Adaptive Sampling Point and Q-Learning–Based Sensing Threshold for Spectrum Energy Detection in Cognitive Radio Networks

Abstract

Spectrum sensing (SS) is a significant processing of cognitive radio networks (CRNs) that enables cognitive users to detect the underutilized or unutilized primary users (PUs) and licensed users spectrum for effectual usage. The threshold value selection is a vital step in determining the state (appearance/non-appearance) of PU in the spectrum sensing, and it has a significant impact on the detection and false-alarm probability. When a targeted sensing parameter is achieved at low SNR, other sensing parameter considerably degrades. In this manuscript, adaptive sampling point with Q-learning–dependent sensing threshold for spectrum energy detection in cognitive radio networks (ASSTQL-STSED-CRNF) is proposed. Adaptive sampling point and Q-learning (ASSTQL) employs an adaptive threshold mechanism that adjusts the detection threshold based on the current noise conditions, thereby improving the accuracy of signal detection. The proposed approach utilizes a Q-learning approach to optimize the sampling points and sensing thresholds. The ASSTQL-STSED-CRNF technique significantly enhances spectral detection performance, especially in scenarios with unpredictable noise levels. The proposed method is simulated in MATLAB. The simulation outcomes demonstrate an increased probability of identification as the signal-to-noise ratio (SNR) rises. The proposed model attains lower power spectral density and high throughput when compared with existing models.