Exploiting Architectural Communities in Early Life Cycle Cost Estimation

System architectures evolve over time. Accordingly, the dynamic properties of architectures reflect how systems respond to change, and this response ultimately impacts cost. In prior work we make an explicit connection between the architectural diagrams of Model-Based Systems Engineering (MBSE), parametric cost estimation, and network science. Specifically, by treating the DoD Architecture Framework (DoDAF) Systems View 3 (SV3) as an adjacency matrix, we assess how the addition of a new subsystem to an immature architecture might grow the existing network. With the subsequent application of parametric cost modeling, we translate anticipated growth into expected cost, thereby quantifying the impact of change. This paper refines that approach. In particular, by using the Girvan-Newman algorithm, the SV3 is initially divided into groups of subsystems such that the number of interfaces is dense within and sparse between groups. Based on this division into “architectural communities” and the prevalence of bridging ties, interfaces generated by the addition of a new subsystem can be faithfully integrated into the existing architecture, adding validity to our growth mechanism. This procedure is illustrated in detail with an example that highlights the importance of this refinement, and it is incorporated within a Monte Carlo simulation that allows the distribution of future costs to be estimated and assessed.