Reliability Analysis of a Controllable Queueing System with Two Heterogeneous Servers Subject to Failures

Description

To make modern communication systems superior in performance and reliability to the previous generation systems they can be supplied with heterogeneous communication links. Such links can differ in availability, link data throughputs, power consumption and reliability characteristics. To model the dynamic behaviour of the links with different properties a queueing system with non-reliable heterogeneous servers can be used. While the first steps in the performance analysis of controllable heterogeneous queueing systems have already been developed for completely reliable servers, a missing link to an applicability of these models is reliability analysis of such queues with servers subject to failures. In this paper we use a matrix transform based method to evaluate reliability measures such as reliability function and mean time to the first failure for each server separately and for the total service facility under the fixed threshold allocation control policy. The reliability functions are obtained in terms of the Laplace transform and numerical inversion algorithm is used to get the time dependent functions. Additionally a new discrete reliability metric which can be treated as a discrete counterpart to the distribution of the time to failure is introduced. This function specifies the distribution of the number of repairs of the server until a complete failure of the service facility occurs. Some numerical examples illustrate the efficiency of the proposed algorithms.

Period	19 Jul 2016
Event title	European Conference on Queueing Theory
Event type	Conference
Location	FranceShow on map