Throughput Maximization for Periodic Packet Routing on Trees and Grids
In the periodic packet routing problem a number of tasks periodically create packets which have to be transported through a network. Due to capacity constraints on the edges, it might not be possible to find a schedule which delivers all packets of all tasks in a feasible way. In this case one aims to find a feasible schedule for as many tasks as possible, or, if weights on the tasks are given, for a subset of tasks of maximal weight. In this paper we investigate this problem on trees and grids with row-column paths. We distinguish between direct schedules (i.e., schedules in which each packet is delayed only in its start vertex) and not necessarily direct schedules. For these settings we present constant factor approximation algorithms, separately for the weighted and the cardinality case.
Our results combine discrete optimization with real-time scheduling. We use new techniques which are specially designed for our problem as well as novel adaptions of existing methods.
KeywordsGreedy Algorithm Throughput Maximization Feasible Schedule Start Vertex Destination Vertex
Unable to display preview. Download preview PDF.
- 11.Peis, B., Stiller, S., Wiese, A.: The periodic packet routing problem. Technical Report 008-2010, Technische Universität Berlin (April 2010)Google Scholar
- 12.Peis, B., Wiese, A.: Throughput maximization for periodic packet scheduling. Technical Report 013-2010, Technische Universität Berlin (June 2010)Google Scholar
- 15.Srinivasan, A.: Improved approximations for edge-disjoint paths, unsplittable flow, and related routing problems. In: Proceedings of the 38th Annual Symposium on Foundations of Computer Science (FOCS), pp. 416–425 (1997)Google Scholar