Coping with Uncertainty in Systems-of-Systems Architecture Modeling on the IoT with SosADL

Abstract

A challenging issue in the architectural design of a System-of-Systems (SoS) is how to cope with the uncertainty raised by the limited knowledge of the operational environment where the SoS will actually be deployed as well as the constituent systems which will concretely participate in the SoS at run-time. It is especially the case of SoSs being architected on the Internet-of-Things (IoT). Indeed, due to the open and dynamic nature of the IoT, on the one hand, at design-time, most often the SoS architects do not know which will be the concrete IoT systems that will become constituents of an SoS, these being predominantly identified at run-time; on the other hand, the correct architecture depends not only on the constituent IoT systems but also, largely, on the operational environment where the SoS will be positioned on the IoT. The consequent research question is thereby how to design and describe the SoS architecture in a way that is flexible enough to cope with these different uncertainties. To address this challenge, this paper investigates the notion of uncertainty in SoS and presents how SosADL, a formal SoS Architecture Description Language (ADL), enables to cope with uncertainty in the architecture modeling of SoSs. It presents the concepts and constructs that makes possible to describe SoS architectures which will operate in unpredictable environments on the IoT based on the SosADL support for dealing with partial knowledge, grounded on concurrent constraints.