Section: New Results

Multiprocessor Real-Time Scheduling

Participants : Salah-Eddine Saidi, Yves Sorel.

During the third year of the PhD thesis of Salah Eddine Saidi, we focused on two aspects. First, we finalized our work on the parallelization on multi-core processors of FMI-based co-simulation of numerical models in order to accelerate its execution. Our approach, based on the transformation of FMU graphs into operation graphs which reveal more parallelism, comprises the following two steps: first acyclic orientation necessary for avoiding that some operations of a same model are executed in parallel and second multi-core offline scheduling of operations [5]. We proposed exact algorithms based on ILP (Integer Linear Programming) and heuristics for performing the acyclic orientation and the multi-core scheduling. Also, we proposed a random generator of synthetic co-simulations. Using these generated co-simulations, we compared the performances of the heuristics and the ILP-based exact algorithm for both the acyclic orientation and the scheduling in terms of execution time and quality of the obtained solution. Tests have been carried out for different sizes of co-simulation and different numbers of cores. Moreover, we compared the performance of our offline approach with an online scheduling approach based on the Intel TBB runtime library. This comparison was acheived by applying both approaches on an industrial use case which consists in a co-simulation of a four cylinder spark ignition engine. The various tests that we performed showed the efficieny of our proposed heuristics. Second, we focused on the parallelization of FMI-based co-simulation under real-time constraints. In particular, we were interested in HiL (Hardware-in-the-Loop) co-simulation where a part of the co-simulation is replaced by its real counterpart that is physically available. The real and simulated parts have to exchange data during the execution of the co-simulation under real-time constraints. In other words, the inputs and ouputs of the real part are sampled periodically, sending and receving data to and from the simulated part. This periodic data exchange defines a set of real-time constraints to be satisfied by the simulated part. We proposed a method for defining these real-time constraints and propagating them to all the operations of the co-simulation (simulated part). In our ongoing work, we are focusing on multi-core scheduling of FMI-based co-simulation under real-time constraints. More precisely, we are working on a heuristic and an ILP-based algorithm that will enable the execution of the co-simulation on a multi-core processor while ensuring the defined real-time constraints are respected.