|dc.description.abstract||The increasing complexity of net-centric warfare requires assets to cooperate to achieve mission success. Such cooperation requires the integration of many heterogeneous systems into an interoperable system-of-systems (SoS). Interoperability can be considered a metric of an architecture, and must be understood as early as the conceptual design phase. This thesis approaches interoperability by first creating a general definition of interoperability, identifying factors that affect it, surveying existing models of interoperability, and identifying fields that can be leveraged to perform a measurement, including reliability theory and graph theory.
The main contribution of this thesis is the development of the Architectural Resource Transfer and Exchange Measurement of Interoperability for Systems of Systems, or ARTEMIS methodology. ARTEMIS first outlines a quantitative measurement of system pair interoperability using reliability in series and in parallel. This step incorporates operational requirements and the capabilities of the system pair. Next, a matrix of interoperability values for each resource exchange in an operational process is constructed. These matrices can be used to calculate the interoperability of a single resource exchange, IResource, and layered to generate a weighted adjacency matrix of the entire SoS. This matrix can be plugged in to a separate model to link interoperability with the mission performance of the system of systems. One output of the M&S is a single value ISoS that can be used to rank architecture alternatives based on their interoperability. This allows decision makers to narrow down a large design space quickly using interoperability as one of several criteria, such as cost, complexity, or risk.
A canonical problem was used to test the methodology. A discrete event simulation was constructed to model a small unmanned aircraft system performing a search and rescue mission. Experiments were performed to understand how changing the systems' interoperability affected the overall interoperability; how the resource transfer matrices were layered; and if the outputs could be calculated without time- and computationally-intensive stochastic modeling. It was found that although a series model of reliability could predict a range of IResource, M&S is required to provide exact values useful for ranking. Overall interoperability ISoS can be predicted using a weighted average of IResource, but the weights must be determined by M&S.
Because a single interoperability value based on performance is not unique to an architecture configuration, network analysis was conducted to assess further properties of a system of systems that may affect cost or vulnerability of the network. The eigenvalue-based Coefficient of Networked Effects (CNE) was assessed and found to be an appropriate measure of network complexity. Using the outputs of the discrete event simulation, it was found that networks with higher interoperability tended to have more networked effects. However, there was not enough correlation between the two metrics to use them interchangeably. ARTEMIS recommends that both metrics be used to assess a networked SoS.
This methodology is of extreme value to decision-makers by enabling trade studies at the SoS level that were not possible previously. It can provide decision-makers with information about an architecture and allow them to compare existing and potential systems of systems during the early phases of acquisition. This method is unique because it does not rely on qualitative assessments of technology maturity or adherence to standards. By enabling a rigorous, objective mathematical measurement of interoperability, decision-makers will better be able to select architecture alternatives that meet interoperability goals and fulfill future capability requirements.||