Complex engineering projects require a variety of skills and people often distributed across multiple locations and time zones. This distribution requires both a partitioning and a consolidation. Tasks, responsibilities, requirements and other project information must be partitioned so that people can work independently. At the same time, the resulting disparate information needs to be consolidated to form the whole.
The various tasks in the work breakdown structure may require parallel working across different timescales to ensure a timely delivery of the developed system. However, these tasks will need to be coordinated and work products integrated to ensure that the different teams are working towards a common goal and that the “pieces all fit together.” Model-based Systems Engineering (MBSE) has been developed to help coordinate different work streams and project teams. However, merging large amounts of complex system data together on a regular basis is a complex undertaking requiring matching of interfaces, types, and systems each time. This constant merging becomes time consuming and error prone, especially when system designs start to diverge over time, as they inevitably do. Systems analysis, parametrics, and simulation via external tools also need to operate on individual models as well as on the system model as a whole. Unless there is a way of tying everything together, such whole-system analysis and simulation will not be possible. With the recent increase in distributed working, new approaches to engineering collaboration are needed.
What we need is a different paradigm for MBSE. Rather than a single federated model (or even worse a set of disconnected models) we need an architecture of Distributed Integrated models and services. Although distributed integration sounds like an oxymoron, it is a common way to manage systems and people. Traditional control systems used a federated architecture with a centralized controller located at a central control room or within a central computer. SCADA and distributed control systems use autonomous controllers distributed throughout the system, often over a large geographical region with a centralized system that manages the system of systems (SoS). Each autonomous controller has a clear set of responsibilities and rules governing its behavior and well-defined interfaces. The centralized system does not need to understand the internal details of the autonomous controller, but only the data interfaces and how it works with the other parts of the SoS. They are in constant communication with the centralized system and use data filters to limit data transmission to significant changes. Autonomous controllers can be modified without changing the others, unless its interfaces change.
The integrated distributed model paradigm would work in a similar manner. An initial SoS model would describe the overall engineering effort with the individual work packages assigned to different teams. Each team would have a well-defined set of requirements and tasks. Each distributed model would have a clear definition of how it fits into the SoS. Rather than merging the entire distributed model into the SoS model, “model wrappers” define functional, structural and parametric interfaces to allow integration of the wrapper into the SoS model. This results in a loosely coupled modeling architecture. For trade-off analysis between systems options, results from the distributed model are integrated into the overall model, obviating the need for complex model and parametric integration. This way of working will require more planning and design ahead of time, but this will result in increased team independence and minimize time spent on integration, development and coordination. This presentation will describe the integrated distributed model architecture, its benefits, how this can be implemented and requirements for standardization.
Matthew Hause is a principal at SSI, the INCOSE representative of the UAF group, and a member of the OMG SysML specification team. He has been developing multi-national complex systems for over 40 years as a systems and software engineer. He started out working in the power systems industry then transitioned to command and control systems, process control, communications, SCADA, distributed control, Military systems, and many other areas of technical and real-time systems. He has been a regular presenter at INCOSE, the IEEE, BCS, the IET, the OMG, AIAA, DoD Enterprise Architecture, Embedded Systems Conference and many other conferences. He was recently a keynote speaker at the Model-based Systems Engineering Symposium at the DSTO in Australia. His role at SSI includes consulting, standards development, presentations at conferences, specification of the UAF profile and developing and presenting training courses.