Optimized Spectrum Sensing Algorithm Using ML

This research enhances spectrum sensing in cog- nitive radio networks using machine learning. We evaluate models like KNN, SVM, LR, RF, CNN, MLP, and LSTM for detecting primary user signals under varying signal-to-noise ratio (SNR) conditions. Feature extraction techniques such as energy detection, differential entropy, geometric mean, and log power improve classification accuracy. Experimental results show that CNN and LSTM perform well in low-SNR environments, while Random Forest offers robust performance across conditions. This study contributes to optimized spectrum utilization for telecommunications, IoT, and radar applications.