A low-complexity optical OFDM physical layer for AI-native 6G visible light communication networks

The sixth generation (6G) networks native to artificial intelligence (AI) are expected to enable intelligent and real-time communication services, which increase the processing needs across the network nodes. Therefore, future visible light communication (VLC) systems based on intensity modulation and direct detection (IM/DD) need low-complexity physical-layer solutions. Conventional optical orthogonal frequency-division multiplexing (O-OFDM) methods have considerable computational complexity because they require Hermitian symmetry, multiple FFT/IFFT operations, and additional signal processing, such as DC biasing or asymmetric clipping. In this paper, we propose a novel Binary-Sign Optical OFDM (BSO-OFDM) scheme as a low-complexity physical-layer solution for AI-native 6G VLC networks. The proposed method uses all available subcarriers and avoids Hermitian symmetry by conveying absolute-value information alongside binary-sign information, thereby avoiding DC bias and asymmetric clipping. We assess the proposed system over an additive white Gaussian noise (AWGN) channel, multipath-AWGN, and physical line-of-sight (LOS) with diffuse VLC channels at a fixed spectral efficiency of 2.5 bits/s/Hz. Theoretical and simulation results show that BSO-OFDM reduces computational complexity by more than 75% compared to ACO-OFDM and more than 45% compared to DCO-OFDM and Flip-OFDM. It also provides about a 4 dB peak-to-average power ratio (PAPR) reduction compared to ACO-OFDM and Flip-OFDM while maintaining similar bit error rate (BER) performance and improved multipath robustness. The results show that BSO-OFDM is a promising, low-complexity physical-layer candidate for future optical wireless networks.