Modeling and resource scheduling of real-time unsplittable data transfers

Abstract

Real-time communication where the timely delivery of the data transfer requests needs to be guaranteed is essential for several applications. This work formally introduces the Real-Time Unsplittable Data Dissemination Problem (RTU/DDP), which is a generalization of the unsplittable flow problem. RTU/DDP problem is proved to be NP-hard. Therefore, heuristic approaches are required to acquire good solutions to the problem. The problem is divided into two sub-problems: path selection and request packing. Each of these sub-problems is formally defined and heuristic algorithms are proposed for both sub-problems. MinMin/FPF, Edge Disjoint MinMin/FPF, MinCon/FPF, and LFL-MinCon/FPF heuristics are proposed for the path selection subproblem. MNOFF and MOFF heuristics are introduced for the latter subproblem. The performances of these algorithms are compared with a genetic algorithm solution proposed in this study and a heuristic from the literature. The results and discussions of the comparisons among the performances of the proposed heuristics are presented