A token-based distributed algorithm for the generalized resource allocation problem

  • Authors:
  • Hirotsugu Kakugawa;Sayaka Kamei

  • Affiliations:
  • Osaka University, Suita, Osaka, Japan;Hiroshima University, Higashihiroshima, Hiroshima, Japan

  • Venue:
  • OPODIS'10 Proceedings of the 14th international conference on Principles of distributed systems
  • Year:
  • 2010

Quantified Score

Hi-index 0.00

Visualization

Abstract

The resource allocation problem is one of the fundamental problems for conflict resolution in distributed systems. In this paper, we consider the (n, m, k, d) -resource allocation problem, proposed by Joung [Joung, Distributed Computing (2010)], which is a generalization of the k-mutual exclusion problem and the group mutual exclusion problem. We propose a fully distributed solution based on tokens for the (n, m, k, d)-resource allocation problem for asynchronous message passing distributed systems. Previous works allow a process to request only one resource at a time. In our algorithm, the amount of resources that a process may request is arbitrary. The algorithm is designed based on leader-follower scheme, and permission to access resources is granted by tokens. The first process that requests resources becomes the leader, and then, other processes become followers and they are granted their requests by the leader. The message complexity of the proposed algorithm is bounded by (O(|Q|) in the worst case and O(1) in the best case, where |Q| is the size of quorums of a coterie that the algorithm uses.