Exploration of Wi-Fi-Based Indoor Positioning System Using Linear Regression and K-Nearest Neighbour

In recent years, the use of Wi-Fi signals and machine learning techniques for indoor positioning has shown promising results. However, challenges such as privacy issues and resource constraints such as the availability of Wi-Fi access points (APs) and cost, need to be addressed to further improve the accuracy and applicability of these systems. In this paper, we aim to explore the performance of the indoor positioning system (IPS) using Wi-Fi signals. Three common machine learning models, linear regression, decision trees, and K-nearest neighbour (KNN) have been proposed for indoor positioning using the received signal strength indicator (RSSI). This study also includes the finding of the K value for the optimum performance of the KNN model in IPS. By comparing the experimental results of these two models, the optimised KNN model with K = 19 outperformed the linear regression and decision trees models with an accuracy of 96% in just 0.9 min, compared with an accuracy of 70 and 82% for the linear regression and decision trees models respectfully in three minutes. This suggests that the optimised KNN model is more effective and efficient in predicting the position of a device based on RSSI values in an IPS.