"NeuroHybrid Spectrum Sensing: Merging Energy and Feature Detection with Deep Learning for Unlicensed Spectrum Access"

Abstract

The ever-growing demand for wireless communication has intensified the need for efficient spectrum utilization. Cognitive Radio Networks (CRNs) have emerged as a promising solution by allowing unlicensed users (secondary users) to opportunistically access underutilized licensed spectrum. However, traditional spectrum sensing techniques like energy detection or feature-based sensing alone often fail to provide the required accuracy and adaptability in dynamic environments. This research proposes NeuroHybrid Spectrum Sensing, a novel hybrid approach that combines energy detection and cyclostationary feature detection with a deep learning-based decision engine. The system leverages the strengths of both physical-layer sensing and data-driven inference to enhance detection accuracy, minimize false alarms, and reduce spectrum sensing latency. Simulations demonstrate the model's superior performance over standalone methods in terms of spectrum occupancy prediction, signal-to-noise robustness, and computational efficiency. This work contributes a scalable and intelligent framework for real-time, reliable unlicensed spectrum access in next-generation wireless networks.