SVM-Based Enhanced Spectrum Prediction Model for Cognitive Radio Networks

Efficient spectrum utilization is a major challenge in cognitive radio networks due to increasing wireless communication demands. To address the limitations of conventional machine learning approaches such as the Support Vector Machine (SVM), this work proposes a Convolutional Neural Network (CNN)-based model for accurate spectrum prediction. The proposed CNN automatically learns complex, nonlinear relationships among key parameters such as transmission power, frequency, and duty cycle, eliminating the need for manual feature extraction. By leveraging convolution and pooling operations, the model effectively captures spatial and temporal dependencies within spectrum data, enabling reliable classification of occupied and idle channels. The CNN is trained and validated using real-world datasets to enhance spectrum availability prediction for dynamic access by secondary users. Experimental results demonstrate that the proposed method achieves significantly higher accuracy, better adaptability to changing environments, and greater robustness to noise compared to traditional algorithms. This deep learning–based approach provides an intelligent and scalable solution for real-time spectrum management, paving the way for efficient communication in next-generation cognitive and 5G wireless networks.

Keywords: Cognitive Radio, Spectrum Prediction, Convolutional Neural Network, Dynamic Spectrum Access, Deep Learning.