Examination Committee

Prof Gary CHAN, CSE/HKUST (Chairperson)
Prof Ling SHI, ECE/HKUST (Thesis Supervisor)
Prof Daniel HO, Department of Mathematics, City University of Hong Kong (External Examiner)
Prof Wai Ho MOW, ECE/HKUST
Prof Ming LIU, ECE/HKUST
Prof Francesco CIUCCI, MAE/HKUST
 
 

Abstract

Cyber physical systems are open to all sorts of attacks due to its reliance on wireless communication. Attackers are searching for optimal attacks that damage the system's performance, and in the presence of a detection mechanism, would want the optimal attack that is stealthy. In this thesis, the concept of epsilon-stealthiness is studied in its entirety. We construct optimal epsilon-stealthy attacks with respect to estimation quality for the control attack and tackle the measurement attack case from two different perspectives.
 
Our secondary objective is to propose a method to counter this attack. We will develop a model under which the estimator can increase the probability of detection. In this model, two ideas are proposed to optimize the threshold distribution and the optimal distribution is constructed accordingly.