Queueing Performance Research Articles

Queuing Performance of Multichannel S-ALOHA Systems With Correlated Arrivals

In this paper, we examine the queuing performance of terminals in a multichannel centralized S-ALOHA system with a finite terminal population and finite queue size in each terminal, employing a uniform backoff (UB) algorithm with retry limit for collision resolution. The performance evaluations focus on uplink traffic from web browsing, which is modeled as a Markov-modulated Bernoulli process with correlated arrivals. We analyze the system performance in terms of system throughput, mean queue length, mean delay, the probability that a packet is dropped by retry limit, the probability that a packet is blocked by a full queue, and the mean and variance of packet retransmission time, in relation to the source correlation, number of channels, window size, and retry limit. For comparison, we also consider bufferless terminals with correlated arrivals. In addition, we evaluate by simulations the performance of terminals with finite queue size when they have perfect knowledge of the backlog size. Results from our study allow the parameters of the UB algorithm to properly be chosen to meet the access-level quality-of-service requirements.

Improved multi-fractal network traffic model and its performance analysis

This paper demonstrates the limitation of the traditional multi-fractal wavelet model (MWM). Through analyzing the multi-resolution behaviors of the real video traffic, we propose an improved MWM model. It synthesizes the traffic traces using another wavelet basis, and can adjust wavelet coefficients and multiplicative coefficients at each time scale, based on the network measurement. Subsequently, multifractal spectra and queue performances of the new model have been analyzed. The simulation proves it can capture the multifractal behaviors of network traces.

Loss probability estimation and control for OFDM/TDMA wireless systems considering multifractal traffic characteristics

In this paper, we analyze the queueing performance in terms of loss rate of an OFDM (orthogonal frequency-division multiplexing)/TDMA (time division multiplexing access) based wireless system taking into account the multifractal behavior of the wireless traffic flows. To this end, first, we show evidences of multifractal characteristics on wireless traffic traces. These findings motivated us to propose a traffic policing and control scheme based on a multifractal envelope process in order to maintain the traffic flows well-behaved, i.e., in accordance to the desired QoS parameters. Furthermore, by assuming a multifractal traffic model, we derive a data loss probability equation for wireless traffic flows that was applied to the OFDM/TDMA based wireless system. Simulations and comparisons to other methods were carried out in order to verify the efficiency of the proposed traffic policing scheme as well as of the loss probability estimation approach.

Scheduling commercial vehicle queues at a Canada–US border crossing

Using the context of queue operations at a major Canada–US commercial border crossing for truck-borne trade flows, we report on a computer simulation study to predict the likely impacts of smoothing those flows. We quantify the operational and resource efficiencies of smoothing for trans-border trucking companies and their trans-border supply chain partners as well as for government authorities with regulatory jurisdiction at border crossings. Our study’s major conclusion is that smoothing can achieve queue performance levels that, in the absence of smoothing, would require significant investment in truck processing capacity at border crossings.

Modeling Cross Correlation in Three-Moment Four-Parameter Decomposition Approximation of Queueing Networks

In two-moment decomposition approximations of queueing networks, the arrival process is modeled as a renewal process, and each station is approximated as a GI/G/1 queue whose mean waiting time is approximated based on the first two moments of the interarrival times and the service times. The departure process is also approximated as a renewal process even though the autocorrelation of this process may significantly affect the performance of the subsequent queue depending on the traffic intensity. When the departure process is split into substreams by Markovian random routing, the split processes typically are modeled as independent renewal processes even though they are correlated with each other. This cross correlation might also have a serious impact on the queueing performance. In this paper, we propose an approach for modeling both the cross correlation and the autocorrelation by a three-moment four-parameter decomposition approximation of queueing networks. The arrival process is modeled as a nonrenewal process by a two-state Markov-modulated Poisson process, viz., MMPP(2). The cross correlation between randomly split streams is accounted for in the second and third moments of the merged process by the innovations method. The main contribution of the present research is that both the cross correlation and the autocorrelation can be modeled in parametric decomposition approximations of queueing networks by integrating the MMPP(2) approximation of the arrival/departure process and the innovations method. We also present numerical results that strongly support our refinements.

Managing dynamic enterprise and urgent workloads on clouds using layered queuing and historical performance models

The automatic allocation of enterprise workload to resources can be enhanced by being able to make what–if response time predictions whilst different allocations are being considered. We experimentally investigate an historical and a layered queuing performance model and show how they can provide a good level of support for a dynamic-urgent cloud environment. Using this we define, implement and experimentally investigate the effectiveness of a prediction-based cloud workload and resource management algorithm. Based on these experimental analyses we: (i) comparatively evaluate the layered queuing and historical techniques; (ii) evaluate the effectiveness of the management algorithm in different operating scenarios; and (iii) provide guidance on using prediction-based workload and resource management.

Practical insights on pitfalls in estimating queue performance

Significant estimation errors, especially in high utilisation or low sample size situations, can occur when using sample data in standard queueing formulas to estimate queueing measures like time in queue. Perhaps surprisingly, these issues are rarely addressed, and especially in materials developed for students and practitioners. We establish error bounds in the case of exponential arrivals and Poisson service times for the single-server system when estimating average time in queue and also provide practical guidance for practitioners.

Tight Upper Bound for Stability of TCP/RED Systems in AQM Routers

Stability of TCP/RED systems can be achieved by tuning random early detection (RED) parameters. The authors present a tight upper bound of the maximum packet drop probability in the RED to obtain a stable queue performance in TCP/RED systems. Simulation results show that the derived stability condition is exact and match up well to our analysis.

Performance analysis of wireless hybrid-ARQ systems with delay-sensitive traffic

The design of wireless communication schemes tailored to real-time traffic requires an analysis framework that goes beyond the traditional criterion of data throughput. This work considers an approach that relates physical system parameters to the queueing performance of wireless links. The potential benefits of multi-rate techniques such as hybrid-ARQ are assessed in the context of delay-sensitive traffic using large deviations. A continuous-time Markov channel model is employed to partition the instantaneous data-rate received at the destination into a finite number of states, each representing a mode of operation of the hybrid-ARQ scheme. The proposed methodology accounts for the correlation of the wireless channel across time, which is computed in terms of level-crossing rates. The tail asymptote governing buffer overflow probabilities at the transmitter is then used to provide a measure of overall performance. This approach leads to a characterization of the effective capacity of the system which, in turn, is applied to quantify the performance advantages of hybrid-ARQ over traditional schemes.

Weighted fair share scheduling for loosely controlled concurrent event systems

In asynchronous event systems, the production of an event is decoupled from its consumption via an event queue. The loose coupling of such systems allows great flexibility as to where and when within the system the event handler of a given event is actually executed, allowing them to scale with increasing number of processors on a given node or across multiple nodes in a distributed system. Although this flexibility is useful in heterogeneous distributed systems, such as SOA, it may appear not to be well suited for real-time systems, which require an upper bound on how long an event can remain unhandled in a queue. However, we show how the weighted fair scheduling algorithms developed for QoS (quality of service) routing can be adapted to event queues to bound the delay between production and consumption. We achieve this, despite the fact that the underlying execution environment is only weakly controlled, through the use of predictive techniques on a calibrated model of the event system. Our choice of algorithms and data structures is driven in part by the highly concurrent nature of the system. We show how a weighted fair scheduler can be built on a lock-free concurrent priority queue. We analyze the performance of various such queues proposed in the literature and show that a very simple linear quantizing queue yields the best performance.

Characterization of long-range dependent traffic regulated by leaky-bucket policers and shapers

Long-range dependence (LRD) is a widely verified property of Internet traffic, which severely affects network queuing performance. An approach for guaranteeing quality-of-service requirements is enforcing the statistical profile of input traffic by policing or shaping regulators. In this paper, it is investigated by thorough simulation how leaky-bucket (LB) policers and shapers affect the LRD of regulated traffic having 1 / f α power spectral density. Unlike previous studies on this subject, spectral analysis of regulated traffic and estimation of its LRD parameter α are carried out in the time domain using the Modified Allan Variance, because of its demonstrated superior accuracy in LRD parameter estimation. Adoption of this tool allows to attain unprecedented precision and fineness in characterizing the LRD of LB regulated traffic. In addition, the probability that LRD traffic is dropped by policers or that it exceeds a delay limit in shapers is also studied. The queuing behaviour of LRD regulated traffic in FIFO schedulers is finally investigated, highlighting conditions under which a service level agreement based on delay bounds can be violated by varying α in input LRD traffic, even if this is controlled by policers or shapers.

Constructions of Linear Compressors, Nonovertaking Delay Lines, and Flexible Delay Lines for Optical Packet Switching

Queueing theory is generally known as the study of the performance of queues. In this paper, we are interested in another aspect of queueing theory, the constructions of queues via switched delay lines (SDLs). We consider three types of discrete-time queues: linear compressors, nonovertaking delay lines, and flexible delay lines. These three types of queues correspond to certain conditional nonblocking switches and (in a strict sense) nonblocking switches in switching theory. Analogous to their counterparts in switching theory, there exist multistage constructions for these three types of queues. Specifically, we develop a two-stage construction of a linear compressor. For nonovertaking delay lines and flexible delay lines, however, we develop a three-stage construction. Moreover, a flexible delay line can also be constructed by a layered Cantor network. As applications of these constructions, we show two optical switch architectures that achieve exact emulation of output-buffered switches.

The impact of production time variability on make-to-stock queue performance

Abstract In this paper, we exploit the distributional Little’s law to obtain the steady-state distribution of the number of customers in a GI/G/1 make-to-stock queueing system. Non-exponential service times in make-to-stock queue modeling are usually avoided or at best, considered in approximations due to difficulties in developing an exact method. By providing a numerical solution of the GI/G/1 make-to-stock queue, we observed the impact of production time variability on optimal inventory control policies. The numerical results prove the degree of errors in the results if an exponential service time distribution were assumed instead of the actual distribution.

Modeling of Single-Step Power Control Scheme in Finite-State Markov Channel and Its Impact on Queuing Performance

In this paper, we analyze the queuing performance in the finite-state Markov channel (FSMC) when the single-step power control (SSPC) scheme is adopted. To start with, an SSPC model and its extended power control error (PCE) model are proposed and described in detail. Such models can emulate the behaviors of the interaction between the SSPC and an FSMC and the variation of the PCE with high accuracy. Furthermore, the packet error rate that has been recently analyzed by Lee and Chang is adopted to evaluate the service rate, the time average queuing length, and the probability mass function (PMF) of the queuing length variation. Based on these results, a queuing variation model is proposed to emulate the queuing variation in the FSMC. The flooring phenomenon of the queuing length is observed under a particular requirement of an overflow probability when the SNR value is greater than a certain value. This indicates that the tradeoff among the SNR, the buffer size of the queue, and the overflow probability exists. Hence, two different approaches to select the optimal target SNR of the SSPC and the maximum buffer size of the queue under the requested queuing buffer overflow probability are proposed. The improvement of the SSPC, its impact on the system, and the validity of the proposed models are shown in this paper as well.

Field Studies of Operations and Conflicts in Drop-Off–Pick-Up Zones

Primary schools frequently have queues that can lead to substantial impacts on the local transportation infrastructure. Queuing issues appear to be critical at primary schools, with substantial queues developing at both pick-up and drop-off times at most schools. Operational and design strategies vary substantially between schools, with some schools providing aggressive supervision and multiple queue lanes. This paper documents the field studies conducted at schools in Texas. These studies included examinations of school bus and vehicle ridership, conflicts, and queuing in parent pick-up and drop-off areas at 13 elementary and five middle schools. The studies were generally performed in both mornings and afternoons. The schools examined included sites in both large and small cities in Texas. The studies examined the effects of various operational and design characteristics on queue performance. Researchers documented the design characteristics of the queuing areas (i.e., number of lanes), operational strategies (i.e., presence of school personnel supervising the loading–unloading operation), and ridership (i.e., the number of students loaded into individual vehicles). The research team measured the performance of the queues through studies of the formation and dissipation of the queues at the observed schools and the incidence of conflicts between pedestrians and motor vehicles.

Effect of queue discipline on the performance of a queueing system

The effect of three queue discipline namely first in, first out (FIFO), last in first out (LIFO) and service in random order (SIRO) on some measures of performance of a single sever queue are examined. The measures of performance are average waiting time and queuing time. The comparison of the systems were carried out by writing an appropriate program in BASIC to simulate the queue discipline. This is due to the versatile nature of simulation and the fact that it is extremely difficult to obtain numerical results mathematically for the single server queue when the queue discipline is not FIFO and the arrival process is non-stationary. The approach adopted is to generate arrival time and service times for n customers through a single sever queuing system under each queue discipline. The measures of performance were calculated for each system using appropriate expressions. The result show that the average queuing time and average waiting time are higher when the queue discipline is LIFO, whereas the total idle time for all the systems were found to be same in most cases. Journal of the Nigerian Association of Mathematical Physics Vol. 10 2006: pp. 453-456

Stability and performance of ant queue inspired task partitioning methods

In this paper, we consider computing systems that have autonomous helper components which fulfill support functions and that possess reconfigurable hardware so that they can specialize to different types of service tasks. Several self-organized task partitioning methods are proposed that can be used by the helper components to decide how to reconfigure and which service tasks to execute. The proposed task partitioning methods are inspired by the so-called ant queue system that can be found in real ants for partitioning tasks between the individuals. The aim of this study is to investigate basic properties of the task partitioning methods, like stability and efficiency, in order to obtain basic insights into the design of task partitioning methods in self-organized service systems. More precisely, the investigations are threefold: (1) discrete event simulations are used to investigate systems, (2) for a simple version of the task partitioning system analytical stability results are obtained by means of delay differential equation systems and (3) by numerically solving initial value problems.

큐잉모델에 기초한 자율 지능 부하 관리 시스템 연구

본 논문은 전력회사와 수용가가 실시간으로 상호 협력하면서 효율적으로 여름철 냉방부하를 관리하여 첨두부하를 저감하기 위한 기법에 대하여 기술한다. 전력회사가 필요 관리량을 수용가 측에 요청하면 부하관리 요청을 받은 수용가, 예를 들면 대규모 아파트 단지의 중앙 지능관리 서버(Central Intelligent Management Server: CIMS)가 자율적인 관리를 통하여 요구받은 관리전력 이내로 부하율을 조정하도록 한다. 부하율을 관리하기 위한 수단으로 CIMS는 필요 관리 부하전력에 따라 냉방기를 가동할 수 있는 권리에 해당하는 적정 수의 토큰(Token)을 발행하고 각 세대내의 냉방기를 제어할 수 있는 다수의 로컬 지능관리 서버(Local Intelligent Management Server: LIMS)들이 각 집안의 상황에 따라 냉방기를 가동하기 원할 경우에 전체 관리 서버에게 냉방기를 가동할 수 있는 권리에 해당하는 토큰을 요청하도록 한다. CIMS는 LIMS들의 토큰 요청을 관리하기 위한 큐(Queue)를 구성하여 관리하고 토큰을 요구한 각 LIMS들은 토큰을 할당받는 대로 자신이 담당하는 세대내의 냉방기를 가동하고 일정조건이 만족되면 자율 또는 타율로 토큰을 반납하여 큐에서 기다리는 다른 LIMS들에게도 토큰이 할당될 수 있도록 한다. 큐 운영 정책에 다양한 냉방기 운용 환경 정보를 반영하고 CIMS와 LIMS간에 원만한 상호 협력을 위하여 지능형 멀티 에이전트(Muti-Agent) 시스템으로 구성하고, 이들을 계층구조의 네트워크로 연결하였다. 제안하는 기법을 서울의 한 아파트단지의 여름철 전력수요 데이터를 예로 하여 예상되는 대기 시간과 지원 받을 수 있는 전력요금 규모를 산정해 보았으며, 비교적 적은 수용가의 불편으로 큰 부하관리 효과를 얻을 수 있는 것을 알 수 있었다.

Steady state analysis of a M<SUP align=right>x</SUP>/G/1 queue with two service modes and multiple vacations

This paper deals with a single server Poisson queue with state dependent service. The service is rendered with different service modes as singly or in bulk according to the queue length. The service starts only if at least one customer is waiting for service. The server switches over from single service to bulk service only at service initiation epoch according to the queue length. If the number of customers waiting for service is ξ (1 ≤ ξ < a), then the server serves one by one. At the service initiation epoch, if the number of waiting customers ξ is at least 'a' (a < b), then the service is rendered according to Neut's general bulk service rule. At the service completion epoch, if there are no customers in the queue, the server avails multiple vacations, until he finds at least one customer in the queue. The Probability Generating Function (PGF) of the queue size and performance measures are obtained. Numerical results are presented in order to illustrate the effectiveness of performance of the state dependent queue.

Improving Inter-server Fairness in Active Queue Management

More than than fifty new active queue management (AQM) algorithms have been proposed since 1999. However, none of them address the inter-server fairness problem. This letter describes the inter-server fairness problem in conventional AQM, and presents a group-based virtual queue algorithm, called GVQ, as a novel AQM approach to solve this problem. The queue complexity and performance of GVQ are evaluated by analysis. The validity of the analytical result and the effectiveness of GVQ are confirmed through ns-2 simulations.