Secure Estimation for Attitude and Heading Reference Systems Under Sparse Attacks

This paper focuses on the problem of secure attitude estimation for autonomous vehicles. Based on the established AHRS measuring model and the attack model, we have decomposed the optimal Kalman estimate into a linear combination of local state estimates. We then propose a convex optimization-based approach, instead of the weighted sum approach, to combine the local estimate into a more secure estimate. It is shown that the proposed secure estimator coincides with the Kalman estimator with certain probability when there is no attack, and can be stable when
$p$
elements of the model state are compromised. Simulations have been conducted to validate the proposed secure filter under single and multiple measurement attacks.