The electric power industry is undergoing profound transformations due to the advances in information technology, stronger interoperability between the physical and cyber layers of the power grid, and deregulations in the energy market. Driven by such transformations, power grids are rapidly growing in scale, complexity, and interconnectivity, evolving towards distributed networks of interconnected platforms. Such highly distributed structures expose the power grid to critical security vulnerabilities, such that malicious attackers bent on subverting grid operations can severely disrupt energy generation, transmission, and distribution. The overarching goal of this proposal is to develop a decision-theoretic approach that includes security as an integral part of design mechanisms in power grids, rendering inherently secure operations.
The key motivation underpinning this research is that current monitoring, protection, and control functions in power grid are designed without regard for inherent security, and security vulnerabilities are addressed as they are discovered. Therefore, while being effective against naturally occurring faults and attacks with known structures, they lack the versatility to counter highly structured and unknown cyber attacks. This research produces a decision-theoretic framework that integrates security assurance as a constraint in designing and optimizing monitoring and control modules while adopting realistic models for adversaries, and incorporating generation, transmission, and operational constraints. This framework facilitates analyzing the tradeoffs among three key measures involved in designing any operation in monitoring and control modules, namely: (i) security guarantees, (ii) quality the operation (through assigning appropriate performance measures that match the goals of the operations), and (iii) amenability to real-time and scalable implementation. These analytical results guide transformational approaches to secure monitoring and protection in future energy grids. These approaches are in contrast to the existing designs in which operations are optimized for attack-free settings, and are augmented by additional modules to enhance their resiliency against known classes of cyber attacks. The methods used to address these problems include mathematical statistics, statistical signal processing, optimization theory, and power engineering.
The educational component of this research is composed of several parts aimed at students ranging from the high school to graduate level, and from several demographics. In collaboration with the Engineering Ambassadors program at Rensselaer, this research program develops interactive presentations for high-school students about modern notions of cyber security, which involve undergraduate students as assistants in delivering these presentations, who will be also involved in research. Other educational components include designing curricular materials for graduate courses on the applications of modern data science in power grids.