Adaptive spectrum sensing in optical OFDM using high-order QAM for VLC applications

Abstract This paper presents a comprehensive performance analysis of spectrum sensing techniques for Optical Orthogonal Frequency Division Multiplexing (Optical OFDM) systems under high-order modulation in Rician and Rayleigh fading channels. Using MATLAB simulations, we evaluate the proposed Hybrid Energy-Cyclostationary Detection with Machine Learning (HECD-ML) algorithm against conventional methods such as Energy Detection (ED), Cyclostationary Detection, Matched Filter (MF), and standard Optical OFDM. Results show that the proposed HECD-ML approach significantly outperforms existing techniques in both detection accuracy and bit error rate (BER). Specifically, the proposed method achieves a probability of detection (Pd) = 1 at −2.5 dB SNR in Rician and −2 dB in Rayleigh channels, offering a detection gain of 2.5–9.5 dB over traditional techniques. Additionally, it achieves a BER of 10 −3 at 6.2 dB (Rician) and 7.2 dB (Rayleigh), demonstrating an SNR gain of 2–8.6 dB. These improvements indicate enhanced sensitivity, faster detection, reduced error rates, and greater spectral efficiency in low-SNR and fading conditions, making the proposed method highly suitable for dynamic visible light communication (VLC) scenarios. However, the increased computational complexity and reliance on simulation-based validation highlight the need for future work focusing on real-world implementation and system optimization under practical constraints.