Tight bounds for double coverage against weak adversaries

N. Bansal, M. Elias, L. Jez, G. Koumoutsos, K. Pruhs

Research output: Contribution to journalArticleAcademicpeer-review

7 Citations (Scopus)
60 Downloads (Pure)

Abstract

We study the Double Coverage (DC) algorithm for the k-server problem in tree metrics in the (h, k)-setting, i.e., when DC with k servers is compared against an offline optimum algorithm with h ≤ k servers. It is well-known that in such metric spaces DC is k-competitive (and thus optimal) for h = k. We prove that even if k > h the competitive ratio of DC does not improve; in fact, it increases slightly as k grows, tending to h + 1. Specifically, we give matching upper and lower bounds of k(h+1)k+1 on the competitive ratio of DC on any tree metric.

Original languageEnglish
Pages (from-to)349-365
Number of pages17
JournalTheory of Computing Systems
Volume62
Issue number2
DOIs
Publication statusPublished - 1 Feb 2018

Keywords

  • Double coverage
  • Resource augmentation
  • Weak adversaries
  • k-server

Fingerprint

Dive into the research topics of 'Tight bounds for double coverage against weak adversaries'. Together they form a unique fingerprint.

Cite this