A coterie‐based mutual exclusion algorithm for distributed systems allowing multiple process failures at arbitrary time

Abstract

AbstractIn this article, we propose a coterie‐based mutual exclusion algorithm for distributed systems allowing process failures at arbitrary time. The proposed algorithm is an extension of Maekawa's algorithm, which is a representative conventional method for distributed mutual exclusion based on coterie. The proposed algorithm realizes mutual exclusion by detecting process failures using timers, updating the coterie dynamically, and introducing the system to a state without failure. It has the following advantages. First, it inherits all advantages of Maekawa's algorithm, because it behaves in the same way as Maekawa's algorithm when there is no process failure. Also, it allows k(<N; N is the number of processes) process failures at arbitrary time, so the assumption on failure is close to real systems. Furthermore, it is assured that the updated coterie is an ND (nondominated) coterie if the initial coterie is an ND coterie. We also describe an extension of this algorithm to the case where failed processes are allowed to return to normal states. © 2002 Wiley Periodicals, Inc. Syst Comp Jpn, 33(12): 87–96, 2002; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/scj.1169