On Validation of Semantic Composability in Data-driven Simulation ∗ Claudia Szabo and Yong Meng Teo Department of Computer Science National University of Singapore Computing 1, 13 Computing Drive Singapore 117417 Email: claudias, teoym@comp.nus.edu.sg Abstract A simulation model composed using reusable components is semantically valid if it produces meaningful results in terms of expressed behaviors and meets the desired objective. This paper focuses on the validation of data-driven component-based modeling and simulation. In data- driven simulation applications, it is necessary to model entity behavior at higher resolution. In simulations such as military training scenarios where entity behavior changes dynamically, additional input data is required to express complex state transitions. This can significantly increase the composed model state space and presents a major challenge in simulation vali- dation. Using a component-based data-driven tactical military simulation, we propose a lay- ered and automated approach for semantic composability validation. While the expressivity of data-driven models increases the semantic equivalence of the validated model, it incurs higher validation cost. 1. Introduction Component-based simulation model development is an appealing approach to the simulation community [14] because it reduces the time and cost of developing complex simulations. Simulation composability [19] can be defined as “the capability to select and assemble simulation components in various combinations to satisfy user requirements”. Component-based frameworks that employ reused simulation components promise shorter development time and increased flexibility in meeting diverse user needs [19]. In modeling and simulation, two main levels of composability have been identified, namely syntactic composability and semantic composability [5]. In syntactic composability, components have to be properly connected and must interoperate, which assumes common communication protocols, data formats, as well as a common understanding of the time management mechanisms employed. In semantic composability, the composition must be meaningful for all components involved. Furthermore, the composed model must be valid [19]. This is because simulation models are widely used to make critical decisions and to answer “what-if” questions [3]. For example, under the current US Department of Defense policy, all models and simulations must undergo a costly verification, validation, and accreditation (VV&A) process [29]. As such, semantically valid simulation models are absolutely necessary [26], and thus component-based simulation frameworks must provide for the validation of semantic composability or at least for the context in which semantic composability can be achieved [14]. ∗ A version of this paper is published in the Proceedings of 24th ACM/IEEE/SCS Workshop on Principles of Advanced and Distributed Simulation, pp. 73-80, IEEE Computer Society Press, Atlanta, USA, May 17-19, 2010.