Context-Driven Testing For Composability Of Simulation Model

In order to enhance the collaborative design of complex products and reinforce the information sharing and interoperation among multi-domain, heterogeneous and distributed simulation systems, cooperative efforts of different domains and organizations are demanded in modeling and simulation (M&S), and resources reuse and composition are emphasized to meet the requirements of multidisciplinary cooperative simulation and multi-system joint simulation.Composable simulation has raised more and more attentions in the field of M&S, whose research focus is the composability of simulation models. It is not only the evaluation measure of composition validity, but also an important design principle of the development of new simulation system. Accordingly, how to evaluate and improve the composability of simulation models is the key of composable simulation.Because of the dynamic characteristic of simulation models and the concurrent characteristic of their composition, wrong behaviors that are induced by bugs are always non-recurable and non-divinable. Moreover, the reuse-oriented modeling attracts more attention than the classical objective-driven modeling, so whether simulation models need to re-evaluated in different contexts turns to be a crucial problem. Therefore, in order to comfirm whether simulation models and their composition can meet the need of simulation application or not, the information related to the use of simulation models should be described explicitly, which can provide the evaluation basis for the composability.The dissertation chooses mission level simulation as its application background. By analyzing the task space of mission, entities that execute tasks and the relationship between entities and tasks, it makes deep insight into the dynamic composability of simulation models from the testing point of view. The basic idea of the dissertation is as following: Firstly, a composition-based M&S framework is constructed. Secondly, the concept and connotation of context and the relationship between the context and the composability are described. Thirdly, the criteria that are used to evaluate the composability are presented. Finally, the context-driven simulation model testing methodology to support evaluation of the dynamic composability is put forward, which includes system-level, composition-level and unit-level testing. The main work of the dissertation is as following:(1) The context of system simulation is explained from the origin of the word context. Then by uniting the development process of composable simulation with the one of traditional simulation, the dissertation constructs a composition-based M&S framework that extends a reuse-oriented one. Finally, based on the conception of the application context and the model context and the relationship between the two contexts and the composability, the criteria that are used to evaluate the composability are presented.(2) By analyzing misconceptions, the dissertation defines what the simulation model testing is, analyzes how the criteria direct it and what problems still exist, and strengthenes the idea that it supports evaluation of the composability. Then based on analyzing the level of entities, the composition-based development process and the relationships among entities, the context of mission level simulation is explained. Finally, the dissertation presents the context-driven testing methodology from structure, process and execution aspects, which guides hierarchical testing activities.(3) Aiming at the system-level testing, by distinguishing the concept of task, the dissertation defines the task model and summarizes the task-modeling methods. Then it analyzes the temporal and spatial properties of tasks, and presents the decomposition principles of task space that are used to determinate and limit simulation models and their behaviors under test. Finally, the description method and the checking arithmetic of task constraints are presented based on the ECA rules.(4) Aiming at the composition-level testing, by analyzing the interactive modes and summarizing interaction-modeling methods, the dissertation defines what the interaction model is and presents the timing generation of atom events. On the basis of summarizing the description and checking of composite events, the dissertation extends the property patterns of interactive behaviors by temporal logic, describes the event model with determinate and indeterminate characteristics completely, and then discusses its the inference rules and the consistency. Finally, the specification and the checking arithmetics of the property patterns based on the QREs are presented.(5) Aiming at the unit-level testing, by analyzing the behavior characteristics of organizational relationship and state transition, the dissertation takes the state reachable set as the constraint modeling method for entity models. Then the gereration arithmetic and the checking arithmetics of the reachable set are presented, which are used to address the reason why the system-level testing or the composition-level testing cannot be satisfied.(6) From the point of the dependency relationship between testing environment and M&S environment, the dissertation designs the architecture of testing environment. By analyzing the constraint modeling tools and the execution tools, the dissertation presents the constraint specification language for different testing level, the content model of test outline and the reconstructive methods of hierarchical simulators.(7) The effectiveness of the proposed methods is demonstrated by the application case of combat simulation for multi-platform anti-ship missile.The main contributions of this dissertation include:By constructing the composition-based M&S framework, the dissertation presents the criteria that are used to evaluate the composability of simulation models. The context-driven simulation model testing methodology is presented, which not only supports evaluation of the composability, but also guides hierarchical testing activities.Based on the decomposition principles of task space and the ECA rules, the description method and the checking arithmetic of task constraints are presented.Toward composable behaviors of entities, the dissertation extends the property patterns with certain and uncertain characteristics that describe interaction constraints of simulaton models, and presents the specification and the checking arithmetics of the property patterns based on the QREs.Toward internal behaviors of entities, the generation arithmetic and the checking arithmetics of the state reachable set are presented, which are used to describe and check internal behavior constraints of entity models respectively.The dissertation designs the architecture of testing environment and presents the constraint specification language, the content model of test outline and the reconstructive methods of hierarchical simulators.The research of the dissertation belongs to the fundamental studies on the theory and method of composable simulation. The criteria and the context-driven testing methodology provide new theoretical framework and methodological set, which can promote the development of composable simulation.