Enhancing Spectrum Sensing in Cognitive Radio Networks using Deep Learning Models: A Solution for Low-SNR Challenges

Spectrum sensing is a fundamental component of cognitive radio networks (CRNs), enabling efficient spectrum utilization by detecting and leveraging underutilized frequency bands. However, low Signal-to-Noise Ratio (SNR) environments present substantial challenges to reliable spectrum detection, leading to increased false alarms and missed detections. In this paper, we propose an enhanced deep learning approach utilizing advanced neural networks, including Generative Adversarial Networks (GANs), to improve spectrum sensing in low-SNR regimes. By generating high-fidelity, synthetic training data across varying SNR levels, our model gains robustness against noise interference, enabling superior feature extraction and detection accuracy. Comparative evaluations with other deep learning models demonstrate that the GAN-augmented spectrum sensing framework achieves significant performance improvements in low-SNR conditions, offering a promising solution to enhance spectrum reliability in CRNs. This work underscores the potential of GAN-based data augmentation and denoising in addressing noise-related challenges in spectrum sensing, with implications for the next generation of intelligent, adaptive CRN architectures.