文献原文-机器服务模型的性能分析.docxVIP

Queueing TheoryQueueing Theory plays a vital part in almost all investigations of service facilities. This chapter aims to provide an introduction to the highlights of queueing theory. Special attention is paid to multiserver systems, matrix analytic techniques and ellects of customer impatience and retrial behaviour. This selection reffects the author’s preference, which is biased in the direction of telephony and call centre applications. It has been tried to achieve a certain balance between theoretical rigor and practical value by providing applicable results for a wide range of queueing problems. Whenever no workable exact solutions are available, reasonable approximations are supplied instead.1.1 Introduction to Queueing Theory1.1.1 HistoryQueueing theory as part of probability theory has evolved from classic tele-traffic engineering in the last decades. In 1909 A.K. Erlang, a Danish teletraffic engineer published a paper called The Theory of Probabilities and Tele-phone Conversations. In the early 1920s he developed the famous Erlang model to evaluate loss probabilities of multi-channel point-to-point conversations. The Erlang model was extended to allow for calculation in finite source input situations by T.O. Engset several years later leading to the Engset model. In 1951 D.G. Kendall published his work about embedded Markovchains, which is the base for the calculation of queueing systems under fairly general input conditions. He also defined a naming convention for queueing systems which is still used. Nearly at the sametime D.V. Lindley developed an equation allowing for results of a queueing system under fairly general in-put and service conditions. In 1957 J.R. Jackson started the investigation of networked queues thus leading to so called queueing network models. With the appearance of computers and computer networks, queueing systems and queueing networks have been identified as a powerful analysis and design tool for various applications.1.1.2 A