Exploiting inherent physical structure of the CPS domains can lead to economically viable and efficient novel algorithms for providing performance, control, synchronization and an alternate approach to CPS security that does not rely solely on cryptography. In each of these systems, regardless of the current state of the network, in the presence of disturbances or adversarial inputs, there is a need to bring the system to desired state for performance and control of the network. This project presents one such novel approach by observing that the CPS applications including smartgrid, coordinating robotics, formation flights in UAV, and synchronization of biological systems including brain networks all exhibit a special physical structure, namely submodularity, with respect to the set of control actions. Submodularity is a diminishing returns property that enables the development of efficient algorithms with provable optimality guarantees and in many cases distributed versions that are locally implementable, and hence scalable. While it has been widely used in the machine learning and discrete optimization communities, the use of submodularity in the context of CPS is a fertile research area. This project initially applies submodularity in the context of smart grid and show how it can lead to greater system stability and attack resilience. By defining suitable metrics that capture the submodular structures underlying the physical dynamics, the researchers develop algorithms that eliminate the time-consuming and computationally expensive verification of control actions through simulation. The fundamental properties of synchronization, convergence, robustness, and attack-resilience considered in this effort have crosscutting applications to multiple CPS domains, which will benefit from the submodular approach that we will research and develop.

Laboratory-on-a-chip (LoC) technology is poised to improve global health through development of low-cost, automated point-of-care testing devices. In countries with few healthcare resources, clinics often have drugs to treat an illness, but lack diagnostic tools to identify patients who need them. To enable low-cost diagnostics with minimal laboratory support, this project will investigate domain-specific LoC programming language and compiler design in conjunction with device fabrication technologies (process flows, sensor integration, etc.). The project will culminate by building a working LoC that controls fluid motion through electronic signals supplied by a host PC; a forensic toxicology immunoassay will be programmed in software and executed on the device. This experiment will demonstrate benefits of programmable LoC technology including miniaturization (reduced reagent consumption), automation (reduced costs and uncertainties associated with human interaction), and general-purpose software-programmability (the device can execute a wide variety of biochemical reactions, all specified in software). Information necessary to reproduce the device, along with all software artifacts developed through this research effort, will be publicly disseminated. This will promote widespread usage of software-programmable LoC technology among researchers in the biological sciences, along with public and industrial sectors including healthcare and public health, biotechnology, water supply management, environmental toxicity monitoring, and many others. This project designs and implements a software-programmable cyber-physical laboratory-on-a-chip (LoC) that can execute a wide variety of biological protocols. By integrating sensors during fabrication, the LoC obtains the capability to send feedback in real-time to the PC controller, which can then make intelligent decisions regarding which biological operations to execute next. To bring this innovative and transformative platform to fruition, the project tackles several formidable research challenges: (1) cyber-physical LoC programming models and compiler design; (2) LoC fabrication, including process flows and cyber-physical sensor integration; and (3) LoC applications that rely on cyber-physical sensory feedback and real-time decision-making. By constructing a working prototype LoC, and programming a representative feedback-driven forensic toxicology immunoassay, the project demonstrates that the proposed system can automatically execute biochemical reactions that require a closed feedback loop. Expected broader impacts of the proposed work include reduced cost and increased reliability of clinical diagnostics, engagement with U.S. companies that use LoC technology, training of graduate and undergraduate students, increased engagement and retention efforts targeting women and underrepresented minorities, student-facilitated peer-instruction at UC Riverside, a summer residential program for underrepresented minority high-school students at the University of Tennessee, collaborations with researchers at the Oak Ridge National Laboratory, and creation, presentation, and dissemination of tutorial materials to promote the adoption and use of software-programmable LoCs among the wider scientific community.

Assistive machines - like powered wheelchairs, myoelectric prostheses and robotic arms - promote independence and ability in those with severe motor impairments. As the state- of-the-art in these assistive Cyber-Physical Systems (CPSs) advances, more dexterous and capable machines hold the promise to revolutionize ways in which those with motor impairments can interact within society and with their loved ones, and to care for themselves with independence. However, as these machines become more capable, they often also become more complex. Which raises the question: how to control this added complexity? A new paradigm is proposed for controlling complex assistive Cyber-Physical Systems (CPSs), like robotic arms mounted on wheelchairs, via simple low-dimensional control interfaces that are accessible to persons with severe motor impairments, like 2-D joysticks or 1-D Sip-N-Puff interfaces. Traditional interfaces cover only a portion of the control space, and during teleoperation it is necessary to switch between different control modes to access the full control space. Robotics automation may be leveraged to anticipate when to switch between different control modes. This approach is a departure from the majority of control sharing approaches within assistive domains, which either partition the control space and allocate different portions to the robot and human, or augment the human's control signals to bridge the dimensionality gap. How to best share control within assistive domains remains an open question, and an appealing characteristic of this approach is that the user is kept maximally in control since their signals are not altered or augmented. The public health impact is significant, by increasing the independence of those with severe motor impairments and/or paralysis. Multiple efforts will facilitate large-scale deployment of our results, including a collaboration with Kinova, a manufacturer of assistive robotic arms, and a partnership with Rehabilitation Institute of Chicago. The proposal introduces a formalism for assistive mode-switching that is grounded in hybrid dynamical systems theory, and aims to ease the burden of teleoperating high-dimensional assistive robots. By modeling this CPS as a hybrid dynamical system, assistance can be modeled as optimization over a desired cost function. The system's uncertainty over the user's goals can be modeled via a Partially Observable Markov Decision Processes. This model provides the natural scaffolding for learning user preferences. Through user studies, this project aims to address the following research questions: (Q1) Expense: How expensive is mode-switching? (Q2) Customization Need: Do we need to learn mode-switching from specific users? (Q3) Learning Assistance: How can we learn mode-switching paradigms from a user? (Q4) Goal Uncertainty: How should the assistance act under goal uncertainty? How will users respond? The proposal leverages the teams shared expertise in manipulation, algorithm development, and deploying real-world robotic systems. The proposal also leverages the teams complementary strengths on deploying advanced manipulation platforms, robotic motion planning and manipulation, and human-robot co-manipulation, and on robot learning from human demonstration, control policy adaptation, and human rehabilitation. The proposed work targets the easier operation of robotic arms by severely paralyzed users. The need to control many degrees of freedom (DoF) gives rise to mode-switching during teleoperation. The switching itself can be cumbersome even with 2- and 3-axis joysticks, and becomes prohibitively so with more limited (1-D) interfaces. Easing the operation of switching not only lowers this burden on those already able to operate robotic arms, but may open use to populations to whom assistive robotic arms are currently inaccessible. This work is clearly synergistic: at the intersection of robotic manipulation, human rehabilitation, control theory, machine learning, human-robot interaction and clinical studies. The project addresses the science of CPS by developing new models of the interaction dynamics between the system and the user, the technology of CPS by developing new interfaces and interaction modalities with strong theoretical foundations, and the engineering of CPS by deploying our algorithms on real robot hardware and extensive studies with able-bodied and users with spinal cord injuries.

The wide-area measurement systems technology using Phasor Measurement Units (PMUs) has been regarded as the key to guaranteeing stability, reliability, state estimation, and control of next-generation power systems. However, with the exponentially increasing number of PMUs, and the resulting explosion in data volume, the design and deployment of an efficient wide-area communication and computing infrastructure is evolving as one of the greatest challenges to the power system and IT communities. The goal of this NSF CPS project is to address this challenge, and construct a massively deployable cyber-physical architecture for wide-area control that is fast, resilient and cost-optimal (FRESCO). The FRESCO grid will consist of a suite of optimal control algorithms for damping oscillations in power flows and voltages, implemented on top of a cost-effective and cyber-secure distributed computing infrastructure connected by high-speed wide-area networks that are dynamically programmable and reconfigurable. The value of constructing FRESCO is twofold (1) If a US-wide communication network capable of transporting gigabit volumes of PMU data for wide-area control indeed needs to be implemented over the next five years then power system operators must have a clear sense of how various forms of delays, packet losses, and security threats affect the stability of these control loops. (2) Moreover, such wide-area communication must be made economically feasible and sustainable via joint decision-making processes between participating utility companies, and testing how controls can play a potential role in facilitating such economics. Currently, there is very limited insight into how the PMU data transport protocols may lead to a variety of such delay patterns, or dictate the economic investments. FRESCO will answer all of these questions, starting from small prototypical grid models to those with tens of thousands of buses. Our eventual goal will be to make FRESCO fully open-source for Transition to Practice (TTP). We will work with two local software companies in Raleigh, namely Green Energy Corporation and Real-Time Innovations, Inc. to develop a scalable, secure middleware using Data-Distribution Service (DDS) technology. Thus, within the scope of the project, we also expect to enrich the state-of-the-art cloud computing and networking technologies with new control and management functions. From a technical perspective, FRESCO will answer three main research questions. First, can wide-area controllers be co-designed in sync with communication delays to make the closed-loop system resilient and delay-aware, rather than just delay-tolerant This is particularly important, as PMU data, in most practical scenarios, will have to be transported over a shared resource, sharing bandwidth with other ongoing applications, giving rise to not only transport delays, but also significant delays due to queuing and routing. Advanced ideas of arbitrated network control designs will be used to address this problem. The second question we address is for cost. Given that there are several participants in this wide-area control, how much is each participant willing to pay in sharing the network cost with others for the sake of supporting a system-wide control objective compared to its current practice of opting for selfish feedback control only Ideas from cooperative game theory will be used to investigate this problem. The final question addresses security how can one develop a scientific methodology to assess risks, and mitigate security attacks in wide-area control? Statistical and structural analysis of attack defense modes using Bayesian and Markov models, game theory, and discrete-event simulation will be used to address this issue. Experimental demos will be carried out using the DETER-WAMS network, showcasing the importance of cyber-innovation for the sustainability of energy infrastructures. Research results will be broadcast through journal publications, and jointly organized graduate courses between NCSU, MIT and USC.

Cyber-physical additive layer manufacturing, e.g., 3D printing, has become a promising technology for providing cost, time, and space effective solution by reducing the gap between designers and manufacturers. However, the concern for the protection of intellectual property is arising in conjunction with the capabilities of supporting massive innovative designs and rapid prototyping. Intellectual property in the additive layer manufacturing system consists of: i) geometric design of an object; ii) attributes of an object; iii) process information; and iv) machine information. This Early-concept Grant for Exploratory Research (EAGER) project seeks to develop defense mechanisms for detecting malware and counterfeit articles using a variety of signals that are observed during the manufacturing process including acoustic, temperature, power, and others. The project is an EAGER because both the uniqueness of the observed signal signatures, and their utilization in securing the manufacturing process are high risk with potential for high reward in thwarting attacks. This project will demonstrate that during the life-cycle of the additive layer manufacturing system, the intellectual property information contained in the cyber domain can be recovered/reconstructed through attacks occurring during the manufacturing process in the physical domain through various non-intrusive techniques. It will then focus on creating both machine-dependent and machine-independent defense mechanisms for avoiding such an attack. This project will significantly impact US competitiveness over technology-oriented manufacturing. The attack model will provide feedback to 3D printer manufacturers and CAD tool designers to build defenses against these new types of attack. Moreover, it will have a significant societal impact to the explosively growing maker and crowd-sourcing community in protecting their intellectual property. In addition, the project's approach can be used in other manufacturing systems, e.g., CNC machines, manufacturing robots, etc. This is possibly the very first approach to create defense for additive layer manufacturing mechanisms against such attacks occurring in the physical domain to get access to information of the cyber domain. This project has three specific objectives: 1) It will demonstrate a proof of concept by presenting a novel attack model constructed using a combination of machine learning, signal processing, and pattern recognition techniques that utilize the side-channel information (power, temperature, acoustic, electromagnetic emission) obtained during the manufacturing process. 2) It will develop a machine-specific defense mechanism against the attack model for the 3D printer. New techniques to add additional physical process encryption, e.g. adding extra information to the G-code to obfuscate the printing process from the attack model between the G-code and the physical manufacturing process, will be demonstrated. 3) It will create a new security-aware 3D-printing algorithm for the machine-independent CAD tools that can protect against such side channel attacks. The 3D-printing algorithm will slice the STL and generate layer description language (e.g. G-code) randomly so that for the same 3D object, different instructions will be sent to the 3D printer and eventually different physical features will be extracted by the attackers.