Advancing Wideband Spectrum Sensing in Cognitive Radio Networks Using a Deep Learning Approach

Abstract

A novel spectrum sensing method is proposed for application in cognitive radio networks (CRNs). Wideband spectrum sensing is essential for locating and capitalizing on spectral holes, thereby facilitating effective spectrum use. However, problems such as noise uncertainty and signal-to-noise ratio (SNR) variations typically plague conventional sensing methods, resulting in less-than-ideal performance. Here, we employ convolutional neural networks (CNNs) to automatically acquire and extract important spectral features from raw spectrum data. The MATLAB-generated dataset of spectrum measurements was used to fine-tune the CNN model's ability to detect and categorize spectrum holes with more precise SNR estimates. The findings show that the deep learning-driven wideband spectrum sensing method is better than traditional approaches. CNN-based sensing efficiently characterizes spectrum utilization and improves detection of primary user (PU) activities. The proposed solution significantly improves the spectrum efficiency and resilience of CRNs by harnessing the power of ML, which leads to more effective resource utilization and less interference. Simulation results show that our proposed strategy produces more accurate spectrum occupancy assessments than existing methods, with a validation accuracy of 82.4%. This study introduces a cutting-edge deep learning model for wideband spectrum sensing in CRNs. Our deep learning methodology uses convolutional neural networks to automatically learn and adapt to dynamic and complicated radio environments, improving accuracy and flexibility over classic spectrum sensing approaches.