Abstract

Visible light communication (VLC) and light fidelity (LiFi) are emerging as promising optical wireless technologies for sixth-generation (6G) networks, offering high-capacity, spatially confined, and electromagneticinterference-free wireless access by exploiting existing lighting infrastructure. Beyond conventional data transmission, the same optical infrastructure can simultaneously support illumination, indoor positioning, and environmental sensing, a convergence formally defined as lighting, sensing, and communication (LiSAC). This review presents an integrated perspective on AI-enabled LiSAC in VLC/LiFi networks for 6G. The fundamentals of VLC/LiFi, including the Lambertian channel model, intensity-modulated direct-detection constraints, and LiFi networking architecture, are first introduced. The evolution from communication-oriented VLC toward coordinated LiSAC, encompassing visible-light positioning achieving decimeter-to-centimeter accuracy and visible-light sensing demonstrating occupancy and gesture recognition above 90% accuracy, is then reviewed. The role of artificial intelligence across channel estimation, signal detection, adaptive dimming, localization, resource allocation, and handover is examined, with deep reinforcement learning demonstrating up to 48% data rate improvement and federated learning enabling privacy-preserving distributed optimization. Photonic and devicelevel enablers, including GaN micro-LEDs with bandwidths exceeding 800 MHz, perovskite quantum dot color converters with bandwidths 40 times greater than conventional phosphors, wavelength-shifting fiber-based detectors, and optical intelligent reflecting surfaces, are reviewed as the hardware foundation for 6G LiSAC systems. Finally, key challenges, including LOS blockage, multi-function trade-offs, dataset scarcity, hardware-algorithm mismatch, real-time implementation, privacy, and standardization, are discussed alongside a structured three-horizon research roadmap. The review highlights that future VLC/LiFi systems must move toward AI-native, photonic-integrated, privacy-aware, and experimentally validated architectures capable of supporting multifunctional 6G services.