Spectrum Sensing Using Fast Slepian Transform for Cognitive Radio Networks

Efficient and accurate spectrum sensing is vital for cognitive radio networks, especially in environments with low Signal-to-Noise Ratio (SNR). Traditional methods like energy detection often suffer from high false alarm rates and poor detection performance under noise. This paper presents a spectrum sensing approach based on the Fast Slepian Transform (FST), which leverages the time-frequency concentration properties of Discrete Prolate Spheroidal Sequences (DPSS) to project received signals into a sparse spectral domain. This enables robust detection of occupied subbands. We evaluated fixed and adaptive thresholding strategies, with the adaptive threshold optimized using the F1 score to balance detection accuracy and false alarms. Additionally, we introduce a computational efficiency metric based on floating-point operations (FLOPs) to evaluate the energy cost of detection. The FST method achieves reliable sensing with reduced complexity, making it well-suited for real-time applications. Simulation results across various SNR conditions demonstrate that the proposed FST-based method outperforms traditional techniques like Welch’s method, achieving higher detection probability, lower false alarm rates, and improved energy efficiency. These results confirm the method’s effectiveness for reliable and low-power spectrum sensing in cognitive radio networks.