Mean waiting time approximations in cyclic-service systems with exhaustive limited service policy

Abstract

In this paper, we consider a cyclic-service system with exhaustive limited service policy (or k-limited). Under this policy, the server may serve up to a maximum of k customers during his visit at the queue. The major result here is a new approximation procedure to compute the individual mean waiting times. This approximation is based on the pseudo-conservation law and on the concept of conditional cycle times. Extensive numerical examples show that this new approximation is more accurate and robust than the two other heuristic models reported in the literature. Furthermore, its accuracy improves significantly as the number of queues increases and its performance does not seem to be affected by k. The overall performance characteristic, thus, allows this approximation to model token-passing network protocols.