Optimal Deep Learning Empowered Malicious User Detection for Spectrum Sensing in Cognitive Radio Networks

Abstract

Malicious user recognition for spectrum sensing in Cognitive Radio Networks (CRNs) is a serious safety feature to safeguard effective and trustworthy process of these systems. Spectrum sensing permits CRNs to identify and employ accessible spectrum bands. As well as it is available to prospective interference and mischievous actions. To preserve network integrity, recognition of malicious consumers is vital. Deep learning (DL) based malicious consumer classification powers advanced neural network frameworks to recognize and flag possible threats inside a network. By examining numerous amounts of information, DL techniques can distinguish patterns as well as anomalies that are connected with malicious user performance plus system intrusions, scams or irregular action. This technique provides flexibility benefit that permits a network to learn and develop in evolving threats. It also offers an effectual revenue of improving network security in the difficult and active digital landscape. Therefore, this article develops an Optimal Deep Learning Empowered Malicious User Detection for Spectrum Sensing (ODL-MUDSS) in the CRN. The main intention of ODL-MUDSS model focused on automated identification and classification of MUs in CRN. To accomplish this, the ODL-MUDSS model primarily applies deep belief network (DBN) methodology for automated and accurate detection of MUs. In addition, recognition performance of DBN technique can be enhanced by use of sand cat swarm optimization (SCSO) algorithm and thereby improves the detection results. The performance validation of ODL-MUDSS technique is observed under different processes. The comprehensive outcomes stated enhanced performance of ODL-MUDSS model over other existing models with maximum accuracy of 97.75%, precision of 97.75%, recall of 97.75%, and F-score of 97.75%.