Functional mockup Interface extension with support for Discrete Event Languages

Starting date
February 1, 2017
Duration (months)
24
Departments
Computer Science, Department of Engineering for Innovation Medicine
Managers or local contacts
Fummi Franco

Cyber-Physical Systems (CPS) mangle aspects of the physical world (usually continuous) with cyber reactive systems (usually discrete). Nowadays, each domain is covered by dedicated tools offering simulation capabilities (e.g., ANSYS Fluent for fluid dynamics or Scade for cyber controllers). In order to understand the whole system, co-simulation, where the different models/simulators are jointly working, is required.

The last years showed a growing interest in the Functional Mockup interface (FMI), a tool-independent industry-supported standard to co-simulate multi-domain systems. FMI provides a standardized interface for the different models and their simulators. Then, an integrator is in charge of manually writing the so called master algorithm, which is used to convey data and ensure time consistency between the different models/simulators at specific discrete time points. This does not fit well cyber models for digital hardware or network where a discrete-event simulator is used. The current solution consists in forcing simulators to work synchronously at the speed of the slowest one thus limiting simulation speed. Finally, even if FMI is currently supported by 89 tools from 16 companies and research institutions, common hardware description languages (e.g., VHDL, Verilog and SystemC) are not supported.

To address the highlighted problems the project has the following objectives: 1) a methodology to import HDL models into FMI co-simulation; 2) use of models information to automatically generate an efficient and correct by construction master algorithm; 3) identify limits of current FMI interface to propose a revision of the standard. The project outcomes will be demonstrated on a control scenario in which the cyber part is a virtual platform executing the control policy while the plant is modeled by an industry-level tool such as Matlab/Simulink and Scilab/Xcos.

Sponsors:

Funds: assigned and managed by the department

Project participants

Franco Fummi
Full Professor

Activities

Research facilities

Share