Threat Modeling for AI Analysis: Towards the Usage of Misuse Case Templates and UML Diagrams for AI Experiment Description and Trajectory Generation
Wellßow, Arlena and Kohlisch-Posega, Julian and Veith, Eric M.S.P. and Uslar, Mathias
Proceedings of the 2024 13th International Conference on Informatics, Environment, Energy and Applications
As critical national infrastructures (CNI) become more and more cyber-physical systems (CPS), threat modeling is increasingly relevant in this area. This also applies to energy grids with the introduction of information and communications technologies (ICT). Previously, the redundancy concept could adequately manage the risk of malfunctions and outages of the physical energy grid (N-1 rule). Failures of the physical systems can lead to the failure of the ICT systems. However, with the increasing intertwining of ICT, the opposite direction also becomes possible. Although it is possible to describe the misbehavior of systems by a template belonging to the misuse case methodology, specific information is missing for an AI experiment description. This paper describes how the misuse case methodology based on the use case methodology (IEC 62559) has to be extended to prepare such situations for analysis by learning agents. Therefore, a UML-based approach is taken, and the necessary data is analyzed. Further, this data is mapped from Misuse Cases to AI experiment data. We use a specific tool called palaestrAI to show exemplary AI experiment data. This work shows the missing parts to integrate expert knowledge in AI experiments and proposes a toolchain.
2024
inproceedings
Association for Computing Machinery
IEEA '24
7–16
RESili8
Resilience for Cyber-Physical Energy Systems
Resilience for Cyber-Physical Energy Systems
