Digital Twins are seen as a potential solution for many difficult problems in developing military platforms; by keeping a physical asset and a simulated virtual twin in constant communication with each other, they offer the potential to improve operational performance and reduce the cost of development, test and evaluation, production, maintenance, training and support of complicated ‘systems of systems’, with the added benefit of providing flexibility in today's rapidly evolving threat environment.

In the development of Digital Twins, the requirement to have validated and verified models of components which accurately reflect real world physical assets and processes as well as the aggregated systems containing them is key to their use. Despite massive advancements in compute power it nevertheless is not possible to represent everything at the highest level of fidelity for many applications, and the alternative of having different models which are not related to each other poses significant problems as well.

This separation of models ‘at birth’ poses a threat to the implementation of viable, interconnected Digital Twins, due to cost and configuration issues, particularly in the military context where the sharing of computer models, simulations and physical platforms with Allies and with Multi-Domain Integration across Land, Air, Sea and Space is increasingly important to sustain a global force able to flexibly adapt to changing situations.

In the context of a 9 nation research task being undertaken by the NATO Modelling and Simulation Group and elsewhere, this paper analyses the lessons of the past as they apply to model abstraction, with an analysis of potential solutions including the use of Artificial Intelligence techniques which will enhance the development of interconnected, interoperable and useful Digital Twins for the military end user.

Keywords
EMERGING TECHNOLOGIES, INTEROPERABILITY, M&S, MODELING, SIMULATIONS

Additional Keywords
Digital Twins, Artificial Intelligence, Model, Abstraction