Improved bounds for queues with delayed arrivals (Q1275367)

This paper deals with a single server queueing system in which the time between arrivals is the sum of a constant \(p\) and an exponentially distributed random variable \(X\). The inter arrival times and the service times constitute two independent families of i.i.d. random variables. For this system, improved bounds on the customer waiting times are developed using some of the results from the article by \textit{D. J. Daley}, \textit{A. Ya. Kreinin} and \textit{C. D. Trengove} [in: Queueing and related models. Oxf. Stat. Sci. Ser. 9, 117-223 (1992; Zbl 0784.60086)], which contains a comprehensive review of a wide variety of bounds. For moderate to heavy traffic, a simple improved upper bound, which only uses the first two moments of the service time distribution, is obtained. The approach is extended to obtain bounds for two other types of delayed arrival queues. For very light traffic, asymptotically tight bounds are obtained using more information about the service time distribution. The bounds are compared for a queue with exponential service times, for which exact results are available.