Reliability and Resilience of Systems of Systems

Systems of systems (SoSs), which are complex structures characterized by many interconnected entities performing different functions, pose new challenges to reliability engineers. Emergent behavior, coupling between different time scales, operation near instability limits, occurrence of extreme events, and dynamic failure propagation are characteristics that are not easily described with classical reliability engineering methods. The IEEE Reliability Society has, therefore, created a technical committee on systems of systems in order to advance the state of the art. The goals, results to date, and future plans of that committee are presented in this article. The case is made that beyond reliability, resilience should be a key framework for SoS assessment. Two use cases are described: a dual-function power station example from Hydro-Québec and a multimodal urban transit system example from the Washington Metropolitan Area Transit Authority (WMATA). Enhancing the resilience of critical infrastructures is also discussed. This article concludes with future directions, which include artificial intelligence (AI)/machine learning (ML), specifically graph neural networks (GNNs), as well as aggregation methods inspired by statistical physics.