Study On Multi-level Composability Theory Of Simulation Models And Approach To Composable Simulation Integration

The research area of Modeling and Simulation (M&S) has moved to the complex huge systems comprised of many sub-systems, whose structure and behavior are keeping evolving. The ability to quickly respond to such evolvements for keeping valid and sound to the real systems is required to avoid developing new simulation systems from scratch when the requirement, structure and behavior of them has changed. The way to build simulation systems like the industrialized making, in which the components were assembled flexibly and agilely to form different target systems is urgently required to reduce the cost and risk of developing a new simulation systems. In this situation, composable simulation was proposed, as a new thinking and paradigm of simulation development, to tackling these current problems, and it has been the research hotspot of M&S of complex systems.The issue about composing and re-composing of simulation models is focus of composable simulation. But current approaches to composable simulation are mostly concerned about the implement details. Without the theory of composable simulation, the composition of simulation models and analysis of model composablity are still very hard. How to improve composability of simulation models has been the key problem and urgently required in the composable simulation.This dissertation makes the general demonstratiion of satellites navigation systems as the application background. To improving composability of simulation models, this dissertation expands its key contents to several correlative directions focusing on simulation model representation and analysis of composability. The original work and the main contributions of this dissertation include:(1) Core conceptual problems of composable simulation are systematically studied. Focusing on multi-level representation of simulation models, the Composable Simulation Conceptual Framework(CSCF) is proposed, in which several kinds of models entities, process entities, and the relationships between these entities are definitely defined and five composition patterns are provided to assemble models. The Reference Model of Composability of simulation model(RMC) was established, which differentiates related concepts such as interoperability, reusability ,etc. The analysis framework of composability is proposed on the concept of composition correctness.(2) Syntactical model representation and composability analysis based on type theory are systematically studied. Firstly four syntactical composition mismatching problems which roots in type mismatching are presented, then Simulation Model Definition Languages(SMDL) was extended with a SMDL simple Type System(STS). Based on the type judgement and type inferring rules, three type matching relations, including exact match, subtype match and the relaxed match, are formally defined. The formal methods of analyzing compatibility and substitutability of models, as well as the key algorithms, are established on these type matching. (3) Semantic representation of simulation models and composability analysis based on Hoare logic are studied. The classification framework of model semantics shows that explicit, formal semantic description is the keys to prompt semantic model composability. The approach to semantic description is proposed on Pre-Post assertions to form the composition reasoning rules. Three compatibility rules and six substitutability rules are constructed to analyze the composability by checking the logical relations between the pre-and post predicates, and the algorithms are provided for these rules.(4) Pragmatic composability concerns about how to use model and what's the behavior of model, i.e. model dynamic semantic, which represents its interactions with environment or other models. The formal semantic of DEVS specification is defined, and further the behavioral equivalence checking rules of DEVS models based on bi-simulation is provided. The Coupling DEVS specification essentially concerns about syntactical composability between ports, so DEVS is extended by Role-Protocol-Collaboration(RPC) framework to form RPC-DEVS specification. In RPC-DEVS, the roles and its protocols describe the detailed behavior of DEVS port, and the binding of roles characterizes the behavior of the composition of DEVS ports. The approach to analyzing dynamic semantic composability and the algorithms are established on the compatibility and substitutability between roles.(5) In order to effectively support multi-level modeling and simulation integration, the Model driven Composable Simulation Integration Framework (MCSIF) is proposed. MCSIF framework supports rigorous model definition at syntactic, semantic and pragmatic level, and further prompts model composability and reusability. In MCSIF, the Precise Modeling Framework(PMF) was constructed by using Meta-modeling Languages Families (MLF), and the meta-model of MSSD, SCM and REM are designed based on PMF for multi-level modeling. The mapping relationship between these meta-models is studied, which assures the consistency of multi-level model descriptions, and plays the calibration role in multi-level composability analyzing.(6) According to requirements of model composability and simulation integration in general simulation of satellite navigation system, the CSimTools prototype was design and implemented. CSimTools provides series of tools to supports multi-level modeling, composing and simulation integration, and provides a uniform schedule mechanism to support simulation execution. The theory and approach of this dissertation is demonstrated by an example of satellite navigation Service Volume Simulation development.The research topics of this dissertation belong to the fundamental M&S theory category. Above work and contributions can advance the research of simulation methodology, and promote the interoperability and reusability of models in general simulation of satellite navigation system, and can direct the integration of other complex system simulations.