Zhi Quan

Priority queueing analysis of self-similar in high-speed networks

Differentiated services (DiffServ) networking technologies are under development with the objective to support diverse classes of traffic that require different quality of service (QoS) guarantees. Recent studies have shown that real network traffic exhibits self-similarity or long-range dependence (LRD) in high-speed communication networks, which has a deteriorating impact on the network performance. To assist the development of admission control mechanisms, which can accommodate heterogeneous traffic including short-range independence and long-range dependence, this paper proposes a measurement-based approach to estimate the buffer overflow probability for each priority queue in a multiplexer deploying the head-of-the-line (HOL) priority discipline. The accuracy and effectiveness of this model has been verified by simulations. The results of this paper will provide a practical insight into the buffer dimensioning and admission control design of a HOL multiplexer.

A statistical framework for EDF scheduling

Earliest deadline first (EDF) has become one of the most promising scheduling schemes for providing quality-of-service differentiation over high speed networks. In this letter, we study the deadline violation (loss) probability at an EDF scheduling switch. An analytical framework has been developed for estimating the loss probabilities for the aggregated traffic and the individual flows. This enables us to determine whether a given flow can meet its deadline with the required loss probability. As illustrated from the simulation results using real network traffic, the asymptotic approximations presented are accurate enough to predict the real metrics.

Statistical admission control for real-time services under earliest deadline first scheduling

The design of call admission control (CAC) mechanisms has been a critical issue in providing quality-of-service (QoS) guarantees for heterogeneous traffic flows over integrated service (IntServ) and/or differentiated service (DiffServ) networks. Earliest deadline first (EDF) is an ideal scheduler for real-time services because of its optimal admissible region and delay bound properties. The major difficulty in developing an effective and efficient CAC algorithm for statistical services is the analysis of per-class deadline violation (loss) probabilities with respect to the delay bounds. In this paper, we provide an analytical approach to evaluate the aggregate and per-class deadline violation probabilities of an EDF scheduler. Based on these theoretical foundations, we derive the admission control conditions and then propose a CAC algorithm for statistic services under EDF scheduling. In addition, we show that the QoS metrics that an EDF scheduler actually guarantees have an asymptotic ordering property, which provides an important insight into the design and control of EDF networks. The effectiveness and performance of our proposed algorithm have been validated by trace-driven simulation experiments using MPEG and H.263 encoded video sources.

Asymptotic loss of real-time traffic in wireless mobile networks with selective-repeat ARQ

We study the issue of quality of service (QoS) for real-time traffic over a wireless channel deploying automatic repeat request (ARQ) error control. An analytical model has been derived to evaluate the queueing related loss and the wireless channel related loss. In contrast to previous work, this model quantifies the interaction between the network layer and the physical layer, and then it enables the admission controllers of wireless networks to improve utilization while satisfying the traffic QoS constraints through cross-layer design techniques.

Analysis of packet loss for real-time traffic in wireless mobile networks with ARQ feedback

In this paper, the provision of quality-of-service (QoS) for real-time traffic over a wireless channel deploying automatic repeat request (ARQ) error control is investigated. By introducing the concepts of ARQ capacity and effective capacity, an analytic model has been derived to evaluate the loss probabilities in both the network layer and the physical layer. In contrast to the previous results, this model quantifies the interaction between the network and physical layers. As shown by the simulation experiments, our analysis can predict the real metrics under a wide range of conditions. This enables the call admission controller in wireless networks to control and optimize traffic QoS using instantaneous channel status information.

Impact of self-similarity on performance evaluation in differential service networks

Priority queueing has currently become a popular research topic as a low-cost effective method to enable both quality of service (QoS) and differential services (DS) over large-scale high-speed networks. However, there are so many uncertain issues that require further investigation before its deployment into real wide area networks (WAN). The impacts of self-similar traffic on network performance has received much attention recently, however the effects of DS deployment on self-similar traffic networks has not been fully investigated yet. Therefore, the objective of this paper is to study QoS measures in priority queues under self-similar traffic. We first introduce the techniques that we use to generate self-similar network traffic, as well as the methods of how to design and implement priority queues through OPNET. Then we present our statistical results from a simulation study with synthesized self-similar traffic. Our study results show that self-similar traffic, compared with traditional short-range dependent (SRD) models, requires longer queues and thus larger buffers in the DS network design.

An analytical framework for EDF schedulers based on the dominant time scale

Earliest deadline first (EDF) has become one of the most promising scheduling schemes for providing quality-of-service (QoS) differentiation over high speed networks. We study the deadline violation (loss) probability at an EDF scheduling switch. An analytical framework based on the dominant time scale (DTS) has been developed for estimating the deadline violation probabilities of the aggregated traffic and the individual flows. This enables us to determine whether a given flow can meet its deadline with the required loss probability. As shown by simulations using real network traffic, the developed asymptotic approximations are accurate enough to predict the real metrics. The framework can serve as the basis for the design of call admission control (CAC) mechanisms which are targeted to provide statistical guarantees on transmission delays and/or loss.

Admission control for probabilistic services with earliest deadline first scheduling

Future high speed packet-switching networks deploying integrated service (IntServ) or differentiated service (DiffServ) architectures are expected to provide heterogeneous quality-of-service (QoS) guarantees for a variety of applications. Call admission control (CAC) plays a critical role in achieving this goal and is an integration of the traffic models, scheduling disciplines, and QoS specifications. Its major task is to decide whether a new connection should be granted while the QoS requirements of all the connections are to be satisfied. However, it is well known that developing an effective and efficient CAC algorithm for a stochastic system such as an EDF scheduler is generally very difficult due to the intractability of per-class QoS analysis. A robust control mechanism is necessary for the long-range dependent traffic with infinite variance. In this paper, we present an admission control algorithm for probabilistic services scheduled by an EDF scheduler. In deriving the admission condition, we consider all the connections with similar QoS constraints as an aggregate traffic class. A statistical framework is also developed to analyze the per-class QoS metrics.

Performance analysis of WMPLS signaling and control in ad hoc networks

The framework of wireless multiprotocol label switching (WMPLS) technology in applications of ad hoc networking and mobile ad hoc networking (MANET) are presented in this paper. WMPLS has been designed to be a homogeneous protocol to MPLS, GMPLS, and MPLambdaS, which are the strongest candidates for the next generation WAN technologies. This paper provides the framework of WMPLS and the signaling/control protocols to establish connection-oriented and connectionless label switched paths (LSPs) in mobile communication networks and ad hoc networks, and also provides a performance analysis compared to other wireless technologies.

A novel analysis of queue length in differentiated services networks with self-similar arrival processes

It is well known that traditional analytic methods of queueing systems, are based on the distributions of interarrival time and service time. However, it is quite inconvenient to, employ these methods directly to the analysis of current self-similar traffic models. In this paper, we first derive a novel analytic model based on the arrival rate and the service rate for single-class steady-state queueing systems. Then the derivations are extended to provide upper and, lower boundary conditions for multipriority queues in networks deploying differentiated services (DS). In addition, the analytical model is also applied to the analysis of DS effects on self-similar traffic. The results illustrate the performance gain in queue length of the priority classes that DS can provide compared to a network that does not deploy DS. Additionally, the upper lower boundary conditions of the queue lengths for each priority class also, serves as a system design guideline.