Development Interdependency Modeling for System-of-Systems (SoS) using Bayesian Networks: SoS Management Strategy Planning

Abstract

Managing the development and evolution of a system-of-systems (SoS) capability remains a challenge due to, among other reasons, the complex interdependencies between participating systems. One form of complexity stems from the tendency of interdependencies to propagate between systems; disruptions in the development of one system may propagate to other dependent systems in successive cycles, creating schedule and cost overruns. Event tree methods and Bayesian Networks (BNs) are used in this paper to quantify development interdependencies between systems and assess cascading development risks. In addition the approach also allows inputs (e.g. development failure rates of participating systems) to be updated automatically for better decision-making. A primary output of the approach is the quantification of development interdependencies and the identification of critical systems with respect to propagating effect levels. This method when applied to a synthetic problem, as a proof-of- concept, demonstrates the robustness of the proposed approach in tackling risk interaction that arises from the cascading effects of development disruptions and clearly illustrates that the propagating effects depend not only on SoS architecture, but on development failure rates of participating systems as well. The outcomes of the analysis provide a support for decision makers to manage risk in development of a SoS with complex interdependencies.