On the sojourn times for many-queue head-of-the-line Processor-sharing systems with permanent customers

We consider a single server system consisting of e queues with different types of customers (Poisson streams) andk permanent customers. The permanent customers and those at the head of the queues are served in processor-sharing by the service facility (head-of-the-line processor-sharing). The stability condition and a pseudo work conservation law will be given for arbitrary service time distributions; for exponential service times a pseudo conservation law for the mean sojourn tunes can be derived. In case of two queues and exponential service times, the generating function of the stationary occupancy distribution satisfies a functional equation being a Riemann-Hilbert problem which can be reduced to a Dirichlet problem for a circle. The solution yields the mean sojourn times as an elliptic integral, which can be computed numerically very efficiently. In case ofn 2 a numerical algorithm for computing the performance measures is presented, which is efficient forn 3. Since forn 4 an exact analytical or/and numerical treatment is too complex a heuristic approximation for the mean sojourn times of the different types of customers is given, which in case of a (completely) symmetric system is exact. The numerical and simulation results show that, over a wide range of parameters, the approximation works well.This work was supported by a grant from the Siemens AG.     

A multiserver retrial queue: regenerative stability analysis

We consider a multiserver retrial GI/G/m queue with renewal input of primary customers, interarrival time τ with rate , service time S, and exponential retrial times of customers blocked in the orbit. In the model, an arriving primary customer enters the system and gets a service immediately if there is an empty server, otherwise (if all m servers are busy) he joins the orbit and attempts to enter the system after an exponentially distributed time. Exploiting the regenerative structure of the (non-Markovian) stochastic process representing the total number of customers in the system (in service and in orbit), we determine stability conditions of the system and some of its variations. More precisely, we consider a discrete-time process embedded at the input instants and prove that if and , then the regeneration period is aperiodic with a finite mean. Consequently, this queue has a stationary distribution under the same conditions as a standard multiserver queue GI/G/m with infinite buffer. To establish this result, we apply a renewal technique and a characterization of the limiting behavior of the forward renewal time in the (renewal) process of regenerations. The key step in the proof is to show that the service discipline is asymptotically work-conserving as the orbit size increases. Included are extensions of this stability analysis to continuous-time processes, a retrial system with impatient customers, a system with a general retrial rate, and a system with finite buffer for waiting primary customers. We also consider the regenerative structure of a multi-dimensional Markov process describing the system. This work is supported by Russian Foundation for Basic Research under grants 04-07-90115 and 07-07-00088.     

An approximate analysis for a class of assembly-like queues

Assembly-like queues model assembly operations where separate input processes deliver different types of component (customer) and the service station assembles (serves) these input requests only when the correct mix of components (customers) is present at the input. In this work, we develop an effective approximate analytical solution for an assembly-like queueing system withN(N 2) classes of customers formingN independent Poisson arrival streams with rates {_i=1,...,N} The arrival of a class of customers is turned off whenever the number of customers of that class in the system exceeds the number for any of the other classes by a certain amount. The approximation is based on the decomposition of the originalN input stream stage into a cascade ofN-1 two-input stream stages. This allows one to refer to the theory of paired customer systems as a foundation of the analysis, and makes the problem computationally tractable. Performance measures such as server utilization, throughput, average delays, etc., can then be easily computed. For illustrative purposes, the theory and techniques presented are applied to the approximate analysis of a system withN = 3. Numerical examples show that the approximation is very accurate over a wide range of parameters of interest.     

Control and recovery from rare congestion events in a large multi-server system

We develop deterministic fluid approximations to describe the recovery from rare congestion events in a large multi-server system in which customer holding times have a general distribution. There are two cases, depending on whether or not we exploit the age distribution (the distribution of elapsed holding times of customers in service). If we do not exploit the age distribution, then the rare congestion event is a large number of customers present. If we do exploit the age distribution, then the rare event is an unusual age distribution, possibly accompanied by a large number of customers present. As an approximation, we represent the large multi-server system as an M/G/∞ model. We prove that, under regularity conditions, the fluid approximations are asymptotically correct as the arrival rate increases. The fluid approximations show the impact upon the recovery time of the holding-time distribution beyond its mean. The recovery time may or not be affected by the holding-time distribution having a long tail, depending on the precise definition of recovery. The fluid approximations can be used to analyze various overload control schemes, such as reducing the arrival rate or interrupting services in progress. We also establish large deviations principles to show that the two kinds of rare events have the same exponentially small order. We give numerical examples showing the effect of the holding-time distribution and the age distribution, focusing especially on the consequences of long-tail distributions. This revised version was published online in June 2006 with corrections to the Cover Date.     

Dynamic Routing and Admission Control in High-Volume Service Systems: Asymptotic Analysis via Multi-Scale Fluid Limits

Motivated by applications in telephone call centers, we consider a service system model with m customer classes and r server pools. The model is one with doubly stochastic arrivals, which means that the m-vector λ of instantaneous arrival rates is allowed to vary both temporally and stochastically. Two levels of dynamic control are considered: customers may be either blocked or accepted at the time of their arrival, and then accepted customers of each class must be routed, either immediately upon acceptance or after some period of waiting, to a server pool that is qualified to handle that class. Customers who are made to wait before commencement of their service are liable to defect. The objective is to minimize the expected sum of blocking costs, waiting costs and defection costs over a fixed and finite planning horizon. We consider an asymptotic parameter regime in which (i) the arrival rates, service rates and defection rates are uniformly accelerated by a large factor κ, then (ii) arrival rates are increased by an additional factor g(κ), and the number of servers in each pool is increased by g(κ) as well. This produces a separation of time scales, justifying a pointwise stationary stochastic fluid approximation for our original system model. In the stochastic fluid approximation, optimal admission control and routing decisions are determined by a simple linear program that uses the current arrival rate vector λ as data. We explain how to implement the fluid model's optimal control policy in our original service system context, and prove that the proposed implementation is asymptotically optimal in the first-order sense. AMS subject classification: 60K30, 90B15, 90B36     

Waiting-Time Asymptotics for the M/G/2 Queue with Heterogeneous Servers

In this paper, we consider a c-server queuing model in which customers arrive according to a batch Markovian arrival process (BMAP). These customers are served in groups of varying sizes ranging from a predetermined value L through a maximum size, K. The service times are exponentially distributed. Any customer not entering into service immediately orbit in an infinite space. These orbiting customers compete for service by sending out signals that are exponentially distributed with parameter . Under a full access policy freed servers offer services to orbiting customers in groups of varying sizes. This multi-server retrial queue under the full access policy is a QBD process and the steady state analysis of the model is performed by exploiting the structure of the coefficient matrices. Some interesting numerical examples are discussed.     

A critically loaded multiclass Erlang loss system

We consider a generalization of the classical Erlang loss model to multiple classes of customers. Each of the J customer classes has an associated Poisson arrival process and an arbitrary holding time distribution. Classj customers requireM _j servers simultaneously. We determine the asymptotic form of the blocking probabilities for all customer classes in the regime known as critical loading, where both the number of servers and offered load are large and almost equal. Asymptotically, the blocking probability of classj customers is proportional toM _j . This asymptotic result provides an approximation for the blocking probabilities which is reasonably accurate. We also consider the behavior of the Erlang fixed point (reduced load approximation) for this model under critical loading. Although the blocking probability of classj customers given by the Erlang fixed point is again asymptotically proportional toM _j , the Erlang fixed point typically gives the wrong limit. Next we show that under critical loading the throughputs have a pleasingly simple form of monotonicity with respect to arrival rates: the throughput of classi is increasing in the arrival rate of classi and decreasing in the arrival rate of classj forji. Finally, we compare two simple control policies for this system under critical loading.     

On the Diophantine Queueing System, <Emphasis Type="Italic">D/D</Emphasis><Superscript><Emphasis Type="Italic">a,b</Emphasis></Superscript><Emphasis Type="Italic">/</Emphasis>1

This paper considers the solution of a deterministic queueing system. In this system, the single server provides service in bulk with a threshold for the acceptance of customers into service. Analytic results are given for the steady-state probabilities of the number of customers in the system and in the queue for random and pre-arrival epochs. The solution of this system is a prerequisite to a four-point approximation to the model GI/G ^a,b /1. The paper demonstrates that the solution of such a system is not a trivial problem and can produce interesting results. The graphical solution discussed in the literature requires that the traffic intensity be a rational number. The results so generated may be misleading in practice when a control policy is imposed, even when the probability distributions for the interarrival and service times are both deterministic.     

Transient and asymptotic behavior of synchronization processes in assembly-like queues

This paper analyzes the synchronization process of an assembly-like queueing system in which two distinct types of items/customers arrive at separate buffers, according to independent Poisson processes, so as to be synchronized into pairs at a synchronization node. Once a pair is synchronized it then queues up for service from a single server on a first-in-first-out basis as pairs. It is assumed that the service times of pairs are exponentially distributed and that the system has infinite capacity. Despite their practical significance, such queueing systems have not been adequately treated in the literature due to their transience or null recurrence. We first investigate the transient and asymptotic properties of the synchronization process’ first two moments, both analytically and numerically. Motivated by the observed asymptotic behavior, we then propose an M/M/1 approximation to describe the behavior of such assembly-like queueing systems. Finally, a numerical study of the proposed approximation reveals that it performs sufficiently well for practical applications.     

Generalised ‘join the shortest queue’ policies for the dynamic routing of jobs to multi-class queues

Jobs or customers arrive and require service that may be provided at one of several different stations. The associated routing problems concern how customers may be assigned to stations in an optimal manner. Much of the classical literature concerns a single class of customers seeking service from a collection of homogeneous stations. We argue that many contemporary application areas call for the analysis of routing problems in which many classes of customer seek service provided at a collection of diverse stations. This paper is the first to consider routing policies in such complex environments which take appropriate account of the degree of congestion at each service station. A simple and intuitive class of policies emerges from a policy improvement approach. In a numerical study, the policies were close to optimal in all cases.     

Multi-server retrial queue with negative customers and disasters

We consider a multi-server retrial queue with waiting places in service area and four types of arrivals, positive customers, disasters and two types of negative customers, one for deleting customers in orbit and the other for deleting customers in service area. The four types of arrivals occur according to a Markovian arrival process with marked transitions (MMAP) which may induce the dependence among the arrival processes of the four types. We derive a necessary and sufficient condition for the system to be positive recurrent by comparing sample paths of auxiliary systems whose stability conditions can be obtained. We use a generalized truncated system that is obtained by modifying the retrial rates for an approximation of stationary queue length distribution and show the convergence of approximation to the original model. An algorithmic solution for the stationary queue length distribution and some numerical results are presented.      

A queueing model for crowdsourcing

Crowdsourcing is getting popular after a number of industries such as food, consumer products, hotels, electronics, and other large retailers bought into this idea of serving customers. In this paper, we introduce a multi-server queueing model in the context of crowdsourcing. We assume that two types, say, Type 1 and Type 2, of customers arrive to a c-server queueing system. A Type 1 customer has to receive service by one of c servers while a Type 2 customer may be served by a Type 1 customer who is available to act as a server soon after getting a service or by one of c servers. We assume that a Type 1 customer will be available for serving a Type 2 customer (provided there is at least one Type 2 customer waiting in the queue at the time of the service completion of that Type 1 customer) with probability \(p, 0 \le p \le 1\). With probability \(q = 1 - p\), a Type 1 customer will opt out of serving a Type 2 customer provided there is at least one Type 2 customer waiting in the system. Upon completion of a service a free server will offer service to a Type 1 customer on an FCFS basis; however, if there are no Type 1 customers waiting in the system, the server will serve a Type 2 customer if there is one present in the queue. If a Type 1 customer decides to serve a Type 2 customer, for our analysis purposes that Type 2 customer will be removed from the system as Type 1 customer will leave the system with that Type 2 customer. Under the assumption of exponential services for both types of customers we study the model in steady state using matrix analytic methods and establish some results including explicit ones for the waiting time distributions. Some illustrative numerical examples are presented.     

An M/G/1 queue with cyclic service times

In this paper we consider a single server queue with Poisson arrivals and general service distributions in which the service distributions are changed cyclically according to customer sequence number. This model extends a previous study that used cyclic exponential service times to the treatment of general service distributions. First, the stationary probability generating function and the average number of customers in the system are found. Then, a single vacation queueing system with aN-limited service policy, in which the server goes on vacation after servingN consecutive customers is analyzed as a particular case of our model. Also, to increase the flexibility of using theM/G/1 model with cyclic service times in optimization problems, an approximation approach is introduced in order to obtain the average number of customers in the system. Finally, using this approximation, the optimalN-limited service policy for a single vacation queueing system is obtained.On leave from the Department of Industrial Engineering, Iran University of Science and Technology, Narmak, Tehran 16844, Iran.     

Delay Distribution of (Im)Patient Customers in a Discrete Time D-MAP/PH/1 Queue with Age-Dependent Service Times

This paper presents an algorithmic procedure to calculate the delay distribution of (im)patient customers in a discrete time D-MAP/PH/1 queue, where the service time distribution of a customer depends on his waiting time. We consider three different situations: impatient customers in the waiting room, impatient customers in the system, that is, if a customer has been in the waiting room, respectively, in the system for a time units it leaves the waiting room, respectively, the system. In the third situation, all customers are patient – that is, they only leave the system after completing service. In all three situations the service time of a customer depends upon the time he has spent in the waiting room. As opposed to the general approach in many queueing systems, we calculate the delay distribution, using matrix analytic methods, without obtaining the steady state probabilities of the queue length. The trick used in this paper, which was also applied by Van Houdt and Blondia [J. Appl. Probab., Vol. 39, No. 1 (2002) pp. 213–222], is to keep track of the age of the customer in service, while remembering the D-MAP state immediately after the customer in service arrived. Possible extentions of this method to more general queues and numerical examples that demonstrate the strength of the algorithm are also included.     

Runge-Kutta Methods Adapted to Manifolds and Based on Rigid Frames

We consider numerical integration methods for differentiable manifolds as proposed by Crouch and Grossman. The differential system is phrased by means of a system of frame vector fields E ₁, ... , E _n on the manifold. The numerical approximation is obtained by composing flows of certain vector fields in the linear span of E ₁, ... , E _n that are tangent to the differential system at various points. The methods reduce to traditional Runge-Kutta methods if the frame vector fields are chosen as the standard basis of euclidean ⁿ . A complete theory for the order conditions involving ordered rooted trees is developed. Examples of explicit and diagonal implicit methods are presented, along with some numerical results.     

Peak congestion in multi-server service systems with slowly varying arrival rates

In this paper we consider the M _t queueing model with infinitely many servers and a nonhomogeneous Poisson arrival process. Our goal is to obtain useful insights and formulas for nonstationary finite-server systems that commonly arise in practice. Here we are primarily concerned with the peak congestion. For the infinite-server model, we focus on the maximum value of the mean number of busy servers and the time lag between when this maximum occurs and the time that the maximum arrival rate occurs. We describe the asymptotic behavior of these quantities as the arrival changes more slowly, obtaining refinements of previous simple approximations. In addition to providing improved approximations, these refinements indicate when the simple approximations should perform well. We obtain an approximate time-dependent distribution for the number of customers in service in associated finite-server models by using the modified-offered-load (MOL) approximation, which is the finite-server steady-state distribution with the infinite-server mean serving as the offered load. We compare the value and lag in peak congestion predicted by the MOL approximation with exact values for M _t/M/s delay models with sinusoidal arrival-rate functions obtained by numerically solving the Chapman–Kolmogorov forward equations. The MOL approximation is remarkably accurate when the delay probability is suitably small. To treat systems with slowly varying arrival rates, we suggest focusing on the form of the arrival-rate function near its peak, in particular, on its second and third derivatives at the peak. We suggest estimating these derivatives from data by fitting a quadratic or cubic polynomial in a suitable interval about the peak. This revised version was published online in June 2006 with corrections to the Cover Date.     

A discrete queue with double thresholds policy and its application to SVC systems

A discrete time Geo/Geo/1 queue with (m, N)-policy is considered in this paper. There are three operation periods being considered: high speed, low speed service periods and idle periods. With double thresholds policy, the server begins to take a working vacation when the number of customers is below m after a service and there is one customer in the system at least. What’s more, if the system becomes empty after a service, the server will take an ordinary vacation. Otherwise, high speed service continues if the number of customers still exceeds m after a service. At the vacation completion instant, servers resume their service if the quantity of customers exceeds N. Vacations can also be interrupted when the system accumulate customers more than the prefixed threshold. Using the quasi birth-death process and matrix-geometric solution methods, we derive the stationary queue length distribution and some system characteristics of interest. Based on these, we apply the queue to a virtual channel switching system and present various numerical experiments for the system. Finally, numerical results are offered to illustrate the optimal (m, N)-policy to minimize cost function and obtain practical consequence on the operation of double thresholds policy.     

专有服务机制下提供免费体验服务的服务系统优化设计研究

通过提供免费的体验服务,服务系统可以吸引潜在顾客成为忠实顾客。本文考虑专有服务机制下提供免费体验服务和付费(常规)服务的服务系统,基于顾客的延时敏感特性,利用排队论的矩阵分析方法和谱扩展方法,研究服务系统的相关性能指标以及服务系统的优化设计,进而构建服务提供商利润函数并通过数值实例来获得免费体验服务的最优服务速率以及常规服务收取的最优服务费用,并为服务提供商提供相应的管理启示。研究表明,当越来越多的体验顾客转为付费顾客时,服务提供商需要降低体验服务的服务速率,来缓解系统的拥堵情况,减少顾客的逗留时间,并且服务提供商需要降低常规服务的服务费来弥补顾客因拥堵而造成的服务延迟。新到达顾客选择体验服务的人数越多时,服务提供商需要大幅度降低常规服务的收费标准,来吸引体验顾客成为付费顾客。     

Approximate analysis of a gordon-newell like non-product-form queueing network

This paper is tutorial in nature. It demonstrates how a particular heuristic extension of the arrival theorem, which was introduced earlier for very special network topologies, can be effectively applied (in an essentially unchanged manner) to obtain all mean performance measures for a rich class of Gordon-Newell like non-product-form queueing networks (QNs). All nodes in the class of queueing networks discussed are either FIFO or IS (pure delay), there is a single closed chain with probabilistic routing and each FIFO node also processes customers from a dedicated open chain. The number of FIFO nodesK, the number of IS nodesL and the closed chain populationN are finite but arbitrary and closed chain customers route probabilistically according to an arbitrary routing matrixQ. The think time distribution at an IS node is general, the service time distribution for both closed chain and open chain customers at the FIFO nodes is exponential with distinct service times for each, and both IS think times and FIFO service times are node dependent.The approximation technique is enhanced by an analytic study which demonstrates that it mirrors the expected behavior of the QN in many essential respects: monotonicity, bottleneck and asymptotic behavior. Moreover, in the case of balanced QNs, the approximation yields simple and explicit expressions for all quantities of interest. The analytic study and the numerical experiments presented complement one another and suggest that this approximation technique captures the essential structure of the QN, insofar as mean performance quantities are concerned.Currently on leave of absence from Bell Laboratories, at The Chinese University of Hong Kong, Department of Information Engineering, New Territories, Hong Kong.