Towards a Methodology to Design Provably Secure Cyber-physical Systems

Abstract

The inordinate financial cost of mitigating post-production cybersecurity vulnerabilities in cyber-physical systems (CPS) is forcing the industry to rethink systems design cycles: greater attention is being given to the design phase - with the goal of reducing the attack surface of systems at an early stage (i.e., before silicon tape out). Fortunately, formal methods have advanced to the point that they can address such needs and contribute towards achieving security certification. However, new methods and tools focusing on industrial scalability and usability for systems engineers are required. In this ongoing research paper, we describe a framework that will help systems engineers to: a) design cyber-assured CPS using a Model Based Engineering (MBE) approach; b) formally map security requirements to different hardware and software blocks in the model; and c) formally verify security requirements. Based on the nature of each requirement, our framework collects formal correctness evidence from different tools: while high-level architectural properties are suitable for a contract- or ontology-based reasoning, more complex properties with rich semantics require the use of model checking or theorem proving techniques.