Design and Demonstration of a Visible Light Based Semantic Communication System

While visible light communication (VLC) is emerging as a crucial technology for meeting high traffic demand in future network, the intrinsically limited modulation bandwidth of light-emitting-diodes (LEDs) necessitates innovative approaches to enhance data throughput. Semantic communication (SemCom), which focuses on transmitting the meaning of data rather than raw bits, is a perfect candidate to overcome this bottleneck in VLC. In this paper, we investigate how to build a visible light based SemCom (VL-SemCom) system, where non-orthogonal multiple access (NOMA) is leveraged to accommodate multiple users. First, we mathematically characterize the model of a VL-SemCom system, which simultaneously takes into account the SemCom transmission features, the NOMA multi-user access principle, and the physical layer constraints of VLC. Second, we address the core issue of power allocation in the VL-SemCom system with the aim to maximize the semantic throughput while considering user fairness and practical constraints of VLC. Finally, we build a preliminary hardware demonstration platform to achieve SemCom with the proposed power allocation strategy over real VLC links. The simulation and preliminary demo results together validate that our proposed power allocation strategy can effectively improve the performance of semantic throughput while satisfying the VLC constraints, and provide a valuable reference for the future practical deployment of VL-SemCom networks.