Abstract
Physical Layer Security (PLS) in Cognitive Radio Networks (CRN) improves the confidentiality, availability, and integrity of the external communication between the devices/ users. The security models for sensing and beamforming reduce the impact of adversaries such as eavesdroppers in the signal processing layer. To such an extent, this article introduces a Secure Channel Estimation Model (SCEM) using Channel State Information (CSI) and Deep Learning (DL) to improve the PLS. In this proposed model, the CSI is exploited to evaluate the channel utilization and actual capacity availability throughout the allocation intervals. The change in channel capacity and utilization augments the need for security through 2-level key shared authentication. The deep learning algorithm verifies the authentication completeness for maximum channel capacity utilization irrespective of adversary interference. This verification follows mutual authentication between the primary and secondary users sharing the maximum capacity channel with high secrecy. The learning monitors the outage secrecy rates to verify failed allocations such that the replacement for allocation is pursued. Thus, the physical layer security between different user categories is administered through maximum CSI exploitation with high beamforming abilities. The proposed model leverages the secrecy rate by 10.77% and the probability of detection by 15.01% and reduces the interference rate by 11.07% for the varying transmit powers.