Acquisition and Development Programs through the Lens of System Complexity

Abstract

The approach of the Department of Defense (DoD) to acquisition programs is strongly based on systems engineering. DoD Directive 5000.01 calls for "the application of a systems engineering approach that optimizes total system performance and minimizes total ownership costs"(DoD, 2007). Even when systems engineering best practices are employed, the cost of large systems is always increasing, and a large part of this increase is due to system complexity (Arena et al., 2008). Part of this system complexity comes from the functionalities of the system, and is thus justified when these functionalities are required. The remaining contribution is due to unnecessary intricacies in the design, to local optimization, and to oversight in the system-level design. This complexity can lead to rising cost and schedule delays, and should be addressed properly. To overcome these issues regarding cost and schedule overruns, researchers have advocated for the adoption of a complexity budget (Sinha, 2014), which can help identify the effects of unintended interfaces between system elements. While most literature seems to agree about the existence of this issue, the solutions to the measurement of complexity are various and based on different approaches. The purpose of this research is to develop metrics that will allow the DoD to evaluate a complexity budget, particularly in the phases of architecture and design development. The metrics are developed using a set of axioms that can be applied to cyber-physical systems, and they assume that the architecture of the system is known. Knowledge of the system architecture allows for a graph representation of the system and uses graph-theoretic approaches to the evaluation of the topology of the system. Concepts such as graph density and graph energy can be used to build metrics that allow to rank architectures, thus helping identify possible sources of complexity. Additionally, this approach allows engineers to look external to the system to identify the complexity required to interoperate with legacy DoD systems and systems under development. This research effort is limited to a snapshot of the state of the system, but can be extended to a dynamical approach with a system changing state or changing its structure.