CRII: SHF: Model-Based Repair of Cyber-Physical Systems for Improving Resiliency

Model-based design offers a promising approach for assisting developers to build reliable and secure cyber-physical systems (CPS) in a systematic manner. However, constructing a behavioral model at design time that offers resiliency for all kinds of attacks and failures is notoriously difficult. There is currently a shortage of inexpensive, automated software that can effectively repair an initial design, and a model-based system developer regularly needs to redesign and reimplement a system from scratch. The project is developing a methodology, along with an associated framework, to assist a designer in repairing an original CPS model so that it continues to satisfy the correctness requirements under modified assumptions. The project’s novelties are as follows. (1) It provides a fresh approach with an end-to-end design and implementation of a software to facilitate model-based repair for improving the resiliency of CPS against unanticipated attacks and failures. (2) It enables a designer to specify resiliency patterns; the investigator is designing an extensible model transformation language for CPS models. (3) The methodology utilizes formal analysis with respect to correctness requirements formalized in signal temporal logic hyper-properties (HyperSTL) at multiple stages. (4) Software tools are being applied on proof-of-concept case studies where the CPS models can be repaired to mitigate practical attacks. The project’s impacts are in (1) developing new technologies and state-of-the-art software tools to enforce the safety, reliability, security, and resiliency of CPS and (2) strengthening mentorship, skill-building, and workforce readiness for CPS engineering in the Southwest Ohio region and nationally. The proposed framework involves the design, implementation, evaluation, and integration of two main tools: a Model Transformation and a Model Analyzer. A Model Transformation tool consistently incorporates an original state-machine-based model, a collection of resiliency patterns (or potential edits), and feedback from analyzers to produce an updated resilient behavioral model. The tool automatically searches through the extensible library of resiliency patterns, written as model transformation scripts, to solve the model repair problem. A Model Analyzer tool analyzes the system correctness requirements at multiple stages, both at design time and during runtime operation. The complete model generated by the Model Transformation is falsified using a static falsifier, while the corresponding implementation is monitored for violations using a runtime monitor tool. To ensure a rich set of specifications, the investigator is utilizing objectives and safety constraints specified via HyperSTL. An additional feature is a counter-example analyzer that produces feedback to a designer for developing new resiliency patterns. Design and implementation of the tool-chain requires theoretical advances in terms of rigorous formalization, computational engines, and heuristics for scalability. The algorithms for model repair, resiliency patterns, and formal analysis developed in this project are contributions of significant interest to the research community in design and analysis of CPS. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.