A Novel Radio Map-Assisted Network Planning Approach for Indoor Multi-Color VLC Networks

Abstract

Network planning requires repeated channel evaluation as access point (AP) positions are iteratively updated. In indoor visible light communication (VLC) scenarios, the rich reflections of light make accurate channel characterization particularly challenging, rendering ray-tracing (RT) or similar methods the main computational bottleneck during the exhaustive search of AP positions. Also, to mitigate the severe co-channel interference in densely deployed single-color VLC networks, multi-color reuse can be introduced, but jointly optimizing AP placement and color assignment further complicates deployment optimization. To address these issues, we propose a novel learning-based radio map fast generation (RMFG) method, where a dual-branch structure is designed to separately capture the distinct characteristics of line-of-sight (LoS) and non-LoS (NLoS) paths to improve RM accuracy. Based on the generated RMs, we further develop a graph coloring-based iterative optimization (GCIO) algorithm for efficient AP placement in multi-color VLC networks. Compared with RT, the proposed RMFG enables GCIO to achieve almost the same network capacity while reducing runtime by up to two orders of magnitude. Under the same RMFG-based channel evaluation, GCIO also outperforms baseline methods such as hippopotamus optimization (HO) in network capacity with significantly lower runtime.