On the efficient design of scalable indoor positioning systems based on Wi-Fi fingerprinting

This thesis investigates the design and implementation of Wi-Fi fingerprinting based Indoor Positioning Systems (IPS), with a focus on enhancing their efficiency, scalability, and accuracy. Wi-Fi fingerprinting, particularly utilising Received Signal Strength Indicator (RSSI) data, offers a cost-effective and non-intrusive method for indoor positioning. Despite its advantages, existing systems encounter challenges such as high computational complexity, the need for frequent manual updates, and difficulties in managing large datasets. The research commences by evaluating various position estimation algorithms, including k-Nearest Neighbour (k-NN) and its weighted variant (Wk-NN), identifying the correlation distance function as a highly effective approach when combined with exponential data representation. This combination was found to balance accuracy with computational simplicity, making it a viable option for efficient IPS. To address scalability and reliability, the thesis introduces a cloud-based Indoor Positioning System (CB-IPS) framework that leverages cloud computing, edge computing, and cache technologies. This framework significantly enhances the management of large fingerprint databases, optimises computational resources, and supports real-time processing, thereby improving the overall performance of the IPS. Furthermore, the research addresses the complexity of database management by implementing data preprocessing techniques, dimensionality reduction through Principal Component Analysis (PCA), and auto-update mechanisms. These strategies effectively reduce computational load and storage requirements, thereby ensuring that the system remains scalable and efficient. The findings demonstrate that the proposed optimisations can substantially enhance the performance of Wi-Fi fingerprinting-based IPS, making them more competitive with state-of-the-art systems. The research contributes to the advancement of indoor positioning technologies, offering practical solutions that address current limitations while laying the foundation for future innovations. This thesis concludes by outlining potential directions for future research, including the integration of advanced machine learning techniques, and further optimisation of real-time implementations. These efforts are essential for fully realising the potential of Wi-Fi fingerprinting-based indoor positioning systems across various real-world applications.