Utilizing Hybrid Machine Learning Approaches to Enhance the Accuracy ofIndoor Positioning Systems

This paper explores the utilization of hybrid machine learning methods to enhance the accuracy of indoorpositioning systems (IPS). Indoor environments present unique challenges for positioning technologies due tofactors such as signal reflection, multipath propagation, and interference from physical obstacles. Thesechallenges limit the effectiveness of traditional positioning techniques like GPS, which are primarilydesigned for outdoor use. In response to these issues, this research investigates the integration of variousmachine learning algorithms with advanced data preprocessing techniques to improve location estimationaccuracy. By employing a hybrid approach, the study aims to leverage the strengths of multiple algorithmswhile mitigating their individual weaknesses. The results indicate that hybrid methods can significantlyenhance user location estimation accuracy compared to traditional approaches. Specifically, theimplementation of these hybrid algorithms resulted in an improvement of up to ۳۰% in positioning accuracy.This enhancement offers promising implications for applications in navigation, asset tracking, and locationbasedservices, suggesting that hybrid machine learning approaches could play a crucial role in advancingindoor positioning technologies.

Keywords:

Indoor Positioning Systems(IPS) , Bluetooth Low Energy (BLE) , Ultra Wide Band(UWB) , machine learning(ML) , Hybrid method , Support Vector Machine (SVM) , Random Forest(RF) , Neural Network(NN)