Strong approximations for Markovian service networks

Inspired by service systems such as telephone call centers, we develop limit theorems for a large class of stochastic service network models. They are a special family of nonstationary Markov processes where parameters like arrival and service rates, routing topologies for the network, and the number of servers at a given node are all functions of time as well as the current state of the system. Included in our modeling framework are networks of M _t /M _t /n _t queues with abandonment and retrials. The asymptotic limiting regime that we explore for these networks has a natural interpretation of scaling up the number of servers in response to a similar scaling up of the arrival rate for the customers. The individual service rates, however, are not scaled. We employ the theory of strong approximations to obtain functional strong laws of large numbers and functional central limit theorems for these networks. This gives us a tractable set of network fluid and diffusion approximations. A common theme for service network models with features like many servers, priorities, or abandonment is “non-smooth” state dependence that has not been covered systematically by previous work. We prove our central limit theorems in the presence of this non-smoothness by using a new notion of derivative. This revised version was published online in June 2006 with corrections to the Cover Date.     

Estimating variance of output from cyclic exponential queueing systems

Production systems, particularly those making use of a pull production control mechanism, are well-modeled as closed queueing networks. Average throughput is clearly one important performance measure for these systems. However, many control decisions require information concerning the variability of the output process as well as throughput. Because of this, the standard deviation of the number of outputs during a specified interval is a practical performance measure for production systems.In this paper, we consider the standard deviation of the number of outputs during a time interval from a closed queueing network consisting ofM single server exponential queues. Because computing this quantity exactly is extremely cumbersome, we introduce a simple approximation that makes use of (1) known results for the variance of the time a marked job takes to complete a round trip and (2) an approximate correction term for the covariance between successive round trips. We show through comparisons with simulation that our method is quite accurate under a variety of conditions.     

Numerical methods for controlled and uncontrolled multiplexing and queueing systems

We deal with a very useful numerical method for both controlled and uncontrolled queuing and multiplexing type systems. The basic idea starts with a heavy traffic approximation, but it is shown that the results are very good even when working far from the heavy traffic regime. The underlying numerical method is a version of what is known as the Markov chain approximation method. It is a powerful methodology for controlled and uncontrolled stochastic systems, which can be approximated by diffusion or reflected diffusion type systems, and has been used with success on many other problems in stochastic control. We give a complete development of the relevant details, with an emphasis on multiplexing and particular queueing systems. The approximating process is a controlled or uncontrolled Markov chain which retains certain essential features of the original problem. This problem is generally substantially simpler than the original physical problem, and there are associated convergence theorems. The non-classical associated ergodic cost problem is derived, and put into a form such that reliable and good numerical algorithms, based on multigrid type ideas, can be used. Data for both controlled and uncontrolled problems shows the value of the method.Supported by ARO contract DAAL-03-92-G-0115, AFOSR contract F49620-92-J-0081, and DARPA contract AFOSR-91-0375.Formerly at Brown University. Supported by DARPA contract AFOSR-91-0375.     

A Markovian queueing model for ambulance offload delays

Ambulance offload delays are a growing concern for health care providers in many countries. Offload delays occur when ambulance paramedics arriving at a hospital Emergency Department (ED) cannot transfer patient care to staff in the ED immediately. This is typically caused by overcrowding in the ED. Using queueing theory, we model the interface between a regional Emergency Medical Services (EMS) provider and multiple EDs that serve both ambulance and walk-in patients. We introduce Markov chain models for the system and solve for the steady state probability distributions of queue lengths and waiting times using matrix-analytic methods. We develop several algorithms for computing performance measures for the system, particularly the offload delays for ambulance patients. Using these algorithms, we analyze several three-hospital systems and assess the impact of system resources on offload delays. In addition, simulation is used to validate model assumptions.     

A sequential technique for the control of traffic intensity in Markovian queues

A sequential parameter control technique previously introduced by the author is modified in this paper so as to make it simple in practice. The detailed procedure involving two phases, a warning phase with control limits and a testing phase using an appropriate test is illustrated for a queueing system with an embedded Markov chain. Operating characteristics of the procedure are also examined.     

Cooperation in Markovian queueing models

In this paper we study some cooperative models in Markovian queues. We stress the case of several agents agreeing to maintain a common server for their populations in which a priority scheme with preemption has been established. In this situation we propose and characterize an allocation rule for the holding costs that provides core allocations.     

Empirical Bayes estimation for queueing systems and networks

Empirical Bayes estimators are derived for standardM/M/1 queues,M/M/1 queues with state-dependent arrival and service rates, finite capacityM/M/1 queues with state-dependent rates and for open Jackson networks. The asymptotic properties of the empirical Bayes estimators are derived both with respect to the conditional distribution of the observations given the parameters, and with respect to the joint distribution of the observations and the parameters.     

Approximate analysis of arbitrary configurations of open queueing networks with blocking

An algorithm for analyzing approximately open exponential queueing networks with blocking is presented. The algorithm decomposes a queueing network with blocking into individual queues with revised capacity, and revised arrival and service processes. These individual queues are then analyzed in isolation. Numerical experience with this algorithm is reported for three-node and four-node queueing networks. The approximate results obtained were compared against exact numerical data, and they seem to have an acceptable error level.Supported in part by a grant from CAIP Center, Rutgers University.Supported in part by the National Science Foundation under Grant DCR-85-02540.     

排队系统中的泰勒展开方法

综述了排队系统中的泰勒展开方法。它由Gong和Hu在1990s首次提出,并在最近几年里有了一些新的发展。首先,通过GI/GI/1队列的简单例子介绍其基本原理;其次,展示如何应用该方法分析相关性队列和离去过程;然后,阐述如何基于该方法发展排队网络近似的高阶矩方法;最后,讨论未来的几个可能研究方向。     

Optimal prices for finite capacity queueing systems

We prove a lower bound on the optimal price for a fairly large class of blocking systems with general arrival and service processes, determine optimal price expressions for M/M/1/m and M/GI/s/s systems, and investigate how optimal prices change with changes in the size of the waiting room and service capacity.     

A heavy-traffic analysis of a closed queueing system with a GI/∞ service center

This paper studies the heavy-traffic behavior of a closed system consisting of two service stations. The first station is an infinite server and the second is a single server whose service rate depends on the size of the queue at the station. We consider the regime when both the number of customers, n, and the service rate at the single-server station go to infinity while the service rate at the infinite-server station is held fixed. We show that, as n→∞, the process of the number of customers at the infinite-server station normalized by n converges in probability to a deterministic function satisfying a Volterra integral equation. The deviations of the normalized queue from its deterministic limit multiplied by √n converge in distribution to the solution of a stochastic Volterra equation. The proof uses a new approach to studying infinite-server queues in heavy traffic whose main novelty is to express the number of customers at the infinite server as a time-space integral with respect to a time-changed sequential empirical process. This gives a new insight into the structure of the limit processes and makes the end results easy to interpret. Also the approach allows us to give a version of the classical heavy-traffic limit theorem for the G/GI/∞ queue which, in particular, reconciles the limits obtained earlier by Iglehart and Borovkov. This revised version was published online in June 2006 with corrections to the Cover Date.     

On the optimal design of tandem queueing systems with finite buffers

We consider tandem queueing systems that can be formulated as a continuous-time Markov chain, and investigate how to maximize the throughput when the queue capacities are limited. We consider various constrained optimization problems where the decision variables are of one or more of the following types: (1) expected service times, (2) queue capacities, and (3) the number of servers at the respective stations. After surveying our previous studies of this kind, we open up consideration of three new problems by presenting some numerical results that should give some insight into the general form of the optimal design.     

Markov-modulated queueing systems

Markov-modulated queueing systems are those in which the primary arrival and service mechanisms are influenced by changes of phase in a secondary Markov process. This influence may be external or internal, and may represent factors such as changes in environment or service interruptions. An important example of such a model arises in packet switching, where the calls generating packets are identified as customers being served at an infinite server system. In this paper we first survey a number of different models for Markov-modulated queueing systems. We then analyze a model in which the workload process and the secondary process together constitute a Markov compound Poisson process. We derive the properties of the waiting time, idle time and busy period, using techniques based on infinitesimal generators. This model was first investigated by G.J.K. Regterschot and J.H.A. de Smit using Wiener-Hopf techniques, their primary interest being the queue-length and waiting time.     

Approximations for the conditional waiting times in the GI/G/c queue

This note presents a two-moment approximation for the conditional average waiting time in the standard multi-server queue and an approximation for the tail probabilities of the conditional waiting time distribution in the standard single-server queue. These approximations have been tested by extensive numerical experiments.     

Bounds and heuristics for assembly-like queues

We consider an assembly system with exponential service times, and derive bounds for its average throughput and inventories. We also present an easily computed approximation for the throughput, and compare it to an existing approximation.     

A graphical investigation of error bounds for moment-based queueing approximations

Many approximations of queueing performance measures are based on moment matching. Empirical and theoretical results show that although approximations based on two moments are often accurate, two-moment approximations can be arbitrarily bad and sometimes three-moment approximations are far better. In this paper, we investigate graphically error bounds for two- and three-moment approximations of three performance measures forGI/M/ · type models. Our graphical analysis provides insight into the adequacy of two- and three-moment approximations as a function of standardized moments of the interarrival-time distribution. We also discuss how the behavior of these approximations varies with other model parameters and with the performance measure being approximated.     

Monotone control of queueing networks

This paper uses submodularity to obtain monotonicity results for a class of Markovian queueing network service rate control problems. Nonlinear costs of queueing and service are allowed. In contrast to Weber and Stidham [14], our monotonicity theorem considers arbitrary directions in the state space (not just control directions), arrival routing problems, and certain uncontrolled service rates. We also show that, without service costs, transition-monotone controls can be described by simple control regions and switching functions. The theory is applied to queueing networks that arise in a manufacturing system that produces to a forecast of customer demand, and also to assembly and disassembly networks.     

Containment of socially optimal policies in multiple-facility Markovian queueing systems

We consider a Markovian queueing system with N heterogeneous service facilities, each of which has multiple servers available, linear holding costs, a fixed value of service and a first-come-first-serve queue discipline. Customers arriving in the system can be either rejected or sent to one of the N facilities. Two different types of control policies are considered, which we refer to as ‘selfishly optimal’ and ‘socially optimal’. We prove the equivalence of two different Markov Decision Process formulations, and then show that classical M/M/1 queue results from the early literature on behavioural queueing theory can be generalized to multiple dimensions in an elegant way. In particular, the state space of the continuous-time Markov process induced by a socially optimal policy is contained within that of the selfishly optimal policy. We also show that this result holds when customers are divided into an arbitrary number of heterogeneous classes, provided that the service rates remain non-discriminatory.     

On the transient behavior of the processor sharing queue

We analyze the transient behavior of the single server queue under the processor sharing discipline. Under fairly general assumptions, we give the rate of growth of the number of customers in the queue as well as the asymptotic behavior of the residual service times described in terms of a renormalized point process.     

Bounds and approximations for continuous-time Markovian transition probabilities and large systems

We propose new bounds and approximations for the transition probabilities of a continuous-time Markov process with finite but large state-space. The bounding and approximating procedures have been exposed in another paper [S. Mercier, Numerical bounds for semi-Markovian quantities and applications to reliability, in revision for Methodology and Computing in Applied Probability] in the more general context of a continuous-time semi-Markov process with countable state-space. Such procedures are here specialized to the Markovian finite case, leading to much simpler algorithms. The aim of this paper is to test such algorithms versus other algorithms from the literature near from ours, such as forward Euler approximation, external uniformization and a finite volume method from [C. Cocozza-Thivent, R. Eymard, Approximation of the marginal distributions of a semi-Markov process using a finite volume scheme, ESAIM: M2AN 38(5) (2004) 853–875].