Improved indoor visible light positioning system using machine learning

Abstract

In this study, we present a novel Visible Light Positioning (VLP) method to reduce the localization error in an indoor environment. Machine Learning (ML) methods including Decision Tree (DT), Support Vector Machine (SVM), and Neural Networks (NNs) are used in combination with the LED Received Signal Strength (RSS) and the angle of a steerable laser. Zemax optics studio simulator is used to build a real indoor scene. Orange data mining software is utilized to apply ML techniques. Our numerical findings show that the suggested system outperforms the other RSS Visible Light Communication (VLC)-based models by reducing the localization error by more than 90% in some areas.

Artificial Intelligence in Visible Light Positioning for Indoor IoT: A Methodological Review

“…The model is trained in a simulated underwater environment and tested in a 3D simulation room, thus is viable for both air and underwater systems, obtaining an average positioning error of 11 cm. Authors in [62] various ML methods (i.e. DT, SVM, and NNs) are presented and compared for 3D indoor localization of PD receiver, based on RSS fingerprints and the angles of a steerable laser.…”

Kalman Filter-Driven Blind Source Localization for Passive 3D ToF Imaging

“…Blind 3D source localization has emerged as a fundamental requirement for many tasks such as passive imaging, where its appearance has been promoted by advancements in lighting infrastructure, which can potentially be exploited as an opportunity illuminator [1], wireless communication for better resource allocation [2], indoor positioning [3] and robot navigation [4]- [6], and Internetof-Things (IoT) [7]. Despite recent activity around the topic, the source location remains an open question for passive imaging and resource allocation for wireless communication from both a theoretical and practical perspective.…”

A MIMO-Enabled Free Space Optical Link under Log-Normal Fading/Gamma-Gamma Channel: Exploring an Optimal Modulation Scheme

“…Today's realistic workspaces require extraordinary bandwidth and connectivity performance levels to distribute data to exploit and provide a resource advantage. Te land, air, maritime, and space areas have common bandwidth and connectivity requirements [1][2][3]. However, wireless communication has an inherent challenge associated with its dependence on radio frequency (or RF); its nature makes it very vulnerable due to weather conditions.…”