A Framework for Component Based Modelling and Simulation using BOMs and Semantic Web Technology

University dissertation from Stockholm : KTH

Abstract: Modelling and Simulation (M&S) is a multi-disciplinary field that is widely used in various domains. It provides a means to study complex systems before actual physical prototyping and helps lowering, amongst others, manufacturing and training costs. However, as M&S gains more popularity, the demand on reducing time and resource costs associated with development and validation of simulation models has also increased. Composing simulation models of reusable and validated simulation components is one approach for addressing the above demand. This approach, which is still an open research issue in M&S, requires a composition process that is able to support a modeller with discovery and identification of components as well as giving feedback on feasibility of a composition. Combining components in order to build new simulations raise the non-trivial issue of composability.Composability has been defined as the capability to select and assemble reusable simulation components in various combinations into simulation systems to meet user requirements. There are three main types of composability, syntactic, semantic and pragmatic. Syntactic composability is concerned with the compatibility of implementation details, such as parameter passing mechanisms, external data accesses, and timing mechanisms. It is the question of whether a set of components can be combined. Semantic composability, on the other hand, is concerned with the validity of the composition, and whether the composed simulation is meaningful. Pragmatic composability is yet another type which is concerned with the context of the simulation, and whether the composed simulation meets the intended purpose of the modeller. Of these three types syntactic composability is easiest to accomplish and some significant progresses on this issue have been reported in the literature. Semantic and pragmatic composability are much harder to achieve and has inspired many researchers to conduct both theoretical and experimental research.The Base Object Model (BOM) is a new concept identified within M&S community as a potential facilitator for providing reusable model components for the rapid construction and modification of simulations. Although BOMs exhibit good capabilities for reuse and composability they lack the required semantic information for semantic matching and composition. There is little support for defining concepts and terms in order to avoid ambiguity, and there is no method for matching behaviour of conceptual models (i.e., state machines of the components), which is required for reasoning about the validity of BOM compositions.In this work we have developed a framework for component-based model development that supports both syntactic and semantic composability of simulation models by extending the BOM concept using ontologies, Semantic Web and Web Services technologies, and developing a rule-based method for reasoning about BOM compositions. The issue of pragmatic composability has not been the focus of this work, and it has only been partly addressed. The framework utilises intelligent agents to perform discovery and composition of components, according to the modeller needs. It includes a collaborative environment, a semantic distributed repository and an execution environment to support model development and execution process.The basic assumption of this work is that semantic composability should be achieved at conceptual level. Through precise definition and specification of components’ semantic and syntax one can capture the basic requirements for matching and semantically meaningful composition of those components. This requires a common methodology for specification of simulation components. The specification methodology consists of meta-models describing simulation components at different levels. In order to enable automatic matching of meta-models they are formalized and structured using Semantic Web technology in OWL (Web Ontology Language). Hence, the models are based on ontologies to avoid misunderstanding and to provide unambiguous definitions as a basis for reasoning about syntactic and semantic validity of compositions.