Combination-based multivariate time-series clustering for collision analysis of platooning system of systems

Abstract

A platooning system of systems (Platooning SoS) is a system composed of multiple vehicles that communicate with each other. Reliability should be ensured for a platooning SoS since it is a safety-critical system that operates in a physical environment. To ensure the safety of a platooning SoS, an operational test should be conducted. However, since it is hard to analyze failures of a complex system such as platooning SoS, an efficient fault localization technique is required. Although multivariate time-series clustering approaches have been widely used in complex systems, existing approaches are not directly applicable to platooning SoS. Therefore, this study proposes an extended multivariate time-series clustering method for a platooning SoS. The research evaluated the proposed approach with a platooning simulation dataset and showed that the approach can find meaningful information about collisions.