We survey recent advances in theories and models for Internet Quality of Service (QoS). We start with the theory of network calculus, which lays the foundation for support of deterministic performance guarantees in networks, and illustrate its applications to integrated services, differentiated services, and streaming media playback delays. We also present mechanisms and architecture for scalable support of guaranteed services in the Internet, based on the concept of a stateless core. Methods for scalable control operations are also briefly discussed. We then turn our attention to statistical performance guarantees, and describe several new probabilistic results that can be used for a statistical dimensioning of differentiated services. Lastly, we review recent proposals and results in supporting performance guarantees in a best effort context. These include models for elastic throughput guarantees based on TCP performance modeling, techniques for some quality of service differentiation without access control, and methods that allow an application to control the performance it receives, in the absence of network support.

We present a statistical network calculus in a setting where both arrivals and service are specified interms of probabilistic bounds. We provide explicit bounds on delay, backlog, and output burstiness in a network. By formulating well-known effective bandwidth expressions in terms of envelope functions,we are able to apply our calculus to a wide range of traffic source models, including Fractional Brownian Motion. We present probabilistic lower bounds on the service for three scheduling algorithms: Static Priority (SP), Earliest Deadline First (EDF), and Generalized Processor Sharing (GPS).

We consider the problem of bounding the probability of buffer overflow in a network node fed with independent arrival processes that are each constrained by arrival curves, but that are served as an aggregate. Existing results (for example [1] and [2]) assume that the node is a constant rate server. However, in practice, one finds complex network nodes that do not provide a constant service rate, and thus to which the existing bounds do not apply. Now many nodes can be adequately abstracted by a service curve property. We extend the results in [1] and [2] to such cases. As a by-product, we also provide a slight improvement to the bound in [2]. Our bounds are valid for both discrete and continuous time models. Index Terms---Statistical multiplexing, scheduling, queuing analysis I.

Packet Scale Rate Guarantee (PSRG) is a generic node model which underlies the definition of Expedited Forwarding (EF) proposed in the context of Internet Differentiated Services. For the case of FIFO nodes, PSRG is equivalent to the well-understood concept of adaptive service curve. However, in practice, many devices do not necessarily preserve the FIFO property, and therefore known FIFO results do not hold. This paper analyzes the properties of PSRG in the absence of FIFO assumption. Our analysis is based on a novel characterization of PSRG which avoids the use of virtual finish times; it is obtained by min-max algebra. We use it to show that delay bounds previously obtained for the FIFO case are still valid; in contrast, we find that this is not true for the characterization of the concatenation of two nodes.

In this paper we propose an edge-based quality of service architecture aimed at site-to-site private networks over the Internet. We extend the traditional point-to-point service model to a point-to-set service model, assuming a finite, bounded set of destination sites. Instead of provisioning point-to-point links between a source and its set of destinations, a point-to-set service allows the user to have an allocated bandwidth, which could be flexibly assigned to traffic going toward any destination within the set. The proposed point-to-set service provides low loss rates and flexibility to users while allowing providers to obtain multiplexing gains by employing a probabilistic admission control test. The model is demonstrated to...

Carriers increasingly di#erentiate their wide-area connectivity offerings by means of customized services, such as virtual private networks (VPN) with Quality of Service (QoS) guarantees, or QVPNs. The key challenge faced by carriers is to maximize the number of QVPNs admitted by exploiting the statistical multiplexing nature of input tra#c. While existing measurement-based admission control algorithms utilize statistical multiplexing along the bandwidth dimension, they do not satisfactorily exploit statistical multiplexing along the delay dimension to guarantee distinct per-QVPN delay bounds. This paper presents Delay Distribution Measurement (DDM) based admission control algorithm, the first measurement-based approach that e#ectively exploits statistical multiplexing along the delay dimension. In other words, DDM exploits the well known fact that the actual delay experienced by most packets of a QVPN is usually far smaller than its worst-case delay bound requirement since multiple QVPNs rarely send traffic bursts at the same time. Additionally, DDM supports QVPNs with distinct probabilistic delay guarantees -- QVPNs that can tolerate more delay violations can reserve fewer resource than those that tolerate less, even though they require the same delay bound. A comprehensive performance evaluation using Voice over IP traces shows that, when compared to deterministic admission control, DDM can potentially increase the number of admitted QVPNs (and link utilization) by up to a factor of 3.0 even when the delay violation probability is as small as 10 -5 .

Abstract. In this paper we study the impact of statistical multiplexing on leaky-bucket regulated traffic streams as they pass through the network. In particular we show that the burstiness of a flow is randomized as it transits through the nodes with mean equal to its initial burstiness value at the ingress. We then show that the random burstiness for a single flow converges to a constant equal to the initial value at the ingress when the flow is multiplexed with a large number of sources. The results do not depend on independence or homogeniety between flows. We conclude by providing some simulation results that confirm the theory. 1

Distributed real-time embedded (DRE) systems are key components of critical infrastructure including surveillance, target tracking, electric grid management, traffic control, avionics, and communications systems. They require (1) the coordinated management of multiple resources, such as the CPU, network, and disk, (2) end-to-end (E2E) real-time guarantees across the use of multiple resources, and (3) feedback control across multiple resources. None of these properties is supported as a first-class feature within the state-of-theart real-time operating systems, but are left out as an inconvenient detail to be managed by DRE application programmers. In this paper, we shed light on this fundamental problem and make the case for greater research into the development of theory and a runtime systems for coordinated allocation and scheduling of multiple resources in real-time operating systems. We also present the outlines of our proposed solution approach, called the Multiple Resource Allocation and Scheduling (MURALS) framework, that aims to bridge this gap between the need for E2E timing requirements and the techniques to coordinate the use of multiple resources. 1.

We describe recent advances in theories and architecture that support performance guarantees needed for quality of service networks. We start with deterministic computations and give applications to integrated services, differentiated services, and playback delays. We review the methods used for obtaining a scalable integrated services support, based on the concept of a stateless core. New probabilistic results that can be used for a statistical dimensioning of differentiated services are explained; some are based on classical queuing theory, while others capitalize on the deterministic results. Then we discuss performance guarantees in a best effort context; we review: methods to provide some quality of service in a pure best effort environment; methods to provide some quality of service differentiation without access control, and methods that allow an application to control the performance it receives, in the absence of network support. Keywords---Quality of Service, Performance Guarantees, Network Calculus, Elastic Services, Differentiated Services, Integrated Services, Scalability I.

In this paper, we present performance bounds for multiplexed traffic streams, that are leaky bucket regulated with peak rate, mean rate and burst size constraints. We consider independent, heterogeneous streams, which are multiplexed in a common buffer. We derive bounds on the tail of the probability distribution function of the buffer content in the stationary regime. We identify in particular the extremal traffic from the point of view of buffer overflow in the single input case. We then consider the case when the number of sources is large and the buffer is also scaled according to the number of streams. We finally show that an M/G/1 bound, where service times and the intensity of the input Poisson process depend on the (σ, ρ) parameters of each source, holds in the unscaled buffer case.