The main purpose of this dissertation is to define and solve problems on optimal sensing and actuating policies in Cyber-Physical Systems (CPSs). Cyber-physical system is a term that was introduced recently to define the increasing complexity of the interactions between computational hardwares and their physical environments. The problem of designing the ``cyber'' part may not be trivial but can be solved from scratch. However, the ``physical'' part, usually a natural physical process, is inherently given and has to be identified in order to propose an appropriate ``cyber'' part to be adopted. Therefore, one of the first steps in designing a CPS is to identify its ``physical'' part. The ``physical'' part can belong to a large array of system classes. Among the possible candidates, we focus our interest on Distributed Parameter Systems (DPSs) whose dynamics can be modeled by Partial Differential Equations (PDE). DPSs are by nature very challenging to observe as their states are distributed throughout the spatial domain of interest. Therefore, systematic approaches have to be developed to obtain the optimal locations of sensors to optimally estimate the parameters of a given DPS. In this dissertation, we first review the recent methods from the literature as the foundations of our contributions. Then, we define new research problems within the above optimal parameter estimation framework. Two different yet important problems considered are the optimal mobile sensor trajectory planning and the accuracy effects and allocation of heterogeneous sensors. Under the remote sensing setting, we are able to determine the optimal trajectories of remote sensors. The problem of optimal robust estimation is then introduced and solved using an interlaced ``online'' or ``real-time'' scheme. Actuation policies are introduced into the framework to improve the estimation by providing the best stimulation of the DPS for optimal parameter identification, where trajectories of both sensors and actuators are optimized simultaneously. We also introduce a new methodology to solving fractional-order optimal control problems, with which we demonstrate that we can solve optimal sensing policy problems when sensors move in complex media, displaying fractional dynamics. We consider and solve the problem of optimal scale reconciliation using satellite imagery, ground measurements, and Unmanned Aerial Vehicles (UAV)-based personal remote sensing. Finally, to provide the reader with all the necessary background, the appendices contain important concepts and theorems from the literature as well as the Matlab codes used to numerically solve some of the described problems.

This dissertation presents the design, simulation, and execution of multi-unmanned aerial vehicles (UAV)-based cooperative source seeking and contour mapping with cyber-physical systems(CPS). Both fixed-wing UAVs and vertical takeoff and landing (VTOL) UAVs are developed as the flight platforms. Compared with single UAV, multi-UAV formation with advanced cooperative control strategies has more advantages for radiative source detection, especially in urgent tasks, for example, detecting nuclear radiation before deploying the salvage. A contour mapping algorithm for the nuclear radiation is proposed, and practical predefined waypoint-based formation flights are realized. Next, four scenarios are developed for multi-UAV-based cooperative source seeking and contour mapping of radiative signal fields. the fixed wing UAVs are more suitable for widespread detection, while VTOL UAVs are better for accurate detection. The flight control of each VTOL UAV is very critical, and the designed fractional order controller for pitch-loop control guarantees the stable flight. The conclusion of this dissertation and suggestions for future research are presented in the end.

Les Systèmes Cyber Physiques (CPS) intègrent des composants programmables afin de contrôler un processus physique. Ils sont désormais largement répandus dans différentes industries comme l’énergie, l’aéronautique, l’automobile ou l’industrie chimique. Parmi les différents CPS existants, les systèmes SCADA (Supervisory Control And Data Acquisition) permettent le contrôle et la supervision des installations industrielles critiques. Leur dysfonctionnement peut engendrer des impacts néfastes sur l’installation et son environnement.Les systèmes SCADA ont d’abord été isolés et basés sur des composants et standards propriétaires. Afin de faciliter la supervision du processus industriel et réduire les coûts, ils intègrent de plus en plus les technologies de communication et de l’information (TIC). Ceci les rend plus complexes et les expose à des cyber-attaques qui exploitent les vulnérabilités existantes des TIC. Ces attaques peuvent modifier le fonctionnement du système et nuire à sa sûreté.On associe dans la suite la sûreté aux risques de nature accidentelle provenant du système, et la sécurité aux risques d’origine malveillante et en particulier les cyber-attaques. Dans ce contexte où les infrastructures industrielles sont contrôlées par les nouveaux systèmes SCADA, les risques et les exigences liés à la sûreté et à la sécurité convergent et peuvent avoir des interactions mutuelles. Une analyse de risque qui couvre à la fois la sûreté et la sécurité est indispensable pour l’identification de ces interactions ce qui conditionne l’optimalité de la gestion de risque.Dans cette thèse, on donne d’abord un état de l’art complet des approches qui traitent la sûreté et la sécurité des systèmes industriels et on souligne leur carences par rapport aux quatre critères suivants qu’on juge nécessaires pour une bonne approche basée sur les modèles : formelle, automatique, qualitative et quantitative, et robuste (i.e. intègre facilement dans le modèle des variations d’hypothèses sur le système).On propose ensuite une nouvelle approche orientée modèle d’analyse conjointe de la sûreté et de la sécurité : S-cube (SCADA Safety and Security modeling), qui satisfait les critères ci-dessus. Elle permet une modélisation formelle des CPS et génère l’analyse de risque qualitative et quantitative associée. Grâce à une modélisation graphique de l’architecture du système, S-cube permet de prendre en compte différentes hypothèses et de générer automatiquement les scenarios de risque liés à la sûreté et à la sécurité qui amènent à un évènement indésirable donné, avec une estimation de leurs probabilités.L’approche S-cube est basée sur une base de connaissance (BDC) qui décrit les composants typiques des architectures industrielles incluant les systèmes d’information, le contrôle et la supervision, et l’instrumentation. Cette BDC a été conçue sur la base d’une taxonomie d’attaques et modes de défaillances et un mécanisme de raisonnement hiérarchique. Elle a été mise en œuvre à l’aide du langage de modélisation Figaro et ses outils associés. Afin de construire le modèle du système, l’utilisateur saisit graphiquement l’architecture physique et fonctionnelle (logiciels et flux de données) du système. L’association entre la BDC et ce modèle produit un modèle d’états dynamiques : une chaîne de Markov à temps continu. Pour limiter l’explosion combinatoire, cette chaîne n’est pas construite mais peut être explorée de deux façons : recherche de séquences amenant à un évènement indésirable ou simulation de Monte Carlo, ce qui génère des résultats qualitatifs et quantitatifs.On illustre enfin l’approche S-cube sur un cas d’étude réaliste : un système de stockage d’énergie par pompage, et on montre sa capacité à générer une analyse holistique couvrant les risques liés à la sûreté et à la sécurité. Les résultats sont ensuite analysés afin d’identifier les interactions potentielles entre sûreté et sécurité et de donner des recommandations. / Cyber physical systems (CPS) denote systems that embed programmable components in order to control a physical process or infrastructure. CPS are henceforth widely used in different industries like energy, aeronautics, automotive, medical or chemical industry. Among the variety of existing CPS stand SCADA (Supervisory Control And Data Acquisition) systems that offer the necessary means to control and supervise critical infrastructures. Their failure or malfunction can engender adverse consequences on the system and its environment.SCADA systems used to be isolated and based on simple components and proprietary standards. They are nowadays increasingly integrating information and communication technologies (ICT) in order to facilitate supervision and control of the industrial process and to reduce exploitation costs. This trend induces more complexity in SCADA systems and exposes them to cyber-attacks that exploit vulnerabilities already existent in the ICT components. Such attacks can reach some critical components within the system and alter its functioning causing safety harms.We associate throughout this dissertation safety with accidental risks originating from the system and security with malicious risks with a focus on cyber-attacks. In this context of industrial systems supervised by new SCADA systems, safety and security requirements and risks converge and can have mutual interactions. A joint risk analysis covering both safety and security aspects would be necessary to identify these interactions and optimize the risk management.In this thesis, we give first a comprehensive survey of existing approaches considering both safety and security issues for industrial systems, and highlight their shortcomings according to the four following criteria that we believe essential for a good model-based approach: formal, automatic, qualitative and quantitative and robust (i.e. easily integrates changes on system into the model).Next, we propose a new model-based approach for a safety and security joint risk analysis: S-cube (SCADA Safety and Security modeling), that satisfies all the above criteria. The S-cube approach enables to formally model CPS and yields the associated qualitative and quantitative risk analysis. Thanks to graphical modeling, S-cube enables to input the system architecture and to easily consider different hypothesis about it. It enables next to automatically generate safety and security risk scenarios likely to happen on this architecture and that lead to a given undesirable event, with an estimation of their probabilities.The S-cube approach is based on a knowledge base that describes the typical components of industrial architectures encompassing information, process control and instrumentation levels. This knowledge base has been built upon a taxonomy of attacks and failure modes and a hierarchical top-down reasoning mechanism. It has been implemented using the Figaro modeling language and the associated tools. In order to build the model of a system, the user only has to describe graphically the physical and functional (in terms of software and data flows) architectures of the system. The association of the knowledge base and the system architecture produces a dynamic state based model: a Continuous Time Markov Chain. Because of the combinatorial explosion of the states, this CTMC cannot be exhaustively built, but it can be explored in two ways: by a search of sequences leading to an undesirable event, or by Monte Carlo simulation. This yields both qualitative and quantitative results.We finally illustrate the S-cube approach on a realistic case study: a pumped storage hydroelectric plant, in order to show its ability to yield a holistic analysis encompassing safety and security risks on such a system. We investigate the results obtained in order to identify potential safety and security interactions and give recommendations.

The operation of the smart grid is expected to be heavily reliant on microprocessor-based control. Thus, there is a strong need for interoperability standards to address the heterogeneous nature of the data in the smart grid. In this research, we analyzed in detail the security threats of the Generic Object Oriented Substation Events (GOOSE) and Sampled Measured Values (SMV) protocol mappings of the IEC 61850 data modeling standard, which is the most widely industry-accepted standard for power system automation and control. We found that there is a strong need for security solutions that are capable of defending the grid against cyber-attacks, minimizing the damage in case a cyber-incident occurs, and restoring services within minimal time. To address these risks, we focused on correlating cyber security algorithms with physical characteristics of the power system by developing intelligent agents that use this knowledge as an important second line of defense in detecting malicious activity. This will complement the cyber security methods, including encryption and authentication. Firstly, we developed a physical-model-checking algorithm, which uses artificial neural networks to identify switching-related attacks on power systems based on load flow characteristics. Secondly, the feasibility of using neural network forecasters to detect spoofed sampled values was investigated. We showed that although such forecasters have high spoofed-data-detection accuracy, they are prone to the accumulation of forecasting error. In this research, we proposed an algorithm to detect the accumulation of the forecasting error based on lightweight statistical indicators. The effectiveness of the proposed algorithms was experimentally verified on the Smart Grid testbed at FIU. The test results showed that the proposed techniques have a minimal detection latency, in the range of microseconds. Also, in this research we developed a network-in-the-loop co-simulation platform that seamlessly integrates the components of the smart grid together, especially since they are governed by different regulations and owned by different entities. Power system simulation software, microcontrollers, and a real communication infrastructure were combined together to provide a cohesive smart grid platform. A data-centric communication scheme was selected to provide an interoperability layer between multi-vendor devices, software packages, and to bridge different protocols together.

No description available.

No description available.

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.

Internet of Things (IoT) is a promising and thriving technology that incorporates a variety of smart devices that provide enhanced services for remote communication and interaction between humans and physical items. The number of deployed IoT devices will increase to 41.6 billion in 2025, as predicted by International Data Corporation. With such a large population, assaults on IoT networks will harm a vast number of users and IoT devices. In light of this, we explore security from physical and network viewpoints in this thesis. To preserve privacy in IoT environment, this thesis begins by proposing RASA, a context-sensitive access authorization approach. We evaluate the promise of RASA-generated policies against a heuristic rule-based policy. The decisions of the RASA and that of the policy are more than 99% consistent. Furthermore, not only physical attacks but also cybercrimes will threaten IoT networks; consequently, this thesis proposes various Network Intrusion Detection System (NIDS) to identify network intrusions. In this thesis, we firstly examine traditional attacks in the NSL-KDD dataset that can impact sensor networks. Furthermore, in order to detect the introduced attacks, we study eleven machine learning algorithms, among which, XGBoost ranks the first with 97% accuracy. As attack tactics continue to evolve, Advanced Persistent Threat (APT) poses a greater risk to IoT networks than traditional incursions. This thesis presents SCVIC-APT-2021 to define a APT benchmark. Following upon this, an ML-based Attack Centric Method (ACM) is introduced achieving 9.4% improvement with respect to the baseline performance. This thesis proposes a Combined Intrusion Detection System (CIDS) that takes network and host information into consideration to reduce data noise and improve the performance of IDS. Two new CIDS datasets, SCVIC-CIDS-2021 and SCVIC-CIDS-2022, are generated. We further propose CIDS-Net to incorporate network and host related data. CIDS-Net boost the macro F1 score of the best baseline by 5.8% (up to 99.95%) and 5.1% (up to 91.3%), respectively on the two datasets. Besides of detection performance, timely response is considered as a critical metric of NIDS. This thesis introduces Multivariate Time Series (MTS) early detection into NIDS . We form TS-CICIDS2017 which is a time series based NIDS dataset and a new deep learning-based early detection model called Multi-Domain Transformer (MDT) is proposed, resulting in a 84.1% macro F-score with only few of the initial packets. To reduce the size of NIDS inputs, this work proposes a deep learning-based lossy time series compressor (Deep Dict) to achieve a high compression ratio while limiting the decompression error within a desired range. As demonstrated by the results, Deep Dict outperforms the compression ratio of the state-of-the-art lossy compression methods by up to 53.66%.

The infrastructure of cyber-physical systems (CPS) is based on a meta-concept of cybermanufacturing systems (CMS) that synchronizes the Industrial Internet of Things (IIoTs), Cloud Computing, Industrial Control Systems (ICSs), and Big Data analytics in manufacturing operations. Artificial Intelligence (AI) can be incorporated to make intelligent decisions in the day-to-day operations of CMS. Cyberattack spaces in AI-based cybermanufacturing operations pose significant challenges, including unauthorized modification of systems, loss of historical data, destructive malware, software malfunctioning, etc. However, a cybersecurity framework can be implemented to prevent unauthorized access, theft, damage, or other harmful attacks on electronic equipment, networks, and sensitive data. The five main cybersecurity framework steps are divided into procedures and countermeasure efforts, including identifying, protecting, detecting, responding, and recovering. Given the major challenges in AI-enabled cybermanufacturing systems, three research objectives are proposed in this dissertation by incorporating cybersecurity frameworks. The first research aims to detect the in-situ additive manufacturing (AM) process authentication problem using high-volume video streaming data. A side-channel monitoring approach based on an in-situ optical imaging system is established, and a tensor-based layer-wise texture descriptor is constructed to describe the observed printing path. Subsequently, multilinear principal component analysis (MPCA) is leveraged to reduce the dimension of the tensor-based texture descriptor, and low-dimensional features can be extracted for detecting attack-induced alterations. The second research work seeks to address the high-volume data stream problems in multi-channel sensor fusion for diverse bearing fault diagnosis. This second approach proposes a new multi-channel sensor fusion method by integrating acoustics and vibration signals with different sampling rates and limited training data. The frequency-domain tensor is decomposed by MPCA, resulting in low-dimensional process features for diverse bearing fault diagnosis by incorporating a Neural Network classifier. By linking the second proposed method, the third research endeavor is aligned to recovery systems of multi-channel sensing signals when a substantial amount of missing data exists due to sensor malfunction or transmission issues. This study has leveraged a fully Bayesian CANDECOMP/PARAFAC (FBCP) factorization method that enables to capture of multi-linear interaction (channels × signals) among latent factors of sensor signals and imputes missing entries based on observed signals.

In the past, there were perceived delineations between the cyber world and the physical world. We are becoming increasingly aware of the overlap between these two worlds, and the overlap itself is increasing. The overlap between these two worlds is known as cyber-physical systems. There have been several incidents involving cyber-physical systems and the number of these incidents is increasing dramatically. In the past there has been no effort to identify methods for describing these incidents in the unique context of cyber-physical systems. This research provides a taxonomy for classifying these incidents that focuses on cross domain, impact oriented analysis. A repository for information about these incidents has also been created as part of this research.