Robust Indoor Localization in VLC Systems Using Machine Learning: A Comparative Study under Environmental Variability
DOI:
https://doi.org/10.65419/albahit.v5i3.159Abstract
Indoor localization has a growing demand, primarily due to its widespread applications in areas such as robot navigation, indoor navigation, virtual reality, and asset tracking. With recent advancements in solid-state devices, Visible Light Communication (VLC) has emerged as a promising solution for high-speed data communication and accurate indoor localization, harnessing existing LED lighting infrastructure. This research incorporates five machine learning approaches for location estimation using visible light commissioning Received Signal Strength (RSS) as a parameter for location estimation. The study evaluates both static and dynamic scenarios under noise-free and noisy conditions. An indoor environment is simulated with a 5 × 5 × 3 m3 room and a 3 × 3 LED grid. Five Machine Learning (ML) models, namely Linear Regression (LR), Random Forest, K-nearest neighbours (KNN), Support Vector Regression (SVR), and Multi-Layer Perceptron (MLP), leverage RSS data to generate an estimation of the position of the object in a 2D environment. Evaluation parameters used to determine the performance are Root-Mean-Square Error (RMSE), training/inference time, and spatial error maps. Simulation results show superior performance of the KNN and Random Forest models in a noisy environment. In addition to that, results show that ML offers a robust alternative to analytical methods for Visible Light Positioning (VLP). Future work includes real-world validation and exploration of deep learning techniques
References
[1] J. Singh, N. Tyagi, S. Singh, F. Ali, and D. Kwak, “A systematic review of contemporary indoor positioning systems: Taxonomy, techniques, and algorithms,” IEEE Internet of Things Journal, 2024.
[2] R. Wang, G. Niu, Q. Cao, C. S. Chen, and S.-W. Ho, “A survey of visible-light-communication-based indoor positioning systems,” Sensors, vol. 24, no. 16, p. 5197, 2024.
[3] Z. Zhu, Y. Yang, M. Chen, C. Guo, J. Cheng, and S. Cui, “A survey on indoor visible light positioning systems: Fundamentals, applications, and challenges,” IEEE Communications Surveys & Tutorials, 2024.
[4] S. Bastiaens, M. Alijani, W. Joseph, and D. Plets, “Visible light positioning as a next-generation indoor positioning technology: A tutorial,” IEEE Communications Surveys & Tutorials, 2024.
[5] M. Saadi, T. Ahmad, M. Kamran Saleem, and L. Wuttisittikulkij, “Visible light communication–an architectural perspective on the applications and data rate improvement strategies,” Transactions on emerging telecommunications technologies, vol. 30, no. 2, p. e3436, 2019.
[6] Q. Chen, F. Wang, B. Deng, L. Qin, and X. Hu, “Indoor visible light positioning system based on memristive convolutional neural network,” Optics Communications, vol. 577, p. 131340, 2025.
[7] M. Saadi, A. Bajpai, D. Z. Rodriguez, and L. Wuttisittikulkij, “Investigating the role of channel state information for mimo based visible light communication system,” in 2022 37th International Technical Conference on Circuits/Systems, Computers and Communications (ITC-CSCC). IEEE, 2022, pp. 1–4.
[8] B. A. Vijayalakshmi, P. Gandhimathi, and M. Nesasudha, “Vlc channel characteristics and data transmission model in indoor environment for future communication: an overview,” Journal of Optics, vol. 53, no. 2, pp. 933–939, 2024.
[9] K. Zhang, Z. Zhang, and B. Zhu, “Beacon led coordinates estimator for easy deployment of visible light positioning systems,” IEEE Transactions on Wireless Communications, vol. 21, no. 12, pp. 10208–10223, 2022.
[10] H. Q. Tran and C. Ha, “Machine learning in indoor visible light positioning systems: A review,” Neurocomputing, vol. 491, pp. 117–131, 2022.
[11] N. Liu, L.-H. Hong, P. Feng, H.-N. Yang, and J.-Y. Wang, “Performance analysis for visible light communications with gaussian-plus-laplacian additive noise,” Telecommunication Systems, vol. 85, no. 3, pp. 373–387, 2024.
[12] T. O. Hodson, “Root mean square error (rmse) or mean absolute error (mae): When to use them or not,” Geoscientific Model Development Discussions, vol. 2022, pp. 1–10, 2022.
