Distributed approximation of k-service assignment

Magnús M. Halldórsson, Sven Köhler, Dror Rawitz

Research output: Contribution to journalArticlepeer-review

1 Scopus citations


We consider the k-Service Assignment problem (k-SA). The input consists of a network that contains servers and clients. Associated with each client is a demand and a profit. In addition, each client c has a service requirement[InlineEquation not available: see fulltext.], where κ(c) is a positive integer. A client c is satisfied only if its demand is handled by exactly κ(c) neighboring servers. The objective is to maximize the total profit of satisfied clients, while obeying the given capacity limits of the servers. We focus here on the more challenging case of hard constraints, where no profit is granted for partially satisfied clients. This models, e.g., when a client wants, for reasons of fault tolerance, a file to be stored at κ(c) or more nearby servers. Other motivations from the literature include resource allocation in 4G cellular networks and machine scheduling on related machines with assignment restrictions. In the r-restricted version of k-SA, no client requires more than an r-fraction of the capacity of any adjacent server. We present a (centralized) polynomial-time [InlineEquation not available: see fulltext.]-approximation algorithm for r-restricted k-SA. A variant of this algorithm achieves an approximation ratio of [InlineEquation not available: see fulltext.] when given a resource augmentation factor of 1 + r. We use the latter result to present a [InlineEquation not available: see fulltext.]-approximation algorithm for k-SA. In the distributed setting, we present: (i) a [InlineEquation not available: see fulltext.]-approximation algorithm for r-restricted k-SA, (ii) a [InlineEquation not available: see fulltext.]-approximation algorithm that uses a resource augmentation factor of 1 + r for r-restricted k-SA, both for any constant ε> 0 , and (iii) an [InlineEquation not available: see fulltext.]-approximation algorithm for k-SA (in expectation). The three distributed algorithms run in O(k2ε- 2log 3n) synchronous rounds (with high probability). In particular, this yields the first distributed [InlineEquation not available: see fulltext.]-approximation of 1 -SA.

Original languageEnglish
Pages (from-to)27-40
Number of pages14
JournalDistributed Computing
Issue number1
StatePublished - 12 Feb 2019

Bibliographical note

Funding Information:
A preliminary version was presented at the 19th International Conference on Principles of Distributed Systems (OPODIS) 2015. M. M. Halldórsson supported in part by the Icelandic Research Fund (Grant nos. 120032011 and 152679-051). S. Köhler supported in part by the Sustainability Center Freiburg, a cooperation of the Fraunhofer Society and the University of Freiburg, supported by grants from the Baden-Württemberg Ministry of Economics and the Baden-Württemberg Ministry of Science, Research and the Arts. D. Rawitz supported in part by a grant from the Israeli Ministry of Science, Technology, and Space (Grant no. 3-10996) and by the Israel Science Foundation (Grant no. 497/14).

Publisher Copyright:
© 2017, Springer-Verlag GmbH Germany, part of Springer Nature.


Dive into the research topics of 'Distributed approximation of k-service assignment'. Together they form a unique fingerprint.

Cite this