Precision indoor location tracking using RSSI fingerprinting and machine learning

Abstract

In this project, the objective is to determine the effectiveness of using fingerprinting method with machine learning for indoor Wi-Fi localization. Research was done on methods of collecting Wi-Fi data. Using a python script, the author collected Wi-Fi RSSI data and his coordinates. The author also researched into Machine learning algorithms such as KNN regression and classification, and the steps needed to model the data. Using KNN regression, the author trained the model with collected datasets. Results from processing through the algorithm shows a low MSE and predictions of new data points are relatively accurate. With more Wi-Fi APs and more data, the author believes that this model can be improved to a better accuracy and can be implemented in future applications.