing scheme has been proposed, which enhances the detection accuracy under low SNR scenario while reducing the computational complexity involved in threshold computation required in conventional energy detection approach. ANN, SVM ML based approaches are used as a classical binary classifier to classify the received signal samples into two classes and subsequently provides the information about the presence of the primary user in the given band. Sensing performance is presented in terms of Probability of detection, accuracy and error rate as given in the simulation results.