Machine Learning Based Indoor Localization using Wi-Fi and Smartphone in a Shopping Malls

Abstract

The availability of sensors in smartphones has led to indoor positioning solutions. However, the accuracy of these techniques remains uneven as a straightforward solution to indoor positioning. Solutions based on Wi-Fi signal strength work in favor of the idea of controlling infrastructure costs. Our work attempts to explore other learning algorithms and make more robust trade-offs between accuracy and power.Our work also focuses on using classification-based learning algorithms to achieve higher accuracy. By using methods to select the appropriate model and using more complex on-device learning algorithms. Accurate indoor positioning, based on general sensors and user permission, allows for a great location based experience. Machine learning (ML) based methods are also used to improve the quality and efficiency of services.To verify the accuracy of the models, we reviewed the algorithms using several comparisons between a variety of machine learning approaches.We have verified the system’s performance using measurements of a smartphone’s Wi-Fi RSS (Really Simple Syndication) sensor. Evaluation results show that the gradient boosting method achieves the best internal feature localization accuracy of more than 95%.