Indoor Localization with Ensemble Machine Learning via Visible Light Communication Channels

Authors

Alzahraa M. Ghonim, Mechatronics Engineering Department, High Institute of Engineering and Technology (HIET), ElMahala Elkobra, EgyptFollow
Wessam Salama, Department of Basic Science, Faculty of Engineering, Pharos University, Alexandria, EgyptFollow

Abstract

An indoor localization system based on received signal strength, visible light communication (VLC) and several machine learning approaches is proposed in this paper. Our proposed framework is divided into two strategies. The first one is consisting of gathering our dataset based on MATLAB software to create indoor VLC channel model. While the second phase is depending on training the gained dataset using ensemble machine learning models. Specifically, random forest, decision tree and gradient boosting models. In order to evaluate the robustness of the proposed framework, several evaluation metrics are applied, specifically, training time, testing time, classification accuracy (CA), area under curve (AUC), F1-score, precision, recall, logloss, and specificity.

It is observed that the proposed framework achieves 0.991 for AUC, 0.986 for CA, F1-score, precision and recall. While logloss, and specificity obtains 0.077 and 0.991, respectively. Moreover, additional several evaluation error metrics are applied, mean square error (MSE), root mean square error (RMSE), mean absolute error (MAE) and coefficient of derivation of RMSE (CVRMSE). The gained results are 0.003 for MSE, RMSE is 0.055, MAE is 0.041 and CVRMSE 1.316 is Our proposed framework achieves the best performance based on different evaluation metrics.

Recommended Citation

Ghonim, Alzahraa M. and Salama, Wessam (2024) "Indoor Localization with Ensemble Machine Learning via Visible Light Communication Channels," Journal of Engineering Research: Vol. 8: Iss. 6, Article 16.
Available at: https://digitalcommons.aaru.edu.jo/erjeng/vol8/iss6/16