Trust aware particle filters for autonomous vehicles

Cyber-Physical Systems have been widely employed in safety critical applications including intelligent highways, autonomous vehicles and robotic systems. State estimation is crucial for Cyber-Physical Systems because control commands that are sent to physical systems depend on the estimated states. The particle filter is a good candidate for state estimation due to its applicability to nonlinear and/or non-Gaussian dynamic systems. However, classical particle filters are not robust against false data injection from sensors compromised by attackers. In this paper, we propose a novel particle filter algorithm, trust aware particle filter, that is robust to false data injection attacks. We develop a framework in which a state estimator assigns trust values to sensors based on the measurements and we utilize the trust values in the state estimation. Simulation results demonstrate the robustness of the trust aware particle filter in the presence of false data injection attacks.