Admission control with advance reservations in simple networks

Research output: Contribution to journalArticlepeer-review

2 Scopus citations

Abstract

In the admission control problem we are given a network and a set of connection requests, each of which is associated with a path, a time interval, a bandwidth requirement, and a weight. A feasible schedule is a set of connection requests such that at any given time, the total bandwidth requirement on every link in the network is at most 1. Our goal is to find a feasible schedule with maximum total weight. We consider the admission control problem in two simple topologies: the line and the tree. We present a 12c-approximation algorithm for the line topology, where c is the maximum number of requests on a link at some time instance. This result implies a 12c-approximation algorithm for the rectangle packing problem, where c is the maximum number of rectangles that cover simultaneously a point in the plane. We also present an O (log t)-approximation algorithm for the tree topology, where t is the size of the tree. We consider the loss minimization version of the admission control problem in which the goal is to minimize the weight of unscheduled requests. We present a c-approximation algorithm for loss minimization problem in the tree topology. This result is based on an approximation algorithm for a generalization of set cover, in which each element has a covering requirement, and each set has a covering potential. The approximation ratio of this algorithm is Δ, where Δ is the maximum number of sets that contain the same element.

Original languageEnglish
Pages (from-to)491-500
Number of pages10
JournalJournal of Discrete Algorithms
Volume5
Issue number3
DOIs
StatePublished - Sep 2007
Externally publishedYes

Keywords

  • Admission control
  • Approximation algorithms
  • Axis parallel rectangles
  • Local ratio
  • Scheduling

Fingerprint

Dive into the research topics of 'Admission control with advance reservations in simple networks'. Together they form a unique fingerprint.

Cite this