| Chair | Hans Vangheluwe
School of Computer Science McGill University Tel. 514 398 44 46 |
Co-Chair | Pieter J. Mosterman
Institute of Robotics and Mechatronics DLR Oberpfaffenhofen Tel. +49 8153 282434 |
Specialized computer automated tools for each of the domains in computer aided control system design are very helpful or even indispensable to carry out the related tasks. However, because these tools hardly ever are compatible, the sharing and coordinating of information flow between project teams inevitably leads to a lot of overhead in terms of collaboration and is very error prone, inefficient, and expensive. Moreover, similar tasks may be carried out multiple times and even simultaneously. For example, compared to control design approaches, fault detection and isolation (FDI) methods often rely on different, more detailed, models of the same system, which, as a consequence, is modeled by the control design team as well as the FDI team.
Paper 1 [Pieter J. Mosterman and Hans Vangheluwe] discusses the issues of multi paradigm modeling in control design engineering. It gives an overview of the meta modeling approach that is applied in much of the work in facilitating multi paradigm modeling and contrasts this approach with the alternative of one generic unifying formalism.
Paper 2 [Johannes Ernst and Scott Washburn] addresses concurrent systems design and collaboration issues through an internet and meta-model-based approach to data management based on commercial modeling tools, called Zero-latency EngineeringTM. As a consequence, every engineer can use a best-of-class modeling tool for a component modeled in a certain paradigm, while the team can understand and view the system as a whole, built from parts in different tools using different modeling paradigms. This prevents inconsistencies at the boundaries between engineering groups, engineering software and modeling paradigms that are one of the major reasons of delayed development projects today. Furthermore, it dramatically reduces the need for face-to-face information exchange and thus enables Zero-latency Engineering.
Paper 3 [John R. James] presents an overview of modeling paradigms to analyze military operations. The specification of these systems includes many tasks (e.g., concurrent operation, planning, and scripting) with particular behaviors (e.g., discrete state behavior, continuous behavior evolution, and resource assignment), and a number of dedicated formalisms and modeling paradigms are used to conveniently specify one or more such behaviors to allow the system designer sufficient flexibility and elegance. These paradigms may include differential equation continuous dynamics models to capture the physical behavior of agents combined with specification formalisms to model certain scenarios and discrete event behavior. This requires one to deal with the issue of combining these modeling paradigms and formalisms in a heterogeneous representation amenable to analysis.
Specification software such as CASE (Computer Aided Systems Engineering) tools are specifically designed to handle the complexity of present day complex systems. These tools are required to comprise many formalisms, e.g., data flow diagrams, state charts, and block diagrams. This leads to two complications: (i) a tool developer needs to be sufficiently flexible in its product offering (especially considering the widely proliferated industrial custom to use proprietary formalisms) and (ii) consistency between these formalisms needs to be ensured.
Paper 4 [Eric Engstrom and Jonathan Krueger] addresses these two issues by modeling the specification formalisms at a higher level by means of a meta model specification language. This language defines the entities, relations, attributes, etc., of each of the particular formalisms. Such a meta model can then be instantiated into a dedicated specification formalism. Moreover, because specified in the same meta language, consistency constraints between the formalisms can be captured and automatically enforced by the tool as well.
Paper 5 [Gabor Karsai, Greg Nordstrom, Akos Ledeczi and Janos Sztipanovits] discusses how, at a meta level, syntactic and semantic modeling language constraints can be specified to ensure a certain degree of model fidelity. For example, in the bond graph formalism for modeling physical systems, conservation of energy is an inherent constraint of each of the basic elements. Therefore, constraints captured at the metalevel are used to ensure all bond graph models built with these elements satisfy this physical law. Another example is the use of very domain-specific rules. For example, at the metalevel an aircraft can be specified to consist of two wings, one body, two to four engines, and two computer control systems. Again, the control systems may be best specified in detail by a discrete event modeling formalism such as statecharts, whereas the engines are more suitable for a component based decomposition, resulting in heterogeneous refinement and the need for an integrated modeling capability. Examples such as this lead to a discussion of using metamodel-based modeling environments to model large-scale, heterogeneous computer-based systems.
Paper 6 [Jie Liu and Edward A. Lee] discusses how each of these modeling formalisms may have a different model of computation. For example, discrete event models have a global notion of time, while finite state machine is an untimed model. In order to associate execution semantics with heterogeneous models, a component-based framework is proposed. The framework uses hierarchical composition to hide the implementation details of one component from other components, and keeps the components at the same level of hierarchy interacting in the same way. The framework is easy to scale up and enhances the possibility of design reuse.