Date of Award:

5-2020

Document Type:

Dissertation

Degree Name:

Doctor of Philosophy (PhD)

Department:

Electrical and Computer Engineering

Committee

Ryan Gerdes

Committee

Rajnikant Sharma

Committee

Jake Gunther

Abstract

The work in this dissertation focuses on two examples of Cyber-Physical Systems (CPS), integrations of communication and monitoring capabilities to control a physical system, that operate in adversarial environments. That is to say, it is possible for individuals with malicious intent to gain access to various components of the CPS, disrupt normal operation, and induce harmful impacts. Such a deliberate action will be referred to as an attack. Therefore, some possible attacks against two CPSs will be studied in this dissertation and, when possible, solutions to handle such attacks will also be suggested.

The first CPS of interest is vehicular platoons wherein it is possible for a number of partially-automated vehicles to drive autonomously towards a certain destination with as little human driver involvement as possible. Such technology will ultimately allow passengers to focus on other tasks, such as reading or watching a movie, rather than on driving. In this dissertation three possible attacks against such platoons are studied. The first is called ”the disbanding attack” wherein the attacker is capable of disrupting one platoon and also inducing collisions in another intact (non-attacked) platoon vehicles. To handle such an attack, two solutions are suggested: The first solution is formulated using Model Predictive Control (MPC) optimal technique, while the other uses a heuristic approach. The second attack is False-Data Injection (FDI) against the platooning vehicular sensors is analyzed using the reachability analysis. This analysis allows us to validate whether or not it is possible for FDI attacks to drive a platoon towards accidents. Finally, mitigation strategies are suggested to prevent an attacker-controlled vehicle, one which operates inside a platoon and drives unpredictably, from causing collisions. These strategies are based on sliding mode control technique and once engaged in the intact vehicles, collisions are reduced and eventual control of those vehicles will be switched from auto to human to further reduce the impacts of the attacker-controlled vehicle.

The second CPS of interest in this dissertation is Heating, Ventilating, and Air Conditioning (HVAC) systems used in smart automated buildings to provide an acceptable indoor environment in terms of thermal comfort and air quality for the occupants For these systems, an MPC technique based controller is formulated in order to track a desired temperature in each zone of the building. Some previous studies indicate the possibility of an attacker to manipulate the measurements of temperature sensors, which are installed at different sections of the building, and thereby cause them to read below or above the real measured temperature. Given enough time, an attacker could monitor the system, understand how it works, and decide which sensor(s) to target. Eventually, the attacker may be able to deceive the controller, which uses the targeted sensor(s) readings and raises the temperature of one or multiple zones to undesirable levels, thereby causing discomfort for occupants in the building. In order to counter such attacks, Moving Target Defense (MTD) technique is utilized in order to constantly change the sensors sets used by the MPC controllers and, as a consequence, reduce the impacts of sensor attacks.

Checksum

4cf0e9c13876c5f9385fd1c96f959402

Share

COinS