Optical IRS-Aided Indoor Visible Light Fingerprint Localization
Guopeng Cheng, Fasong Wang, Ziye Zhang, Xingwang Li, Nguyen Cong Luong, Arumugam Nallanathan · IEEE Transactions on Green Communications and Networking · 2025
In response to the challenge of substantial localization inaccuracies in Line-of-Sight (LoS) obstructed areas within indoor Visible Light Positioning (VLP) systems, this study presents a visible light fingerprint localization technique enhanced by optical Intelligent Reflecting Surfaces (IRS). Initially, a fingerprint matrix database is established utilizing Received Signal Strength (RSS) data. The proposed method, termed IRSPba, employs the Differential Evolution Based on Weight Issue (DE-WI) algorithm alongside the Weighted K-Nearest Neighbor (WKNN) algorithm to compute localization weight vectors. This integration proves particularly advantageous for Internet of Things (IoT) deployments in multipath-rich environments where traditional Radio-Frequency (RF)-based localization underperforms. To further enhance localization precision, the fingerprint matrix database is expanded into a fingerprint tensor database. This extension, in conjunction with the DE-WI and WKNN weight calculation algorithms and a two-step localization strategy, leads to the development of an improved optical IRS-assisted VLP method, designated as IRSPad. Simulation experiments substantiate the efficacy of both proposed localization methods, revealing that the IRSPad method achieves superior localization accuracy relative to the IRSPba method. Furthermore, within the same localization framework, the DE-WI weight calculation algorithm demonstrates superior performance compared to the WKNN algorithm regarding localization outcomes under the conditions examined in this research.