Automatic Parallelization and Scheduling Approches for Co-simulation of Numerical Models on Multi-core Processors

Abstract : When designing complex cyber-physical systems, engineers have to integrate numerical models from different modeling environments in order to simulate the whole system and estimate its global performances. If some parts of the system are physically available, it is possible to connect these parts to the simulation in a Hardware-in-the-Loop (HiL) approach. In this case, the simulation has to be performed in real-time where models execution consists in periodically reacting to the real (physically available) components and providing periodic output updates. The increase of requirements on the simulation accuracy and its validity domain requires more complex models. Using such models, it becomes hard to ensure fast or real-time execution without using multiprocessor architectures. FMI (Functional Mocked-up Interface), an increasingly common standard for model exchange and co-simulation, offers new opportunities for multi-core execution of numerical models. One goal of this thesis is the extraction of potential parallelism in a set of interconnected multi-rate models. We build on the RCOSIM approach that has been previously developed at IFP Energies nouvelles and which allows the parallelization of FMI models on multi-core processors. It is based on representing the co-simulation by a dependence graph model. In the first part of the thesis, improvements have been proposed to overcome the limitations of RCOSIM. In particular, we propose new algorithms in order to allow handling models that exchange data at different rates and schedule them on multi-core processors. Also, the improvements allow handling specific constraints such as mutual exclusion and real-time constraints. Second, we propose algorithms for the allocation and non preemptive scheduling of the dependence graphs, taking into account their real-time, data dependence and allocation constraints. These algorithms aim at accelerating the execution of the co-simulation or ensuring its real-time execution in a HiL approach. The proposed solutions have been tested on randomly generated dependence graphs and validated against an industrial use case which is an internal combustion engine co-simulation. This thesis is part of a joint action IFP Energies nouvelles-Inria, in which Inria brings its real-time systems experience to the numerical simulation challenges of IFP Energies nouvelles.
Complete list of metadatas

Cited literature [89 references]  Display  Hide  Download

https://hal.inria.fr/tel-01895280
Contributor : Brigitte Briot <>
Submitted on : Monday, October 15, 2018 - 8:40:39 AM
Last modification on : Tuesday, January 22, 2019 - 3:51:44 PM
Long-term archiving on : Wednesday, January 16, 2019 - 1:09:17 PM

File

THESE.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : tel-01895280, version 1

Citation

Salah Eddine Saidi. Automatic Parallelization and Scheduling Approches for Co-simulation of Numerical Models on Multi-core Processors. Embedded Systems. Université Sorbonne, 2018. English. ⟨tel-01895280⟩

Share

Metrics

Record views

176

Files downloads

141