Assessing Robustness in Systems of Systems Meta-architectures

AbstractNew Systems of Systems (SoSs) are frequently created through partial participation from independent, existing systems with their own continuing missions. These systems’ contribution to the SoS may be contingent on changing priorities and conditions. Therefore, when devising a SoS architecture, consideration should be given to the SoS robustness to occasional lack of availability of individual systems. Robustness is generally the ability to deliver capability in unknown future conditions. Not knowing which systems will be present would seem to fit this definition for a SoS. A fuzzy approach to defining SoS performance in terms of capabilities provided by each type of system and interfaces, with the robustness defined by the remaining performance when one system is absent, is proposed. Several types of capabilities are typically available from several systems when putting together an SoS. An ideal SoS architecture, while paying due homage to lean principles and affordability, should also avoid single point failures. When a suitable SoS performance model exists, the architect can assess SoS capability measures of performance changes when each individual system is removed. Even low fidelity performance models can help distinguish among alternative SoS architectures. Testing architecture models for robustness can improve overall understanding of the SoS capabilities, and selecting architectures for more distributed performance increases robustness