A Comparative Study of Bluetooth Indoor Positioning Using RSS and Machine Learning Algorithms

This paper presents a comparative study that explores the use of Bluetooth technology for indoor positioning by measuring Received Signal Strength (RSS). It assesses how effective this technology is by employing three machine learning algorithms: K-Nearest Neighbor (KNN), Weighted K-Nearest Neighbor (WKNN), and Convolutional Neural Network (CNN). The research begins with an analysis of the results obtained from field tests, following by some general insights in order to improving the accuracy of the positioning system. These positioning algorithms have shown moderate effectiveness, indicating potential for future business applications. The study is based on experimental data collected using beacons in a controlled indoor environment, along with the three algorithms. Importantly, all the algorithms perform well in our indoor testing setup, achieving an average accuracy of less than 2 meters. Notably, the CNN algorithm outperforms the others, achieving an impressive accuracy of 1.28 meters.