Enhancing Indoor Positioning Accuracy using a Hybrid Li-Fi/Wi-Fi System with Deep Learning Support

This study proposes a new indoor positioning system that utilizes Li-Fi/Wi-Fi technology and the Received Signal Strength (RSS) triangulation method, aided by a Deep Neural Network (DNN) for better system accuracy. The proposed system uses several Light-Emitting Diodes (LEDs) as light emitters and photodetectors as receivers to determine the position of a user in an indoor environment. Photodetectors measure the RSS of a Li-Fi or Wi-Fi signal, which is then used to calculate the distance between the light sources and the user. RSS values are entered into a DL model to improve the accuracy of the positioning system by predicting the location of the user in more detail. The proposed system was experimentally tested and the results show that this method can achieve high positioning accuracy. The main objective of this work was to locate the mobile user within a room equipped with Li-Fi technology and obtain the best possible coverage of service to the user. In the first stage of data simulation, the triangulation technique achieved average errors of 2.174×10-14 cm, 6.450×10-14 cm, and 4.657×10-11 cm for the x, y, and z axes, respectively. This indicates the proximity of the simulation results to the actual ones. In the second stage, when RSS triangulation was applied with noise effects, the average error was 2.060×10-3 cm, 4.565×10-3 cm, and 5.110× 10-3 cm for the x, y, and z axes, respectively. A DL technique was used to handle noise, and the greatest error for the x, y, and z axes was 2.520 cm, 2.260 cm, and 4.230 cm in a 6×4×3 m indoor environment.