Mining raw GPS readings for deep profiling of location contexts - part III

• Main Author: Ong, Daniel De Quan
• Other Authors: Luo Jun
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2024
• Subjects: Computer and Information Science
• Online Access: https://hdl.handle.net/10356/175962
• Institution: Nanyang Technological University

While both outdoor and indoor localization methods are flourishing, how to properly marry them to offer pervasive localizability in urban areas remains open. Recently proposals on indoor-out door detection make the first step towards such an integration, yet complicated urban environments render such a binary classification incompetent. Fortunately, the latest developments in Android have granted us access to raw GNSS measurements, which contain far more information than commonly derived GPS location indicators. Although this ability is limited to several up-to-date smartphones, we believe it will be adopted as a standard in the future. In this project, we intend to explore these newly available measurements in order to better characterize the diversified urban environments. Essentially, we first need to gather a large amount of such raw GPS measurements, from various environments in our city, ranging from indoor to semi-indoor/outdoor, and further to pure outdoor. Then we tackle the challenges introduced by the complex GNSS data by applying a deep learning model to identify representations for respective location contexts. If time allows and if we get a big team on this project, we could also extend the objective to enabling the integration of a pervasive localization service In this project, you are supposed to use Android smartphones or other smartwears (e.g., Smartwatch) to collect raw GPS readings, and then we apply standard deep data analytics upon these data. So you are required to understand Android, and some basic knowledge of machine learning/data analytics would be a plus. For Part III of this project, we mostly focus on semi-outdoor and pure outdoor data.