Secure Estimation for Unmanned Aerial Vehicles against Adversarial Cyber Attacks

摘要

In the coming years, usage of UnmannedAerial Vehicles (UAVs) is expected to grow tremendously.Maintaining security of UAVs under cyber attacks is animportant yet challenging task, as these attacks are oftenerratic and difficult to predict. Secure estimation problemsstudy how to estimate the states of a dynamical systemfrom a set of noisy and maliciously corrupted sensormeasurements. The fewer assumptions that an estimatormakes about the attacker, the larger the set of attacks itcan protect the system against. In this paper, we focus onsensor attacks on UAVs and attempt to design a secureestimator for linear time-invariant systems based on asfew assumptions about the attackers as possible. We proposea computationally efficient estimator that protects thesystem against arbitrary and unbounded attacks, wherethe set of attacked sensors can also change over time.In addition, we propose to combine our secure estimatorwith a Kalman Filter for improved practical performanceand demonstrate its effectiveness through simulations oftwo scenarios where an UAV is under adversarial cyberattack.