Using System of Systems' States for Identifying Emergent Misbehaviors

Coping with unpredictable behavior in complex systems is a very hard task. This becomes more difficult when the system itself is composed of preexisting systems, which is the case of systems of systems (SoS). Often Engineers rely on simulation as the best way to understand and manage the complexity of such a system. In the case of SoS, managing emergent behavior is needed to avoid misbehaviors raised from the composition of the constituent systems. This paper is based on the assumption that an emergent misbehavior leads the SoS to an undesired or inconsistent state. So, we propose to explore simulation results to detect these undesirable SoS states and therefore helping the designer in identifying the corresponding misbehavior and its origin. We use the invariant paradigm to define the acceptable states for the SoS. The violation of the invariants implies that there is undesirable states, and therefore the existence of some emergent misbehaviors. We experimented our approach on a realistic case study concerning the glycemia control, and the results were very promising.