abstract: The cyber-physical systems (CPS) are emerging as the underpinning technology for major industries in the 21-th century. This dissertation is focused on two fundamental issues in cyber-physical systems: network interdependence and information dynamics. It consists of the following two main thrusts. The first thrust is targeted at understanding the impact of network interdependence. It is shown that a cyber-physical system built upon multiple interdependent networks are more vulnerable to attacks since node failures in one network may result in failures in the other network, causing a cascade of failures that would potentially lead to the collapse of the entire infrastructure. There is thus a need to develop a new network science for modeling and quantifying cascading failures in multiple interdependent networks, and to develop network management algorithms that improve network robustness and ensure overall network reliability against cascading failures. To enhance the system robustness, a "regular" allocation strategy is proposed that yields better resistance against cascading failures compared to all possible existing strategies. Furthermore, in view of the load redistribution feature in many physical infrastructure networks, e.g., power grids, a CPS model is developed where the threshold model and the giant connected component model are used to capture the node failures in the physical infrastructure network and the cyber network, respectively. The second thrust is centered around the information dynamics in the CPS. One speculation is that the interconnections over multiple networks can facilitate information diffusion since information propagation in one network can trigger further spread in the other network. With this insight, a theoretical framework is developed to analyze information epidemic across multiple interconnecting networks. It is shown that the conjoining among networks can dramatically speed up message diffusion. Along a different avenue, many cyber-physical systems rely on wireless networks which offer platforms for information exchanges. To optimize the QoS of wireless networks, there is a need to develop a high-throughput and low-complexity scheduling algorithm to control link dynamics. To that end, distributed link scheduling algorithms are explored for multi-hop MIMO networks and two CSMA algorithms under the continuous-time model and the discrete-time model are devised, respectively. / Dissertation/Thesis / Ph.D. Electrical Engineering 2012

State estimation problem for power systems has long been a fundamental issue that demands a variety of methodologies dependent on the system settings. With recent introduction of advanced devices of phasor measurement units (PMUs) and dedicated communication networks, the infrastructure of power grids has been greatly improved. Coupled with the infrastructure improvements are three emerging issues for the state estimation problems, namely, the coexistence of both traditional and PMU measurements, the incomplete information resulting from delayed, missing and quantized measurements due to communication constraints, and the cyber-attacks on the communication channels. Three challenging problems are faced when dealing with the three issues in the state estimation program of power grids: 1) how to include the PMU measurements in the state estimator design, 2) how to account for the phenomena of incomplete information occurring in the measurements and design effective state estimators resilient to such phenomena, and 3) how to identify the system vulnerability in state estimation scheme and protect the estimation system against cyber-attacks. In this thesis, with the aim to solve the above problems, we develop several state estimation algorithms which tackle the issues of mixed measurements and incomplete information, and examine the cyber-security of the dynamic state estimation scheme. • To improve the estimation performance of power grids including PMU measurements, a hybrid extended Kalman filter and particle swarm optimization algorithm is developed, which has the advantages of being scalable to the numbers of the installed PMUs and being compatible with existing dynamic state estimation software as well. • Two kinds of network-induced phenomena, which leads to incomplete information of measurements, are considered. Specifically, the phenomenon of missing measurements is assumed to occur randomly and the missing probability is governed by a random variable, and the quantized nonlinear measurement model of power systems is presented where the quantization is assumed to be of logarithmic type. Then, the impact of the incomplete information on the overall estimation performance is taken into account when designing the estimator. Specifically, a modified extended Kalman filter is developed which is insensitive to the missing measurements in terms of acceptable probability, and a recursive filter is designed for the system with quantized measurements such that an upper bound of the estimation error is guaranteed and also minimized by appropriately designing the filter gain. • With the aim to reduce or eliminate the occurrence of the above-mentioned network-induced phenomena, we propose an event-based state estimation scheme with which communication transmission from the meters to the control centre can be greatly reduced. To ensure the estimation performance, we design the estimator gains by solving constrained optimization problems such that the estimation error covariances are guaranteed to be always less than a finite upper bound. • We examine the cyber-security of the dynamic state estimation system in power grids where the adversary is able to inject false data into the communication channels between PMUs and the control centre. The condition under which the attacks cause unbounded estimation errors is found. Furthermore, for system that is vulnerable to cyber-attacks, we propose a system protection scheme through which only a few (rather than all) communication channels require protection against false data injection attacks.

<div>Cyber-Physical Systems (CPSs) consist of physical and computational components usually interconnected through the internet. This type of systems have found applications in robotic surgery, smart medical services, driverless cars, smart power grids as well as in modern homes and offices. For a CPS to function properly, a reliable and secure communications between the system physical and cyber elements is of utmost importance. Malicious attacks during control signals and output measurements transmission between the physical plant and the control center must be addressed, which is the main research problem studied in this thesis.</div><div><br></div><div>A novel robust observer was proposed to synthesize a combined controller-observer compensator for a class of CPSs with sparse malicious attacks and arbitrary disturbances. The compensator consists of a controller, a norm approximator, and an unknown input observer (UIO). The proposed observer was compared with a norm-based observer given in the literature to show its advantage. To further enhance the proposed observer's performance against arbitrary disturbances, design methods were given that use fictitious output measurements and error correcting code (ECC) approach. The design of the UIO was extended to a bank of UIOs in order to improve the observer's performance against sparse malicious attacks.</div><div><br></div><div>The proposed observer can be used in the design of UIO-based fault detection and isolation (FDI) algorithms as well as in the distributed fault-tolerant control of large-scale interconnected systems. The results of this thesis can be applied to the design of controller-observer compensators for CPSs with modeling uncertainties.</div>

Cyber-physical systems (CPS) need to be designed to deal with various forms of uncertainty associated with data contributing to the system's knowledge of the environment. Dealing with uncertainty requires adopting an appropriate model, which then allows making the right decisions and carrying out the right actions (possibly affecting the environment) based on imperfect information. However, choosing and incorporating a suitable model into CPS design is difficult, because it requires identifying the kind of uncertainty at hand as well as knowledge of suitable models and their application to dealing with the uncertainty. While inspiration can be found in other CPS designs, the details of dealing with uncertainty in another CPS can be confounded by domain-specific terminology, context, and requirements. To make this aspect of CPS design less daunting, we aim at providing an overview of approaches dealing with uncertainty in the design of CPS targeting collective behavior. To this end, we present a systematic review of relevant scientific projects with industrial leadership and synthesis of relations between system features, the kinds of uncertainty, and methods used to deal with it. The results provide an overview of uncertainty across different domains and challenges and reason about a guide for...

<div>Cyber-physical systems, through various sensors and actuators, are used to handle interactions of the cyber-world with the physical-world. Conventionally, the temporal component of the physical-world has been used only for estimating real-time deadlines and responsiveness of control-loop algorithms. However, there are various other applications where the relationship of the temporal component and the cyber-world are of interest. An example is the ability to reconstruct a sequence of past temporal activities from the current state of the cyber-world, which is of obvious interest to cyber-forensic investigators. Another example is the ability to control the temporal components in broadcast communication networks, which leads to new attack and defense capabilities. These relationships have not been explored traditionally.</div><div><br></div><div>To address this gap, this dissertation proposes three systems that cast light on the effect of temporal component of the physical-world on the cyber-world. First, we present Timeliner, a smartphone cyber-forensics technique that recovers past actions from a single static memory image. Following that, we present work on CAN (Controller Area Network), a broadcast communication network used in automotive applications. We show in DUET that the ability to control communication temporally allows two compromised ECUs, an attacker and an accomplice, to stealthily suppress and impersonate a victim ECU, even in the presence of a voltage-based intrusion detection system. In CANDID, we show that the ability to temporally control CAN communication opens up new defensive capabilities that make the CAN much more secure.</div><div><br></div><div>The evaluation results show that Timeliner is very accurate and can reveal past evidence (up to an hour) of user actions across various applications on Android devices. The results also show that DUET is highly effective at impersonating victim ECUs while evading both message-based and voltage-based intrusion detection systems, irrespective of the features and the training algorithms used. Finally, CANDID is able to provide new defensive capabilities to CAN environments with reasonable communication and computational overheads.</div><div><br></div>

Cyber-physical systems, built from the integration of cyber and physical components, are being used in multiple domains ranging from manufacturing and healthcare to traffic con- trol and safety. Ensuring the security of cyber-physical systems is crucial because they provide the foundation of the critical infrastructure, and security incidents can result in catastrophic failures. Recent publications report that machine learning models are vul- nerable to adversarial examples, crafted by adding small perturbations to input data. For the past decade, machine learning security has become a growing interest area, with a significant number of systematic reviews and surveys that have been published. Secu- rity of artificial intelligence in cyber-physical systems is more challenging in comparison to machine learning security, because adversaries have a wider possible attack surface, in both cyber and physical domains. However, comprehensive systematic literature re- views in this research field are not available. Therefore, this work presents a systematic literature review of the adversarial attacks on artificial intelligence in cyber-physical sys- tems, examining 45 scientific papers, selected from 134 publications found in the Scopus database. It provides the classification of attack algorithms and defense methods, the sur- vey of evaluation metrics, an overview of the state of the art in methodologies and tools, and, as the main contribution, identifies open problems and research gaps and highlights future research challenges in this area of interest.

Complex cyber-physical systems have become fundamental to modern society by effectively providing critical services and improving efficiency in various domains. Unfortunately, as systems become more connected and more complex, they also can become more vulnerable and less robust. As a result, various failure modes become more common and easily triggered from both unanticipated and malicious perturbations. Research has been conducted in the area of vulnerability analysis for cyber-physical systems, to assist in locating these possible vulnerabilities before they can fail. I present two case studies on different forms of critical infrastructure systems to identify vulnerabilities and understand how external perturbations can affect them, namely UAV drone swarms and municipal water infrastructure.

This thesis explores how a detection engine using Artificial Neural Networks (ANNs) could be implemented within the DETECT framework. The framework is used for security purposes in Cyber-physical systems. These are critical systems often vital to important infrastructure so discovering new ways of how to defend against threats is of huge importance. However, there are many difficult challenges that needs to be addressed before employing an ANN as a threat detection mechanism. Most notable what kind of ANN to use, training data and issues such as over-fitting. These challenges were addressed in the model that was created for this paper. The model was based on the current literature and previous research. To make informed decisions about the model a literature review was carried out to gather as much information as possible. A key insight from the review was the use of recurrent neural networks for threat detection.

Smart manufacturing promotes the demand of a new interface for communication with virtual entities such as big data analysis model, digital twin, and autonomous software programs. Although ideal smart manufacturing pursues full automation by self-adoption and self-decision of autonomy, converging human intervention and collaboration with the autonomy have shown significant improvements on productivity and quality, and it expects more advancements on current manufacturing trend. This study aims how to combine human and autonomy based on current technical advancement of smart manufacturing. In detail, creating networks between the entities, developing a new interface for human-autonomy collaboration, and demonstrating the effectiveness of the collaboration are main research topics.<br>

Cyber-Physical System (CPS) becomes more and more importance from industrial application (e.g., aircraft control, automation management) to societal challenges (e.g. health caring, environment monitoring). It has traditionally been designed to one specific application domain and to be managed by a single entity, implemented communication between physical world and computational world. However, it still just work within its domain, and not be interoperability. How to make it into scalable? How to make it reusing? These questions become more and more necessary. In this paper, we are trying to developing a common CPS infrastructure, let it be an innovative CPS crossing multiple domains to broad use sensors and actuators. Here, we implement a technique for automatically build a model according to the sensor data in different domains. And based on our approach under continuous situation, it could identify the sensor values right now or estimate next few time step, which we call spatial model or temporal model.