The project defines a DP-Fair scheduling of Periodic Parallel Tasks. We consider identical processors and Gang Tasks that use simultaneously several processors. The schedule is defined by a pattern that will be used in every slice in the schedule that is delimited by two subsequent task deadlines. Two polynomial-time methods are proposed to define the pattern : an optimal linear programming based approach and an heuristic with a performance guarantee of (2-1/m) on resource augmentation (i.e., processor speedup factor). Both methods are implemented in MATLAB and compared through intensive numerical experiments. All source codes and results are provided.


Manager : Joel GOOSSENS, Pascal RICHARD