System-of-Systems: An architectural framework to support development cost strategies

Abstract

A System-of-Systems (SoS) architectural framework has been institutionalized within the Systems Engineering (SE) discipline in recent years in many industries. The objective is to combine independent product elements into collaborative solutions that leverage the technical and performance capabilities of the elements. SoS-based solutions can accelerate data transmission, increase data fidelity, result in safer products and provide new ways of doing business. On the other hand, such solutions typically take longer to develop and test; cost more to implement and maintain; and provide unexpected reliability and performance characteristics. In a few instances, they have even introduced flaws in security. Root-cause analysis of SoS-based systems that failed to meet performance, cost or schedule goals reveals that, for those systems, early assessments of both Technology Readiness Level (TRL) and the scope of integration efforts tended toward optimism. While the overly optimistic TRL assessments enabled the systems to proceed the optimism also leads to initial life-cycle cost estimates (LCCE) that have often proven unrealistically optimistic. Those estimates in turn have resulted in cost overruns and, for Federal programs, Nunn — McCurdy (N-M) breaches for both cost and schedule. This paper describes an approach to help decision makers accurately characterize a SoS during program formulation that supports defendable LCCEs and budgeting well into program execution without excessive growth. It builds on recent work by Mankins, Sauser, Marquez-Ramirez, Bilbro, McCabe and others into system and integration readiness levels (SRLs/IRLs); advancement degree of difficulty (AD2); and cyclomatic complexity. Using their research provides methods for determining the technological maturity of complex systems and elements. We show how TRL, IRL, SRL and complexity metrics can be used to adjust cost and schedule estimates early on. We demonstrate how to estimate a SRL using TRLs and IRLs, adjust for integration complexity, and then define the effort to forecast a realistic system readiness level that supports LCCEs that are reasonable and credible. Our approach provides the Program Manager/System Engineering Practitioner a tool suite that supports defendable program estimates during initial planning. We will be evaluating TRL & IRL growth in programs, both those with overruns and others.