Legacy cyber-physical systems (CPSs) were designed without considering cybersecurity as a primary design tenet especially when considering their evolving operating environment. There are many examples of legacy systems including automotive control, navigation, transportation, and industrial control systems (ICSs), to name a few. To make matters worse, the cost of designing and deploying defenses in existing legacy infrastructure can be overwhelming as millions or even billions of legacy CPS systems are already in use. This economic angle, prevents the use of defenses that are not backward compatible. Moreover, any protection has to operate efficiently in resource constraint environments that are dynamic nature. Hence, the existing approaches that require ex- pensive additional hardware, propose a new protocol from scratch, or rely on complex numerical operations such as strong cryptographic solutions, are less likely to be deployed in practice. In this dissertation, we explore a variety of lightweight solutions for securing different existing CPSs without requiring any modifications to the original system design at hardware or protocol level. In particular, we use fingerprinting, crowdsourcing and deterministic models as alternative backwards- compatible defenses for securing vehicles, global positioning system (GPS) receivers, and a class of ICSs called supervisory control and data acquisition (SCADA) systems, respectively. We use fingerprinting to address the deficiencies in automobile cyber-security from the angle of controller area network (CAN) security. CAN protocol is the de-facto bus standard commonly used in the automotive industry for connecting electronic control units (ECUs) within a vehicle. The broadcast nature of this protocol, along with the lack of authentication or integrity guarantees, create a foothold for adversaries to perform arbitrary data injection or modification and impersonation attacks on the ECUs. We propose SIMPLE, a single-frame based physical layer identification for intrusion detection and prevention on such networks. Physical layer identification or fingerprinting is a method that takes advantage of the manufacturing inconsistencies in the hardware components that generate the analog signal for the CPS of our interest. It translates the manifestation of these inconsistencies, which appear in the analog signals, into unique features called fingerprints which can be used later on for authentication purposes. Our solution is resilient to ambient temperature, supply voltage value variations, or aging. Next, we use fingerprinting and crowdsourcing at two separate protection approaches leveraging two different perspectives for securing GPS receivers against spoofing attacks. GPS, is the most predominant non-authenticated navigation system. The security issues inherent into civilian GPS are exacerbated by the fact that its design and implementation are public knowledge. To address this problem, first we introduce Spotr, a GPS spoofing detection via device fingerprinting, that is able to determine the authenticity of signals based on their physical-layer similarity to the signals that are known to have originated from GPS satellites. More specifically, we are able to detect spoofing activities and track genuine signals over different times and locations and propagation effects related to environmental conditions. In a different approach at a higher level, we put forth Crowdsourcing GPS, a total solution for GPS spoofing detection, recovery and attacker localization. Crowdsourcing is a method where multiple entities share their observations of the environment and get together as a whole to make a more accurate or reliable decision on the status of the system. Crowdsourcing has the advantage of deployment with the less complexity and distributed cost, however its functionality is dependent on the adoption rate by the users. Here, we have two methods for implementing Crowdsourcing GPS. In the first method, the users in the crowd are aware of their approximate distance from other users using Bluetooth. They cross validate this approximate distance with the GPS-derived distance and in case of any discrepancy they report ongoing spoofing activities. This method is a strong candidate when the users in the crowd have a sparse distribution. It is also very effective when tackling multiple coordinated adversaries. For method II, we exploit the angular dispersion of the users with respect to the direction that the adversarial signal is being transmitted from. As a result, the users that are not facing the attacker will be safe. The reason for this is that human body mostly comprises of water and absorbs the weak adversarial GPS signal. The safe users will help the spoofed users find out that there is an ongoing attack and recover from it. Additionally, the angular information is used for localizing the adversary. This method is slightly more complex, and shows the best performance in dense areas. It is also designed based on the assumption that the spoofing attack is only terrestrial. Finally, we propose a tandem IDS to secure SCADA systems. SCADA systems play a critical role in most safety-critical infrastructures of ICSs. The evolution of communications technology has rendered modern SCADA systems and their connecting actuators and sensors vulnerable to malicious attacks on both physical and application layers. The conventional IDS that are built for securing SCADA systems are focused on a single layer of the system. With the tandem IDS we break this habit and propose a strong multi-layer solution which is able to expose a wide range of attack. To be more specific, the tandem IDS comprises of two parts, a traditional network IDS and a shadow replica. We design the shadow replica as a deterministic IDS. It performs a workflow analysis and makes sure the logical flow of the events in the SCADA controller and its connected devices maintain their expected states. Any deviation would be a malicious activity or a reliability issue. To model the application level events, we leverage finite state machines (FSMs) to compute the anticipated states of all of the devices. This is feasible because in many of the existing ICSs the flow of traffic and the resulting states and actions in the connected devices have a deterministic nature. Consequently, it leads to a reliable and free of uncertainty solution. Aside from detecting traditional network attacks, our approach bypasses the attacker in case it succeeds in taking over the devices and also maintains continuous service if the SCADA controller gets compromised. / Doctor of Philosophy / Our lives are entangled with cyber-physical systems (CPSs) on a daily basis. Examples of these systems are vehicles, navigation systems, transportation systems, industrial control systems, etc. CPSs are mostly legacy systems and were built with a focus on performance, overlooking security. Security was not considered in the design of these old systems and now they are dominantly used in our everyday life. After numerous demonstration of cyber hacks, the necessity of protecting the CPSs from adversarial activities is no longer ambiguous. Many of the advanced cryptographic techniques are far too complex to be implemented in the existing CPSs such as cars, satellites, etc. We attempt to secure such resource constraint systems using simple backward compatible techniques in this dissertation. We design cheap lightweight solutions, with no modifications to the original system. In part of our research, we use fingerprinting as a technique to secure passenger cars from being hacked, and GPS receivers from being spoofed. For a brief description of fingerprinting, we use the example of two identical T-shirts with the same size and design. They will always have subtle differences between them no matter how hard the tailor tried to make them identical. This means that there are no two T-shirts that are exactly identical. This idea, when applied to analog signalling on electric devices, is called fingerprinting. Here, we fingerprint the mini computers inside a car, which enables us to identify these computers and prevent hacking. We also use the signal levels to design fingerprints for GPS signals. We use the fingerprints to distinguish counterfeit GPS signals from the ones that have originated from genuine satellites. This summarizes two major contributions in the dissertation. Our earlier contribution to GPS security was effective, but it was heavily dependent on the underlying hardware, requiring extensive training for each radio receiver that it was protecting. To remove this dependence of training for the specific underlying hardware, we design and implement the next framework using defenses that require application-layer access. Thus, we proposed two methods that leverage crowdsourcing approaches to defend against GPS spoofing attacks and, at the same time, improve the accuracy of localization for commodity mobile devices. Crowdsourcing is a method were several devices agree to share their information with each other. In this work, GPS users share their location and direction information, and in case of any discrepancy they figure that they are under attack and cooperate to recover from it. Last, we shift the gear to the industrial control systems (ICSs) and propose a novel IDS to protect them against various cyber attacks. Unlike the conventional IDSs that are focused on one of the layers of the system, our IDS comprises of two main components. A conventional component that exposes traditional attacks and a second component called a shadow replica. The replica mimics the behavior of the system and compares it with that of the actual system in a real-time manner. In case of any deviation between the two, it detects attacks that target the logical flow of the events in the system. Note that such attacks are more sophisticated and difficult to detect because they do not leave any obvious footprints behind. Upon detection of attacks on the original controller, our replica takes over the responsibilities of the original ICS controller and provides service continuity.

No description available.

This thesis proposes a new method for detecting malicious cyber-physical attacks on additive manufacturing (AM) systems. The method makes use of a physical hash, which links digital data to the manufactured part via a disconnected side-channel measurement system. The disconnection ensures that if the network and/or AM system become compromised, the manufacturer can still rely on the measurement system for attack detection. The physical hash takes the form of a QR code that contains a hash string of the nominal process parameters and toolpath. It is manufactured alongside the original geometry for the measurement system to scan and compare to the readings from its sensor suite. By taking measurements in situ, the measurement system can detect in real-time if the part being manufactured matches the designer's specification. A proof-of-concept validation was realized on a material extrusion machine. The implementation was successful and demonstrated the ability of this method to detect the existence (and absence) of malicious attacks on both process parameters and the toolpath. A case study for detecting changes to the toolpath is also presented, which uses a simple measurement of how long each layer takes to build. Given benchmark readings from a 30x30 mm square layer created on a material extrusion system, several modifications were able to be detected. The machine's repeatability and measurement technique's accuracy resulted in the detection of a 1 mm internal void, a 2 mm scaling attack, and a 1 mm skewing attack. Additionally, for a short to moderate length build of an impeller model, it was possible to detect a 0.25 mm change in the fin base thickness. A second case study is also presented wherein dogbone tensile test coupons were manufactured on a material extrusion system at different extrusion temperatures. This process parameter is an example of a setting that can be maliciously modified and have an effect on the final part strength without the operator's knowledge. The performance characteristics (Young's modulus and maximum stress) were determined to be statistically different at different extrusion temperatures (235 and 270 °C). / Master of Science / Additive Manufacturing (AM, also known as 3D printing) machines are cyber-physical systems and are therefore vulnerable to malicious attacks that can cause physical damage to the parts being manufactured or even to the machine itself. This thesis proposes a new method for detecting that an AM system has been hacked. Attacks are identified via a series of measurements taken by a measurement system that is disconnected from the main network. The disconnection ensures that if the network and/or AM system are hacked, the manufacturer can still rely on the measurement system for attack detection. The proposed method uses a physical hash to transfer information to the disconnected measurement system. This physical hash takes the form of a QR code and stores in it the nominal process parameters and toolpath of the build. It is manufactured alongside the original geometry for the measurement system to scan and compare to the readings from its sensor suite. By taking measurements in real-time, the measurement system can detect if the part being manufactured matches the designer’s specification. A proof-of-concept of the proposed method was realized on a common AM system. The implementation was successful and demonstrated the ability of this method to detect the existence of a malicious attack. A case study for detecting changes to the toolpath is also proposed using the simple measurement of how long each layer takes to build. Given benchmark readings of a part manufactured on the same technology as the proof-of-concept implementation, several modifications were able to be detected. The attack types tested were the insertion of an internal void, scaling the part, and skewing the part. A second case study is also presented where components were manufactured at different extrusion temperatures. By measuring the force required to break the parts, it was determined that temperature has an effect on the final part strength. This confirmed that malicious attacks targeting extrusion temperature are a plausible threat, and that the parameter should be measured in the proposed system.

A cyber-physical controller (CPC) uses computing to control a physical process. Example CPCs can be found in self-driving automobiles, unmanned aerial vehicles, and other autonomous systems. They are also used in large-scale industrial control systems (ICSs) manufacturing and utility infrastructure. CPC operations rely on embedded systems having real-time, high-assurance interactions with physical processes. However, recent attacks like Stuxnet have demonstrated that CPC malware is not restricted to networks and general-purpose computers, rather embedded components are targeted as well. General-purpose computing and network approaches to security are failing to protect embedded controllers, which can have the direct effect of process disturbance or destruction. Moreover, as embedded systems increasingly grow in capability and find application in CPCs, embedded leaf node security is gaining priority. This work develops a root-of-trust design architecture, which provides process resilience to cyber attacks on, or from, embedded controllers: the Trustworthy Autonomic Interface Guardian Architecture (TAIGA). We define five trust requirements for building a fine-grained trusted computing component. TAIGA satisfies all requirements and addresses all classes of CPC attacks using an approach distinguished by adding resilience to the embedded controller, rather than seeking to prevent attacks from ever reaching the controller. TAIGA provides an on-chip, digital, security version of classic mechanical interlocks. This last line of defense monitors all of the communications of a controller using configurable or external hardware that is inaccessible to the controller processor. The interface controller is synthesized from C code, formally analyzed, and permits run-time checked, authenticated updates to certain system parameters but not code. TAIGA overrides any controller actions that are inconsistent with system specifications, including prediction and preemption of latent malwares attempts to disrupt system stability and safety. This material is based upon work supported by the National Science Foundation under Grant Number CNS-1222656. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation. We are grateful for donations from Xilinx, Inc. and support from the Georgia Tech Research Institute. / Ph. D.

Autonomous cyber-physical systems (CPSs) such as autonomous connected vehicles (ACVs), unmanned aerial vehicles (UAVs), critical infrastructure (CI), and the Internet of Things (IoT) will be essential to the functioning of our modern economies and societies. Therefore, maintaining the autonomy of CPSs as well as their stability, robustness, and security (SRS) in face of exogenous and disruptive events is a critical challenge. In particular, it is crucial for CPSs to be able to not only operate optimally in the vicinity of a normal state but to also be robust and secure so as to withstand potential failures, malfunctions, and intentional attacks. However, to evaluate and improve the SRS of CPSs one must overcome many technical challenges such as the unpredictable behavior of a CPS's cyber-physical environment, the vulnerability to various disruptive events, and the interdependency between CPSs. The primary goal of this dissertation is, thus, to develop novel foundational analytical tools, that weave together notions from machine learning, game theory, and control theory, in order to study, analyze, and optimize SRS of autonomous CPSs. Towards achieving this overarching goal, this dissertation led to several major contributions. First, a comprehensive control and learning framework was proposed to thwart cyber and physical attacks on ACV networks. This framework brings together new ideas from optimal control and reinforcement learning (RL) to derive a new optimal safe controller for ACVs in order to maximize the street traffic flow while minimizing the risk of accidents. Simulation results show that the proposed optimal safe controller outperforms the current state of the art controllers by maximizing the robustness of ACVs to physical attacks. Furthermore, using techniques from convex optimization and deep RL a joint trajectory and scheduling policy is proposed in UAV-assisted networks that aims at maintaining the freshness of ground node data at the UAV. The analytical and simulation results show that the proposed policy can outperform policies such discretized state RL and value-based methods in terms of maximizing the freshness of data. Second, in the IoT domain, a novel watermarking algorithm, based on long short term memory cells, is proposed for dynamic authentication of IoT signals. The proposed watermarking algorithm is coupled with a game-theoretic framework so as to enable efficient authentication in massive IoT systems. Simulation results show that using our approach, IoT messages can be transmitted from IoT devices with an almost 100% reliability. Next, a brainstorming generative adversarial network (BGAN) framework is proposed. It is shown that this framework can learn to generate real-looking data in a distributed fashion while preserving the privacy of agents (e.g. IoT devices, ACVs, etc). The analytical and simulation results show that the proposed BGAN architecture allows heterogeneous neural network designs for agents, works without reliance on a central controller, and has a lower communication over head compared to other state-of-the-art distributed architectures. Last, but not least, the SRS challenges of interdependent CI (ICI) are addressed. Novel game-theoretic frameworks are proposed that allow the ICI administrator to assign different protection levels on ICI components to maximizing the expected ICI security. The mixed-strategy Nash of the games are derived analytically. Simulation results coupled with theoretical analysis show that, using the proposed games, the administrator can maximize the security level in ICI components. In summary, this dissertation provided major contributions across the areas of CPSs, machine learning, game theory, and control theory with the goal of ensuring SRS across various domains such as autonomous vehicle networks, IoT systems, and ICIs. The proposed approaches provide the necessary fundamentals that can lay the foundations of SRS in CPSs and pave the way toward the practical deployment of autonomous CPSs and applications. / Doctor of Philosophy / In order to deliver innovative technological services to their residents, smart cities will rely on autonomous cyber-physical systems (CPSs) such as cars, drones, sensors, power grids, and other networks of digital devices. Maintaining stability, robustness, and security (SRS) of those smart city CPSs is essential for the functioning of our modern economies and societies. SRS can be defined as the ability of a CPS, such as an autonomous vehicular system, to operate without disruption in its quality of service. In order to guarantee SRS of CPSs one must overcome many technical challenges such as CPSs' vulnerability to various disruptive events such as natural disasters or cyber attacks, limited resources, scale, and interdependency. Such challenges must be considered for CPSs in order to design vehicles that are controlled autonomously and whose motion is robust against unpredictable events in their trajectory, to implement stable Internet of digital devices that work with a minimum communication delay, or to secure critical infrastructure to provide services such as electricity, gas, and water systems. The primary goal of this dissertation is, thus, to develop novel foundational analytical tools, that weave together notions from machine learning, game theory, and control theory, in order to study, analyze, and optimize SRS of autonomous CPSs which eventually will improve the quality of service provided by smart cities. To this end, various frameworks and effective algorithms are proposed in order to enhance the SRS of CPSs and pave the way toward the practical deployment of autonomous CPSs and applications. The results show that the developed solutions can enable a CPS to operate efficiently while maintaining its SRS. As such, the outcomes of this research can be used as a building block for the large deployment of smart city technologies that can be of immense benefit to tomorrow's societies.

Cyber-physical systems (CPSs) are large-scale systems that seamlessly integrate physical and human elements via a cyber layer that enables connectivity, sensing, and data processing. Key examples of CPSs include smart power systems, smart transportation systems, and the Internet of Things (IoT). This wide-scale cyber-physical interconnection introduces various operational benefits and promises to transform cities, infrastructure, and networked systems into more efficient, interactive, and interconnected smart systems. However, this ubiquitous connectivity leaves CPSs vulnerable to menacing security threats as evidenced by the recent discovery of the Stuxnet worm and the Mirai malware, as well as the latest reported security breaches in a number of CPS application domains such as the power grid and the IoT. Addressing these culminating security challenges requires a holistic analysis of CPS security which necessitates: 1) Determining the effects of possible attacks on a CPS and the effectiveness of any implemented defense mechanism, 2) Analyzing the multi-agent interactions -- among humans and automated systems -- that occur within CPSs and which have direct effects on the security state of the system, and 3) Recognizing the role that humans and their decision making processes play in the security of CPSs. Based on these three tenets, the central goal of this dissertation is to enhance the security of CPSs with human actors by developing fool-proof defense strategies founded on novel theoretical frameworks which integrate the engineering principles of CPSs with the mathematical concepts of game theory and human behavioral models. Towards realizing this overarching goal, this dissertation presents a number of key contributions targeting two prominent CPS application domains: the smart electric grid and drone systems. In smart grids, first, a novel analytical framework is developed which generalizes the analysis of a wide set of security attacks targeting the state estimator of the power grid, including observability and data injection attacks. This framework provides a unified basis for solving a broad set of known smart grid security problems. Indeed, the developed tools allow a precise characterization of optimal observability and data injection attack strategies which can target the grid as well as the derivation of optimal defense strategies to thwart these attacks. For instance, the results show that the proposed framework provides an effective and tractable approach for the identification of the sparsest stealthy attacks as well as the minimum sets of measurements to defend for protecting the system. Second, a novel game-theoretic framework is developed to derive optimal defense strategies to thwart stealthy data injection attacks on the smart grid, launched by multiple adversaries, while accounting for the limited resources of the adversaries and the system operator. The analytical results show the existence of a diminishing effect of aggregated multiple attacks which can be leveraged to successfully secure the system; a novel result which leads to more efficiently and effectively protecting the system. Third, a novel analytical framework is developed to enhance the resilience of the smart grid against blackout-inducing cyber attacks by leveraging distributed storage capacity to meet the grid's critical load during emergency events. In this respect, the results demonstrate that the potential subjectivity of storage units' owners plays a key role in shaping their energy storage and trading strategies. As such, financial incentives must be carefully designed, while accounting for this subjectivity, in order to provide effective incentives for storage owners to commit the needed portions of their storage capacity for possible emergency events. Next, the security of time-critical drone-based CPSs is studied. In this regard, a stochastic network interdiction game is developed which addresses pertinent security problems in two prominent time-critical drone systems: drone delivery and anti-drone systems. Using the developed network interdiction framework, the optimal path selection policies for evading attacks and minimizing mission completion times, as well as the optimal interdiction strategies for effectively intercepting the paths of the drones, are analytically characterized. Using advanced notions from Nobel-prize winning prospect theory, the developed framework characterizes the direct impacts of humans' bounded rationality on their chosen strategies and the achieved mission completion times. For instance, the results show that this bounded rationality can lead to mission completion times that significantly surpass the desired target times. Such deviations from the desired target times can lead to detrimental consequences primarily in drone delivery systems used for the carriage of emergency medical products. Finally, a generic security model for CPSs with human actors is proposed to study the diffusion of threats across the cyber and physical realms. This proposed framework can capture several application domains and allows a precise characterization of optimal defense strategies to protect the critical physical components of the system from threats emanating from the cyber layer. The developed framework accounts for the presence of attackers that can have varying skill levels. The results show that considering such differing skills leads to defense strategies which can better protect the system. In a nutshell, this dissertation presents new theoretical foundations for the security of large-scale CPSs, that tightly integrate cyber, physical, and human elements, thus paving the way towards the wide-scale adoption of CPSs in tomorrow's smart cities and critical infrastructure. / Ph. D. / Enhancing the efficiency, sustainability, and resilience of cities, infrastructure, and industrial systems is contingent on their transformation into more interactive and interconnected smart systems. This has led to the emergence of what is known as cyber-physical systems (CPSs). CPSs are widescale distributed and interconnected systems integrating physical components and humans via a cyber layer that enables sensing, connectivity, and data processing. Some of the most prominent examples of CPSs include the smart electric grid, smart cities, intelligent transportation systems, and the Internet of Things. The seamless interconnectivity between the various elements of a CPS introduces a wealth of operational benefits. However, this wide-scale interconnectivity and ubiquitous integration of cyber technologies render CPSs vulnerable to a range of security threats as manifested by recently reported security breaches in a number of CPS application domains. Addressing these culminating security challenges requires the development and implementation of fool-proof defense strategies grounded in solid theoretical foundations. To this end, the central goal of this dissertation is to enhance the security of CPSs by advancing novel analytical frameworks which tightly integrate the cyber, physical, and human elements of a CPS. The developed frameworks and tools enable the derivation of holistic defense strategies by: a) Characterizing the security interdependence between the various elements of a CPS, b) Quantifying the consequences of possible attacks on a CPS and the effectiveness of any implemented defense mechanism, c) Modeling the multi-agent interactions in CPSs, involving humans and automated systems, which have a direct effect on the security state of the system, and d) Capturing the role that human perceptions and decision making processes play in the security of CPSs. The developed tools and performed analyses integrate the engineering principles of CPSs with the mathematical concepts of game theory and human behavioral models and introduce key contributions to a number of CPS application domains such as the smart electric grid and drone systems. The introduced results enable strengthening the security of CPSs, thereby paving the way for their wide-scale adoption in smart cities and critical infrastructure.

The integration of Artificial Intelligence (AI) into water supply systems (WSSs) has revolutionized real-time monitoring, automated operational control, and predictive decision-making analytics. However, AI also introduces security vulnerabilities, such as data poisoning. In this context, data poisoning could involve the malicious manipulation of critical data, including water quality parameters, flow rates, and chemical composition levels. The consequences of such threats are significant, potentially jeopardizing public safety and health due to decisions being made based on poisoned data. This thesis aims to exploit these vulnerabilities in data-driven applications within WSSs. Proposing Water Generative Adversarial Networks, H2OGAN, a time-series GAN-based model designed to synthesize water data. H2OGAN produces water data based on the characteristics within the expected constraints of water data cardinality. This generative model serves multiple purposes, including data augmentation, anomaly detection, risk assessment, cost-effectiveness, predictive model optimization, and understanding complex patterns within water systems. Experiments are conducted in AI and Cyber for Water and Agriculture (ACWA) Lab, a cyber-physical water testbed that generates datasets replicating both operational and adversarial scenarios in WSSs. Identifying adversarial scenarios is particularly importance due to their potential to compromise water security. The datasets consist of 10 physical incidents, including normal conditions, sensor anomalies, and malicious attacks. A recurrent neural network (RNN) model, i.e., gated recurrent unit (GRU), is used to classify and capture the temporal dynamics those events. Subsequently, experiments with real-world data from Alexandria Renew Enterprises (AlexRenew), a wastewater treatment plant in Alexandria, Virginia, are conducted to assess the effectiveness of H2OGAN in real-world applications. / Master of Science / Today, a significant portion of the global population struggles with access to essential services: 25% lack clean water, 50% lack sanitation services, and 30% lack hygiene facilities. In response, AI is being leveraged to tackle these deficiencies within water supply systems. Investments in AI are expected to reach an estimated $6.3 billion by 2030, with potential savings of 20% to 30% in operational expenditures by optimizing chemical usage in water treatment. The flexibility and efficiency of AI applications have fueled optimism about their potential to revolutionize water management. As the era of Industry 4.0 progresses, the role of AI in transforming critical infrastructures, including water supply systems, becomes increasingly vital. However, this technological integration brings with it heightened vulnerabilities. The water sector, recognized as one of the 16 critical infrastructures by the Cybersecurity and Infrastructure Security Agency (CISA), has seen a notable increase in cyberattack incidents. These attacks underscore the urgent need for sophisticated AI-driven security solutions to protect these essential systems against potential compromises that could pose significant public health risks. Addressing these challenges, this thesis introduces H2OGAN, a time-series GAN-based model developed to generate and analyze realistic water data within the expected constraints of water parameter characteristics. H2OGAN supports various functions including data augmentation, anomaly detection, risk assessment, and predictive model optimization, thereby enhancing the security and efficiency of water supply systems. Extensive testing is conducted in ACWA Lab, a cyber-physical testbed that replicates both operational and adversarial scenarios. These experiments utilize a RNN model, specifically a GRU, to classify and analyze the dynamics of various scenarios including normal operations, sensor anomalies, and malicious attacks. Further real-world validation is carried out at AlexRenew, a wastewater treatment facility in Alexandria, Virginia, confirming the effectiveness of H2OGAN in practical applications. This research not only advances the understanding of AI in water management but also emphasizes the critical need for robust security measures to protect against the evolving landscape of cyber threats.

The layer-by-layer nature of additive manufacturing (AM) allows for toolless fabrication of highly complex geometries that cannot be made via traditional processes. AM is unique in its ability to precisely define both the material properties and geometric shape throughout the volume of a part, giving designers unmatched freedom in the creation of new components. However, this freedom of design also creates numerous challenges in the qualification of these parts. As AM processes primitive material in real time to produce each voxel of part volume, manufacturing defects may be dispersed anywhere throughout the part. Many part designs may have complex geometries or material parameters that are challenging for traditional qualification and inspection techniques to inspect for such embedded errors. Even more troubling, this freedom of design also extends to malicious actors, who would then be able to embed intentional targeted defects within the volume of the part. As the AM process is driven almost entirely by computer controlled machines and cyber-domain data, the AM process is uniquely at risk of nearly undetectable cyber-physical attacks, or cyber attacks that can cause physical damage. Additionally, as much of the valuable intellectual property associated with the design and material parameters of parts are stored in digital form, theft of these design files could result in mass replication of lower quality counterfeit parts, putting the supply chain of these AM parts at risk. In order to mitigate these vulnerabilities in the AM process, prior works have focused on in situ monitoring of the manufacturing process in order to ensure the part is constructed as expected. Typically for in situ monitoring, the constructed geometry is compared to the design files associated with the part in question using a monitoring system connected to either the AM machine or the larger network. However, such methods trust the validity of both the design files and monitoring systems used for verification, when either or both may have also been attacked. Therefore, a valid in situ monitoring method needs secure access to a provable set of validation data, while also isolating or air-gapping itself from the network to prevent cyber attacks on the monitoring system itself. Similarly, other works have focused on mitigating the risk of counterfeiting by novel means of part identification tailored for the AM process. Many of these identification methods leverage stochastic or prescribed features, such as surface patterns measured via visible or ultraviolet scanning, or internal porosity features measured via x-ray computed tomography (CT) scanning. However, these surface features are not impacted by alterations or damage to the part in areas away from the specific features being measured, possibly preventing the detection of attacks or damage to other areas of the part in transit. CT scanning can be used to detect damage or alterations to more areas of the part and incorporate this measurement into the identification mechanism, but may be prohibitively expensive while also possibly failing to properly penetrate and measure a sufficiently complex AM part. In this work, efforts to expand the cyber-physical security of the AM process are explored, including (1) a novel method of in situ process validation by means of covertly transmit- ting process quality information to an otherwise air-gapped monitoring system, (2) a novel method of metal AM part identification via a low-cost piezoelectric sensor-actuator able to record a part frequency response that is dependent on the geometry and material properties of the part as a whole, (3) an exploration of part-to-part variation across AM processes, again measured via a piezoelectric sensor-actuator, and (4) a novel means of using the same piezoelectric sensor-actuator for detecting the presence of remaining powder in metal AM parts. / Doctor of Philosophy / Additive manufacturing (AM), or 3D printing, allows for the creation of highly complex parts. AM machines do this by building parts layer-by-layer, processing (e.g., selectively melting metal powder) and placing each segment of a part from the bottom up, allowing it to make internal features which would be impossible with traditional manufacturing processes, such as machining.. However, because these parts can be so complicated, it is difficult to validate that a part is "good", i.e., is free from defects. As the entire volume of the part is built layer- by-layer, any layer anywhere in the part could be defective, with very few techniques being capable of detecting the defect from the outside. Worse, because the AM process is driven by digital design files and other data, cyber attacks have the ability to maliciously change the design of a part before it is made, resulting in physical damage. These cyber-physical attacks can similarly affect existing validation methods, allowing these attacked parts to slip through undetected. Alternatively, part designs can be stolen, allowing the thieves to produce unauthorized and possibly subpar counterfeits. These dangers require new means of validating the AM process and the parts it can produce. In order to detect a cyber-physical attack, previous studies have looked to recording and monitoring the physical actions of the AM process in order to ensure the part is built layer- by-layer as expected. Typically, the part design files are sent to the network-connected monitoring system, which then compares the files to the as-built geometry being recorded. However, in this case, the design files can themselves be attacked, as can the monitoring system recording and comparing the part geometry. In order to detect bad parts without exposing the system to cyber attacks, the monitoring system needs a way to validate the AM part without relying on the part design files directly or being connected to the network. To determine is a part is counterfeit or not, previous studies have tried to create "fingerprints" for parts, allowing a unique part to be identified. However, many of these techniques require changes to the part in question, or rely on features that could be duplicated (i.e. copying the fingerprint) by a skilled attacker. Certain methods using x-ray computed tomography (CT) scanning, while effective at fingerprinting small parts, can be very expensive, and may not work for parts which are too large or complex for x-rays to cleanly pass through. To be successful, a fingerprint needs to be simple to measure, and dependent on the entirety of the part itself, not just a handful of manufactured features. This can be done using the frequency response of the part, or how much the part vibrates over a range of frequencies. This response is dependent on the entire part, including the geometry and the material properties, and can be measured using low-cost equipment, allowing it to be used for a variety of different purposes. In this work, several methods to enhance the cyber-physical security of the AM process are explored. These include (1) a method of validating the AM process by covertly transmitting information to a network disconnected monitoring system, (2) a method of identifying metal AM parts identification using the parts frequency response as a fingerprint, (3) an exploration of part frequency response for fingerprinting across other AM processes, including both metal and polymer parts, and (4) a means of using the frequency response of a part for detecting the presence of residual powder from powder-based AM processes.

In today's interconnected infrastructures, Cyber-Physical Systems (CPSs) play a critical role in domains including water distribution, agricultural production, and energy management. Modern infrastructures rely on a network of cyber-physical components—mechanical actuators, electrical sensors, and internet-connected devices—to supervise and manage operational processes. However, the increasing complexity and connectivity of these systems amplify their vulnerability to cyberattacks, necessitating robust cybersecurity measures and effective Outlier Detection (OD) methods. These methods are essential to prevent infrastructure failures, reduce environmental waste, and mitigate damages caused by malicious activities. Existing approaches often lack the integration of multiple operational metrics and context-driven techniques, hampering their effectiveness in real-world scenarios. In large CPSs—comprising hundreds or thousands of sensors, actuators, PLCs, IoT devices, and complex Control and Protection Switching Gear (CPSG)—the challenge of ensuring data quality, security, and reliability is costly. Cyberattacks frequently appear as outliers or anomalies in the data and are launched with "minimum perturbation," making their detection significantly challenging. This dissertation proposes a novel framework, multiple pipelines, and AI-based methods to develop context-driven, data-driven, and assurance-focused OD solutions. Emphasis is placed on water and agricultural systems, illustrating the proposed framework's effectiveness, particularly through enhanced decision-making, operational efficiency, and cybersecurity measures. A comprehensive survey of OD methods that employ Artificial Intelligence (AI) techniques establishes the foundational understanding of OD. This survey underscores that successful OD depends on domain knowledge, contextual factors, and assurance principles. Synthesizing these insights, the dissertation leverages synthetically generated SCADA data and GAN-produced poisoned data, as well as real-world SCADA data from Wastewater Treatment Plants (WWTPs), to identify outliers and address critical problems—such as forecasting tunnel wastewater overflows under extreme weather conditions—by applying Recurrent Neural Network (RNN)-based Deep Learning (DL) methods. Additionally, an AI-based decision support tool is introduced to detect anomalies in complex plant data and optimize operational set-points, thereby aiding Operation and Maintenance (OandM) in Water Distribution Systems (WDSs). Similarly, in Agricultural Production Systems (APSs), which traditionally rely on reactive policies and short-term solutions, integrating advanced AI-driven OD methods provides farmers with timely, data-informed decisions that account for contextual changes resulting from outlier events. Machine Learning (ML) and DL methods measure associations, correlations, and causations among global and domestic factors, aiding in the accurate prediction of agricultural production. This contextual awareness helps manage policy, optimize resource utilization, and support precision agriculture strategies. The main contributions of this dissertation include introducing a novel framework that integrates OD techniques with AI assurance and context-driven methodologies in CPSs; developing multiple pipelines and DL models that enhance anomaly detection, forecasting accuracy, and proactive decision support in WDSs and APSs; and demonstrating measurable improvements in cybersecurity, operational efficiency, and predictive capability using real-world and synthetic data. These efforts collectively foster more trustworthy and sustainable CPSs. Experimental results are recorded, evaluated, and discussed, revealing that these contributions bridge the gap between complex theoretical constructs and tangible real-world applications. / Doctor of Philosophy / Recent unprecedented AI and sensor technology advancements are transforming all domains, including Water Distribution Systems (WDSs) and Agricultural Production Systems (APSs). With Industry 4.0, WDSs and APSs are undergoing a significant digital transformation to enable data-driven monitoring and control of utility operations. Incorporating cyber elements—such as sensors, actuators, data transmitters, receivers, Programmable Logic Controllers (PLCs), and Internet of Things (IoT) devices—aims to make these Cyber-Physical Systems (CPSs) more effective in Operation and Maintenance (OandM). However, this progress comes with a trade-off, as CPSs become increasingly vulnerable to security and safety threats. For example, in 2013, hackers seized control of a small Florida dam, releasing unprocessed water into nearby communities. Furthermore, on February 5th, 2021, a Florida water treatment plant (in Oldsmar, FL) was compromised when the hacker altered the levels of sodium hydroxide (NaOH) in the water—a chemical that would severely damage human tissue. Recent targeted attacks on infrastructure in Ukraine also highlight the risks facing critical infrastructures worldwide, including WDSs. These events suggest that current control operations are largely exposed, necessitating sophisticated learning algorithms that can estimate system states, detect anomalies, and mitigate the harm caused by such intrusions. Technology has fundamentally transformed agriculture as well, significantly impacting this domain. Agriculture, a vital occupation in numerous countries, now faces increasing global population pressures. The United Nations (UN) projects the population to reach 9.7 billion by 2050, intensifying the strain on limited arable land. With only a 4% increase in cultivable land expected by 2050, farmers must do more with less. Traditional methods are insufficient to meet the soaring demands, as a 60% increase in food production is needed to feed an additional two billion people. This necessity for enhanced productivity and reduced waste drives the integration of AI into the agricultural sector. AI adoption not only accelerates efficiency but also increases production volumes, shortening the time from farm to market. This dissertation proposes novel, data- and context-driven Deep Learning (DL)-based methods and decision-support tools to enhance cybersecurity and anomaly detection within WDSs and APSs. Focusing on these critical infrastructures demonstrates how AI-driven strategies can effectively address real-world challenges and improve resilience, operational efficiency, and overall trustworthiness. The contributions of this dissertation include a framework and pipelines that incorporate contextual insights and AI assurance principles to improve anomaly detection and cybersecurity in these domains; the development of DL models tailored for identifying complex outliers and providing actionable decision-support, thereby optimizing resource allocation and ensuring sustainable operations; and validation of these approaches through experimental evaluations using real-world and synthetic data. Collectively, these efforts highlight significant improvements in reliability, efficiency, and scalability for critical infrastructure management, bridging the gap between theoretical advances in AI-driven anomaly detection and their practical application in WDSs and APSs.

In the era of experience economy, how best to deliver memorable and exciting customer experiences has become a key issue for service providers. This research aims to build a systematical, quantitative and expectation-based mechanism to design and manage service experience and operation for cyber-physical services. Consequently, this study not only analyzes and synthesizes the critical factors by reviewing literatures (that is, customer expectation, service operation and customer emotion) within the background of service science but also establishes a conceptual theoretical framework for designing satisfactory service experiences. Furthermore, this study presents a concept of the Exquisite Technology and a service system (i.e. U2EX) with a customer expectation management engine (including five core methods, Hawk-Dove game, Coopetition, PSO, FCM and expectation measurement model) in the exhibition context to demonstrate the feasibility of implementing the notions of customer expectation management and service experience design. Besides, we integrate the expectation theory with the emotion theory to build a theoretical concept and employ a multimethod (including a single case study, interviews, simulations and questionnaire surveys) to test the relations and research propositions of the theoretical concept. The research results show positive evidences to support our developed theoretical concept. The customer expectation measurement model is one critical element of the proposed engine that can help service providers understand and quantify customer expectation in dynamic and real time environments for appropriate service experiences based on the systematical and theoretical groundings (i.e. Fechner’s law and operation risk). Hence, we use the simulations to verify the reliability of the customer expectation measurement model. Meanwhile, this research also conducts simulation experiments of Hawk-Dove game, PSO, FCM and Coopetition methods to have preliminary evidences for supporting the proposed mechanism. Thus, service providers provide customers with high-quality service experiences to achieve customer satisfaction and co-create values with customers through meticulous service experiences design approaches. The proposed mechanism of expectation-based service experience and operation design and management has been demonstrated in the exhibition service sector. We would like to apply the advantage and usage of the proposed mechanism to the other feasible domains and service sectors. Consequently, this study proposes a S-D based input-output analysis approach in order to find the potential fields that can also adopt the proposed mechanism by measuring the effects of technology spillovers.