Some comments on single-server queuing methods and some new results

Abstract

1. Introduction. Over the past few years, the theory of single-server queues has grown substantially. Much of the effort expended has been applied to the study of transient multivariate queuing processes in continuous time, i.e. to the study of joint time-dependent distributions of two or more system variables given some known initial system configuration. A variety of methods have been employed in these studies to accommodate general interarrival and service time distributions, among them Erlang's method of fictitious phases, the method of supplementary variables, the extended imbedded chain method, and the semi-Markov method. In the first part of this paper an informal review of these transient studies and of the methods employed is given.