Search

The multi-disciplinary nature of the design of cyber-physical systems makes it hard to gain insight in the system behaviour early in the design process. Our aim is to allow the designers to analyse the integration of system components as well as the behaviour of the complete system in an early stage. This is achieved by creating abstract component models and refining them throughout the design process. After every refinement cycle, the models can be co-simulated to analyse the behaviour of the system, supporting design decisions. The co-simulation is created based on existing standards such as HLA and FMI and uses a domain-specific language to construct a co-simulation automatically. This approach is illustrated using a case study which resembles a confidential industrial case.

The inherent multi-disciplinary nature of cyber-physical systems makes it difficult to get early insight in key system properties and trade-offs that have to be made. Our aim is to support system architects of such systems by facilitating the co-simulation of models from different disciplines and design space exploration. This has been achieved by defining a domain-specific language called CoHLA which allows a high-level description of a system architecture and simulation parameters to be specified. A generator has been implemented that generates a co-simulation of component models using an implementation of the HLA standard. Component models that adhere to the FMI standard can be incorporated easily. Moreover, CoHLA includes primitives to express design space parameters and metrics; this information is used to generate tooling for automated design space exploration.

The construction of a co-simulation for large cyber-physical systems can be very time consuming. We have defined a domain specific language called CoHLA that facilitates this construction based on the standards FMI and HLA. Scalability of this approach is investigated by the application to Internet of Things (IoT) systems. Because of the repetitive nature of these systems, we developed a separate domain specific language that allows the user to describe the system and easily generate a co-simulation definition for CoHLA. Additionally, we extended CoHLA to speed up the co-simulation execution by distributing the simulation across multiple nodes. This method also allows the co-simulation to be executed in the cloud easily.

The development of cyber-physical systems (CPSs) is a multi-disciplinary process. A model-based approach during the design of a system is important for making design decisions during the exploration of alternatives. However, all disciplines use different modelling tools and techniques, which makes the integration of these models difficult and time-consuming. The use of the High Level Architecture (HLA) simplifies this problem, but still requires quite an effort to implement. Our work focuses on minimising the effort required to construct co-simulations. We have created a Domain Specific Language (DSL) to define a system design consisting of different types of models. We demonstrate how this DSL can be used to experiment with alternative designs of the system quickly. The DSL allows us to build virtual prototypes of CPSs without the large overhead of constructing the co-simulation.