Cyber-physical systems (CPS), which integrate algorithmic control with physical processes, often consist of physically distributed components communicating over a network. A malfunctioning or compromised component in such a CPS can lead to costly consequences, especially in the context of public infrastructure. It is thus important that unexpected or even malicious behaviour can be detected and reacted to before it is too late.
In this project, we aim to construct invariants (or models) of CPS, and use them for attesting that their physical behaviour is correct (“physical attestation”). To achieve this despite the inherent complexity of CPS, we are investigating a new technique for automatically learning classifiers based on ideas from machine learning and mutation testing. We have performed a preliminary study on a water treatment testbed at SUTD that suggests the efficacy of this approach for protecting against a variety of attacks.
Y. Chen, C. M. Poskitt, and J. Sun. Learning from mutants: Using code mutation to learn and monitor invariants of a cyber-physical system. In Proc. IEEE Symposium on Security and Privacy (S&P 2018), IEEE Computer Society, 2018. To appear. [PDF, Supplementary Material]
Y. Chen, C. M. Poskitt, and J. Sun. Towards learning and verifying invariants of cyber-physical systems by code mutation. In Proc. International Symposium on Formal Methods (FM 2016), volume 9995, LNCS, Springer, 2016, pp. 155-163. [PDF, Supplementary Material]