Competitive queue policies for differentiated services

We consider the setting of a network providing differentiated services. As is often the case in differentiated services, we assume that the packets are tagged as either being a high priority packet or a low priority packet. Outgoing links in the network are serviced by a single FIFO queue.Our model gives a benefit of α1 to each high priority packet and a benefit of 1 to each low priority packet. A queue policy controls which of the arriving packets are dropped and which enter the queue. Once a packet enters the queue it is eventually sent. The aim of a queue policy is to maximize the sum of the benefits of all the packets it sends.We analyze and compare different queue policies for this problem using the competitive analysis approach, where the benefit of the online policy is compared to the benefit of an optimal offline policy. We derive both upper and lower bounds for the policies we consider. We believe that competitive analysis gives important insight to the performance of these queuing policies.     

Many-Sources Delay Asymptotics with Applications to Priority Queues

In this paper, we study discrete-time priority queueing systems fed by a large number of arrival streams. We first provide bounds on the actual delay asymptote in terms of the virtual delay asymptote. Then, under suitable assumptions on the arrival process to the queue, we show that these asymptotes are the same. As an application of this result, we then consider a priority queueing system with two queues. Using the earlier result, we derive an upper bound on the tail probability of the delay. Under certain assumptions on the rate function of the arrival process, we show that the upper bound is tight. We then consider a system with Markovian arrivals and numerically evaluate the delay tail probability and validate these results with simulations.     

Queueing and Fluid Analysis of Partial Message Discarding Policy

We consider a stream of packets that arrive at a queue with a finite buffer. A group of consecutive packets constitutes a frame. We assume that when an arriving packet finds the queue full, not only is the packet lost but also the future packets that belong to the same frame will be rejected. The first part of the paper deals with a detailed packet level queueing model; we obtain exact expressions for the stationary queue length distribution and the goodput ratio (i.e. the fraction of arriving frames that experience no losses). The second part deals with a fluid model and the fluid analysis leads to simple closed form expressions for the stationary workload process and the fluid goodput ratio.     

Provable bounds for the mean queue lengths in a heterogeneous priority queue

We analyze a two-class two-server system with nonpreemptive heterogeneous priority structures. We use matrix–geometric techniques to determine the stationary queue length distributions. Numerical solution of the matrix–geometric model requires that the number of phases be truncated and it is shown how this affects the accuracy of the results. We then establish and prove upper and lower bounds for the mean queue lengths under the assumption that the classes have equal mean service times. This revised version was published online in June 2006 with corrections to the Cover Date.     

Monitoring the packet gap of real-time packet traffic

Real-time packet traffic is characterized by a strict deadline on the end-to-end time delay and an upper bound on the information loss. Due to the high correlation among consecutive packets, the individual packet loss does not well characterize the performance of real-time packet sessions. An additional measure of packet loss is necessary to adequately assess the quality of each real-time connection. The additional measure considered here is the average number of consecutively lost packets, also called the average packet gap. We derive a closed form for the average packet gap for the multiclassG/G/m/B queueing system in equilibrium and show that it only depends on the loss behavior of two consecutive packets. This result considerably simplifies the monitoring process of real-time packet traffic sessions. If the packet loss process is markovian, the consecutive packet loss has a geometric distribution.     

Mean queue size in a queue with discrete autoregressive arrivals of order <Emphasis Type="Italic">p</Emphasis>

We consider a discrete time single server queueing system where the arrival process is governed by a discrete autoregressive process of order p (DAR(p)), and the service time of a customer is one slot. For this queueing system, we give an expression for the mean queue size, which yields upper and lower bounds for the mean queue size. Further we propose two approximation methods for the mean queue size. One is based on the matrix analytic method and the other is based on simulation. We show, by illustrations, that the proposed approximations are very accurate and computationally efficient.     

The M/G/c queue in light traffic

Under light traffic, we investigate the quality of a well‐known approximation for first‐moment performance measures for an M/G/c queue, and, in particular, conditions under which the approximation is either an upper or a lower bound. The approach is to combine known relationships between quantities such as average delay and time‐average work in system with direct sample‐path comparisons of system operation under two modes of operation: conventional FIFO and a version of preemptive LIFO. We then use light traffic limit theorems to show an inequality between time‐average work of the M/G/c queue and that of the approximation. In the process, we obtain new and improved approximations. This revised version was published online in June 2006 with corrections to the Cover Date.     

Analysis of an open tandem queueing network with population constraint and constant service times 1

We consider an open tandem queueing network with population constraint and constant service times. The total number of customers that may be present in the network can not exceed a given value K. Customers arriving at the queueing network when there are more than K customers are forced to wait in an external queue. The arrival process to the queueing network is assumed to be arbitrary. We show that this queueing network can be transformed into a simple network involving only two nodes. Using this simple network, we obtain an upper and lower bound on the mean waiting time. These bounds can be easily calculated. Validations against simulation data establish the tightness of these bounds.     

Improved bounds for queues with delayed arrivals

Improved bounds are developed for a queue where arrivals are delayed by a fixed time. For moderate to heavy traffic, a simple improved upper bound is obtained which only uses the first two moments of the service time distribution. We show that our approach can be extended to obtain bounds for other types of delayed arrival queues. For very light traffic, asymptotically tight bounds can be obtained using more information about the service time distribution. While an improved upper bound can be obtained for light to moderate traffic it is not particularly easy to apply. This revised version was published online in June 2006 with corrections to the Cover Date.     

Connection admission control for constant bit rate traffic at a multi‐buffer multiplexer using the oldest‐cell‐first discipline

Consider an ATM multiplexer where M input links contend for time slots on an output link which transmits C cells per second. Each input link has its own queue of size B cells. The traffic is delay sensitive so B is small (e.g., B=20). We assume that each of the M input links carries Constant Bit Rate (CBR) traffic from a large number of independent Virtual Connections (VCs) which are subject to jitter. The fluctuations of the aggregate traffic arriving at queue i, i=1,...,M, is modeled by a Poisson process with rate λ_i. The Quality of Service (QoS) of one connection is determined in part by the queueing delay across the multiplexer and the Cell Loss Ratio (CLR) or proportion of cells from this connection lost because the buffer is full. The Oldest‐Customer(Cell)‐First (OCF) discipline is a good compromise between competing protocols like round‐robin queueing or serving the longest queue. The OCF discipline minimizes the total cell delay among all cells arriving at the contending queues. Moreover, the CLR is similar to that obtained by serving the longest queue. We develop QoS formulae for this protocol that can be calculated on‐line for Connection Admission Control (CAC). These formulae follow from a simple new expression for the exact asymptotics of a M/D/1 queue. This revised version was published online in June 2006 with corrections to the Cover Date.     

Jitter analysis of an MMPP-2 tagged stream in the presence of an MMPP-2 background stream

We first consider a single-server queue that serves a tagged MMPP-2 stream and a background MMPP-2 stream in a FIFO manner. The service time is exponentially distributed. For this queueing system, we obtain the CDF of the tagged inter-departure time, from which we can calculate the jitter, defined as a percentile of the inter-departure time. The formulation is exact, but the solution is obtained numerically, which introduces an error that has been found to be negligible. Subsequently, we consider a tandem queueing network consisting of N tandem queues, which is traversed by the MMPP-2 tagged stream, and where each queue also serves a local MMPP-2 background stream. For this queueing network, we obtain an upper bound on the CDF of the inter-departure time from the Nth queue using a heavy traffic approximation, and we verify it by simulation.     

Bounding blocking probabilities and throughput in queueing networks with buffer capacity constraints

We propose a new technique for upper and lower bounding of the throughput and blocking probabilities in queueing networks with buffer capacity constraints, i.e., some buffers in the network have finite capacity. By studying the evolution of multinomials of the state of the system in its assumed steady state, we obtain constraints on the possible behavior of the system. Using these constraints, we obtain linear programs whose values upper and lower bound the performance measures of interest, namely throughputs or blocking probabilities. The main advantages of this new technique are that the computational complexity does not increase with the size of the finite buffers and that the technique is applicable to systems in which some buffers have infinite capacity. The technique is demonstrated on examples taken from both manufacturing systems and communication networks. As a special case, for the M/M/s/s queue, we establish the asymptotic exactness of the bounds, i.e., that the bounds on the blocking probability asymptotically approach the exact value as the degree of the multinomials considered is increased to infinity. This revised version was published online in June 2006 with corrections to the Cover Date.     

Discrete-timeGeo X/G/1 queue with place reservation discipline

A discrete-time priority queueing system with place reservation discipline is studied, in which two different types of packets arrive according to batch geometric streams. It is assumed that there is a reserved place in the queue. Whenever a high-priority packet enters the queue, it will seize the reserved place and make a new reservation at the end of the queue. Low-priority arrivals take place at the end of the queue in the usual way. Using the probability generating function method, the joint distribution of system state and the delay distribution for each type are obtained.     

A single server queue with cyclically indexed arrival and service rates

In this paper we consider a queueing model that results from at least two apparently unrelated areas. One motivation to study a system of this type results from a test case of a computer simulation factor screening technique calledfrequency domain methodology. A second motivation comes from manufacturing, where due to cyclic scheduling of upstream machines, the arrival process to downstream machines is periodic. The model is a single server queue with FIFO service discipline and exponential interarrival and service times where the arrival and/or service rates are deterministic cyclic functions of the customer sequence number. We provide steady state results for the mean number in the system for the model with cyclic arrival and fixed service rates and for the model with fixed arrival and cyclic service rates. For the model with both cyclic arrival and service rates, upper and lower bounds are developed for the steady state mean waiting time in the system. Throughout the paper various implications and/or insights derived from the results of this study are discussed for frequency domain methodology.The authors acknowledge the financial support of the CBA/GSB Faculty Research Committee of the College of Business Administration, The University of Texas at Austin.     

On queueing delays of dispersed messages

We study the message queueing delays in a node of a communication system, where a message consists of a block of consecutive packets. The message delay is defined as the time elapsing between the arrival epoch of the first packet of the message to the system until after the transmission of the last packet of that message is completed. We distinguish between two types of message generation processes. The message can be generated as abatch or it can bedispersed over time. In this paper we focus on the dispersed generation model. The main difficulty in the analysis is due to the correlation between the system states observed by different packets of the same message. This paper introduces a new technique to analyze the message delay in such systems for different arrival models and different number of sessions. For anM/M/1 system with variable size messages and for the bursty traffic model, we obtain an explicit expression for the Laplace-Stieltjes transform (LST) of the message delay. Derivations are also provided for anM/G/1 system, for multiple session systems and for fixed message sizes. We show that the correlation has a strong effect on the performance of the system, and that the commonly usedindependence assumption, i.e., the assumption that the delays of packets are independent from packet to packet, can lead to wrong conclusions.     

Structural interpretation and derivation of necessary and sufficient conditions for delay moments in FIFO multiserver queues

Scheller-Wolf [12] established necessary and sufficient conditions for finite stationary delay moments in stable FIFO GI/GI/s queues that incorporate the interaction between service time distribution, traffic intensity (ρ) and the number of servers in the queue. These conditions can be used to show that when the service time has finite first but infinite αth moment, s slow servers can give lower delays than one fast server. In this paper, we derive an alternative derivation of these moment results: Both upper bounds, that serve as sufficient conditions, and lower bounds, that serve as necessary conditions are presented. In addition, we extend the class of service time distributions for which the necessary conditions are valid. Our new derivations provide a structural interpretation of the moment bounds, giving intuition into their origin: We show that FIFO GI/GI/s delay can be represented as the minimum of (s − k) i.i.d. GI/GI/1 delays, when ρ satisfies k < ρ < k+1. AMS Subject Classification 60K25     

New bounds for expected delay in FIFO M/G/c queues

Most bounds for expected delay, E[D], in GI/GI/c queues are modifications of bounds for the GI/GI/1 case. In this paper we exploit a new delay recursion for the GI/GI/c queue to produce bounds of a different sort when the traffic intensity p = λ/μ = E[S]/E[T] is less than the integer portion of the number of servers divided by two. (S AND T denote generic service and interarrival times, respectively.) We derive two different families of new bounds for expected delay, both in terms of moments of S AND T. Our first bound is applicable when E[S₂] < ∞. Our second bound for the first time does not require finite variance of S; it only involves terms of the form E[S^β], where 1 < β < 2. We conclude by comparing our bounds to the best known bound of this type, as well as values obtained from simulation. This revised version was published online in June 2006 with corrections to the Cover Date.     

Sharp and simple bounds for the Erlang delay and loss formulae

We prove some simple and sharp lower and upper bounds for the Erlang delay and loss formulae and for the number of servers that invert the Erlang delay and loss formulae. We also suggest simple and sharp approximations for the number of servers that invert the Erlang delay and loss formulae. We illustrate the importance of these bounds by using them to establish convexity proofs. We show that the probability that the M/M/s queue is empty is a decreasing and convex function of the traffic intensity. We also give a very short proof to show that the Erlang delay formula is convex in the traffic intensity when the number of servers is held constant. The complete proof of this classical result has never been published. We also give a very short proof to show that the Erlang delay formula is a convex function of the (positive integer) number of servers. One of our results is then used to get a sharp bound to the Flow Assignment Problem.     

Relaxation Time for the Discrete D/G/1 Queue

When queueing models are used for performance analysis of some stochastic system, it is usually assumed that the system is in steady-state. Whether or not this is a realistic assumption depends on the speed at which the system tends to its steady-state. A characterization of this speed is known in the queueing literature as relaxation time.The discrete D/G/1 queue has a wide range of applications. We derive relaxation time asymptotics for the discrete D/G/1 queue in a purely analytical way, mostly relying on the saddle point method. We present a simple and useful approximate upper bound which is sharp in case the load on the system is not very high. A sharpening of this upper bound, which involves the complementary error function, is then developed and this covers both the cases of low and high loads.For the discrete D/G/1 queue, the stationary waiting time distribution can be expressed in terms of infinite series that follow from Spitzer’s identity. These series involve convolutions of the probability distribution of a discrete random variable, which makes them suitable for computation. For practical purposes, though, the infinite series should be truncated. The relaxation time asymptotics can be applied to determine an appropriate truncation level based on a sharp estimate of the error caused by truncating.This revised version was published online in June 2005 with corrected coverdate     

Structural properties and stochastic bounds for a buffer problem in packetized voice transmission

We consider a communication channel which carries packetized voice. A fixed number (K) of calls are being transmitted. Each of these calls generates one packet at everyC timeslots and the channel can transmit at most one packet every timeslot. We consider the nontrivial caseK ≤C. We study the effectsK, C and the arrival process have on the number of packets in the buffer. When the call origination epochs in the firstC timeslots of theK calls are uniformly distributed (i.e. when the arrivals during the firstC timeslots have a multinomial distribution) it is shown that the stationary number of calls waiting in the buffer is stochastically increasing and convex in the number of calls. For a fixed average number of calls per slot, it is shown that increasing the number of slots per frame increases the stationary number of packets in the buffer in the sense of increasing convex ordering. Using this, it is shown that the stationary number of packets in the buffer is bounded from above by the number of packets in a stationary discreteM/D/1 queue with arrival rateK/C and unit service time. This bound is in the sense of the increasing convex order.