Explaining Cyber-Physical Systems Using Decision Trees

Start Page
3
End Page
8
Citation
2nd International Workshop on Computation-Aware Algorithmic Design for Cyber-Physical Systems (CAADCPS 2022)
Contribution to Conference
2nd International Workshop on Computation-Aware Algorithmic Design for Cyber-Physical Systems, CAADCPS 2022 
Publisher DOI
10.1109/CAADCPS56132.2022.00006
Scopus ID
2-s2.0-85134358103
Cyber-Physical Systems (CPS) are systems that contain digital embedded devices while depending on environmental influences or external configurations. Identifying relevant influences of a CPS as well as modeling dependencies on external influences is difficult. We propose to learn these dependencies with decision trees in combination with clustering. The approach allows to automatically identify relevant influences and receive a data-related explanation of system behavior involving the system's use-case. Our paper presents a case study of our method for a Real-Time Localization System (RTLS) proving the usefulness of our approach, and discusses further applications of a learned decision tree.