System-of-Systems Engineering (SoSE) has become a constantly growing field within product development for complex systems. Systems are becoming more and more connected with other systems and the operational environment in general. This takes the development process to new levels of complexity where high degrees of uncertainties are expected due to ever occurring changes in the operational environment, and other external factors such as politics, economy, and technology. This creates a need of being able to understand the influence of changes early in the development process and to facilitate the systems? perseverance. The focus of the development shifts from fulfilling specific requirements, to being able to meet needs and deliver capabilities over time. Additionally, modeling and simulation for complex systems and System-of-Systems (SoS) becomes a valued alternative to the economically prohibited and almost impossible prototype testing. In consideration of this problem, the presented work introduces a method for both modeling and simulation of a SoS. The method uses ontology with description logic reasoning to derive and narrow down a SoS design space which is further analyzed using Agent Based Simulation (ABS). A Search and Rescue (SAR) scenario is used as a case study to test the method. Measures of Effectiveness (MoE), based on the time it takes to find a rescue subject and the cost of doing so, are used to evaluate the SoS performances. The presented method is envisioned to be used early in the development of complex systems and SoS to increase the overall understanding of them.

System of Systems (SoS) are everywhere, either created and directed by an organization or simply emerging to fulfill a need. SoS consist of many independent Constituent Systems (CS) with their own life cycles and behavior. One of the big challenges with studying SoS is building models to support the analysis of these large complex systems. For existent SoS, it may be tempting to use available models from subsequent levels and connect them in a common framework, rather than creating new models tailored for the SoS analysis. Typically, the already existing models often have a high level of fidelity and have been created from a certain point of view to best fit its purpose. This paper shows that a framework with higher fidelity models on Sub-Systems (SS) level will impact the performance on SoS level but also take longer time to converge and thus need more simulations in order for the results to be trusted. With an holistic approach to the development of SoS, we use hierarchical integration of CS and SS models to provide inputs to the simulation on SoS level. Using maritime Search and Rescue (SAR) as a case study, the mission performance of two aircraft searching for a lost life raft through Agent Based Simulation (ABS). Two cases are compared where one of them has a higher fidelity representation of the sensor model. Through the case study it is possible to investigate the impact of model fidelity on the SS level to the simulation results on SoS level. The two cases simulate the same scenario but still show a difference in mission performance. A convergence analysis show that with the the High Fidelity Models (HFM) the simulation takes slightly longer to converge. This paper is the beginning of a multi-fidelity model library that will be used to explore the impact of model fidelity on all the levels of SoS and from that find general methods for choosing model fidelity when analyzing SoS.

Studying Systems of Systems (SoS) in relation to Measures of Effectiveness (MoE) is a difficult task. This is due to that SoSs include several system levels and operate in changing environments. There are also computational challenges related to modeling and simulation of SoSs and it is difficult to predict behaviors. Therefore, choosing model fidelity for system models and assessing their impact on MoE is of high relevance to the field. This paper illustrates an approach to modeling and simulation for SoSs that can be used to assess the impact of scenario parameters and to explore when level of fidelity on system level affects the MoE. This is made through a case study based on Search and Rescue (SAR) operations where a Design Of Experiments (DOE) of scenario parameters is performed and simulations of each scenario experiment is performed. The case study shows how Agent Based Simulations (ABS) can be used to obtain the MoE for different SAR missions and how the choice of model fidelity for one of the aircraft’s sensors has different effects for various scenarios. Additionally, a Visual Analytics (VA) approach is introduced and used to create a dashboard for visualizing the obtained simulation results in an interactive way. This allows users to make explorations on the resulting data and see how different scenarios influence the performance of each SoS. Furthermore, the results show that changes in scenario parameters impacts the MoE and that this is difficult to predict, at least using a quantitative method. The results point to the importance of exploration for both the scenario, using an interactive dashboard, but also to the importance of exploring the simulation model to study emerging phenomenon. The results of this approach have raised questions regarding if a more qualitative approach of studying SoSs could be beneficial to study MoE for SoSs and if inspiration could be transferred from a more scientific point of view of systems.