ETD

• Bluetooth Anchor Nodes
• Data Integrity and Security
• Data Poisoning Attacks
• RSSI-based Indoor Positioning Systems (IPS)
• Targeted Trajectory Manipulation

Indoor localization systems are increasingly used for navigation and for tracking people and objects in enclosed environments. Among the most widely adopted solutions, RSSI-based approaches (Received Signal Strength Indicator) are attractive because they do not require additional hardware, but they often overlook crucial security and data-integrity aspects. This thesis investigates the vulnerability of an RSSI-based indoor positioning system to data poisoning attacks, in which an adversary alters signal measurements to influence position estimates. The goal is not only to degrade accuracy, but also to induce the system to reconstruct plausible yet incorrect trajectories, steering the estimated path toward an attacker-desired route. Using a real-world dataset, a system is implemented to estimate the trajectory of a mobile target, assessing the impact of simple manipulations and a targeted attack in which the adversary controls a subset of Bluetooth anchor nodes. In addition, neural networks are explored to automate the targeted attack and compare their effectiveness with more direct approaches. Overall, the results show that an attacker can indeed obtain significant and controlled deviations in the estimated trajectory, confirming the need to integrate data-integrity protection mechanisms into indoor localization systems.