Retracted: A hybrid‐supervised learning model for compressive spectrum sensing in fifth generation(5G) cognitive Radio enabled internet of things (IoT) networks

Abstract

Summary5G Internet of Things (IoT) networks are characterized by wideband radio frequency spectrum utility and are therefore of primary importance for efficient means of sensing the wideband spectrum characterized by high bandwidth. A cognitive radio network (CRN) intelligently does the sensing of authorized users ideal spectrum and allocates the same to the demanded unauthorized user. Conventional energy detection and k‐means schemes associated with CRN perform well for narrowband applications, whereas they are not quite suitable for wideband applications. Hence, a compressive collaborative sensing scheme together with deep neural network learning model (CCS‐DLNN) has been proposed to sense the information from the compressed and reconstructed information signal. Based on the extracted features, decision on presence or absence of primary user (PU) in the received signal has been observed. This paper proposes a deep learning neural network model for learning the dynamic change in the input spectra. Accordingly, this paper also updates the weights associated with the neurons to converge upon the target objective. The performance of the proposed sensing scheme has been evaluated pertaining to probability of detection, sensing error, and accuracy of detection of idle channels. The proposed work will be very useful for the upcoming generation departing to be implemented with 5G IoT networks.