End-to-End Quality of Service E-Book

• Herausgeber: John Wiley & Sons
• Kategorie: Wissenschaft und neue Technologien
• Sprache: Englisch

Seitenzahl: 720

Veröffentlichungsjahr: 2013

Beschreibung

A modern communication network can be described as a large, complex, distributed system composed by higher interoperating, smaller sub-systems. Today, the proliferation and convergence of different types of wired, wireless, and mobile networks are crucial for the success of the next generation networking. However, these networks can hardly meet the requirements of future integrated-service networks, and are expected to carry multimedia traffic with various Quality of Experience (QoE) and Quality of Service (QoS) requirements. Providing all relevant QoS/QoE issues in these heterogeneous networks is then an important challenge for telecommunication operators, manufacturers, and companies. The impressive emergence and the important demand of the rising generation of real-time Multi-service (such as Data, Voice VoD, Video-Conference, etc.) over communication heterogeneous networks, require scalability while considering a continuous QoS. This book presents and explains all the techniques in new generation networks which integrate efficient global control mechanisms in two directions: (1) maintain QoS requirements in order to maximize network resources utilization, and minimize operational costs on all the types of wired-wireless-mobile networks used to transport traffic, and (2) mix the QoS associated with home, access, and core networks in order to provide Quality of Service/Quality of Experience expected by users of new services.

Leseprobe

Table of Contents

Chapter 1: Challenges for End-to-End Quality of Service over Heterogenous Networks

Abdelhamid MELLOUK

1.1. Introduction

1.2. Research challenges in end-to-end QoS

1.3. Contents

1.4. Conclusion

Chapter 2: Principles and Mechanisms for Quality of Service in Networks

Zoubir MAMMERI

2.1. Introduction

2.2. Concepts and definitions

2.3. QoS parameters and application classification

2.4. Mechanisms and functions for QoS provisioning

2.5. Overview of IntServ, DiffServ and MPLS

2.6. Conclusion

2.7. References

Chapter 3: Different Approaches to Guarantee Quality of Service

Pascale MINET

3.1. Introduction to QoS

3.2. Means of managing an end-to-end time constraint

3.3. Evaluation of the end-to-end response time

3.4. Probabilistic guarantee of the end-to-end response time

3.5. QoS support in a mobile ad hoc network

3.6. Conclusion and perspectives

3.7. References

Chapter 4: Quality of Service-based Adaptive Routing Approaches.

Abdelhamid MELLOUK and Saïd HOCEINI

4.1. Introduction

4.2. QoS-based routing algorithms

4.3. QoS-based routing approaches

4.4. Inductive approaches based on machine learning paradigms

4.5. Neural net-based approach for adaptive routing policy

4.6. State-dependent KOQRA algorithm

4.7. Conclusion

4.8. References

Chapter 5: Optical Networks: New Challenges and Paradigms for Quality of Service

Ken CHEN and Wisssam FAWAZ

5.1. Introduction

5.2. Optical communication: from transmission to networking

5.3. Optical networks as a pillar for future network infrastructure

5.4. Routing and wavelength assignment

5.5. GMPLS

5.6. Towards a new optical link-based architecture

5.7. Protection against link failures

5.8. Optical packet switch and optical burst switch

5.9. Conclusion

5.10. References

Chapter 6: Pushing Quality of Service Across Inter-domain Boundaries

Bingjie FU, Cristel PELSSER, Steve UHLIG

6.1. Introduction

6.2. Background

6.3. RSVP-TE extensions to support inter-domain LSPs

6.4. State of the art in inter-domain PCE

6.5.Towards inter-AS QoS

6.6. Conclusion and perspectives

6.7. Acknowledgments

6.8. References

Chapter 7: Internet-based Collaborative Teleoperation: Towards Tailorable Groupware for Teleoperation

Samir OTMANE, Nader CHEAIB and Malik MALLEM

7.1. Introduction

7.2. Teleoperation via the World Wide Web

7.3. ARITI-C: a groupware for collaborative teleoperation via the Internet

7.4. Integrating QoS in designing tailorable collaborative teleoperation systems

7.5. Conclusion

7.6. References

Chapter 8: Survivability-Oriented Quality of Service in Optical Networks

Wissam FAWAZ and Ken CHEN

8.1. Introduction

8.2. Optical transport network failures

8.3. Optical network survivability evolution

8.4. Optical WDM-layer survivability mechanisms

8.5. Conclusion

8.6. References

Chapter 9: MAC Protocols for Quality of Service Provisioning in Mobile Ad Hoc Networks

Ghalem BOUDOUR, Mahboub A. BALI and Cédric TEYSSIÉ

9.1. Introduction

9.2. IEEE 802.11 standard basics

9.3. Prioritization-oriented MAC protocols

9.4. Reservation-oriented protocols

9.5. Available bandwidth estimation methods for ad hoc networks

9.6. Conclusion

9.7. References

Chapter 10: Quality of Service Scheduling Mechanisms in Mobile Networks

Mohamed BRAHMA, Abdelhafid ABOUAÏSSA and Pascal LORENZ

10.1. Introduction

10.2. Quality of Service

10.3. Buffer and energy-based scheduling

10.4. Simulations and numerical results

10.5. Conclusion

10.6. References

Chapter 11: Quality of Service inWireless Ad Hoc and Sensor Networks

Azzedine BOUKERCHE, Horacio A.B.F. OLIVEIRA, Eduardo F. NAKAMURA, Richard W.N. PAZZI and Antonio A.F. LOUREIRO

11.1. Challenges for QoS in ad hoc and sensor networks

11.2. QoS parameters in ad hoc and sensor networks

11.3. Components of a QoS system

11.4. MACmeasurement and reservation

11.5. QoS routing discovery and maintenance

11.6. Conclusions

11.7. References

Chapter 12: Quality of Service Challenges in WiMAX Networks

Sahar GHAZAL and Jalel BEN-OTHMAN

12.1. Introduction

12.2.QoS limitations in wireless networks

12.3.QoS features in WiMAXnetworks

12.4. QoS parameter set and management messages

12.5. MAC layer and QoS architecture

12.6. PHY layer supports QoS

12.7. QoS previous proposed solutions for WiMAX

12.8. Conclusion

12.9. References

Chapter 13: Quality of Service Support for MPLS-based Wired-Wireless Domains

Scott FOWLER, Sherali ZEADALLY and Abdelhamid MELLOUK

13.1.Abstract

13.2. Introduction

13.3. MPLS technology

13.4. Mobility and MPLS

13.5.Hierarchical MIP

13.6. Extending MPLS from wired networks to wireless networks

13.7. Multimedia support over MPLS-based networks

13.8. Emerging trends of MPLS-based networks

13.9. Conclusion

13.10. References

13.11. Appendix – list of acronyms

Chapter 14: Quality of Service Control in Voice-over IP Applications

Vincent LECUIRE and Mouna BENAISSA

14.1. Introduction

14.2. General structure of VoIP applications

14.3. End-to-end delay analysis

14.4. Quality of Service requirements for VoIP

14.5. Algorithms for adaptive playout buffering

14.6. Forward error correction mechanisms for packet loss repair

14.7. Joint playout buffering and packet-level FEC algorithms

14.8. Conclusion

14.9. References

Chapter 15: Towards Collaborative Teleoperation Based On Human-Scale Networked Mixed Reality Environments

Samir OTMANE, Nassima OURAMDANE and Malik MALLEM

15.1. Introduction

15.2. Teleoperation and telerobotics

15.3. Augmented reality assisted teleoperation

15.4. Human-scale collaborative teleoperation

15.5. Synthesis and problematics

15.6. References

Chapter 16: QoS-driven Context Awareness Using Semantic Sensors Infrastructure

Abdelghani CHIBANI and Yacine AMIRAT

16.1. Introduction

16.2. Context-aware pervasive computing

16.3. Service agent middleware for decentralized context management

16.4. Context service discovery

16.5. Semantic context sensor scenarios

16.6. Conclusion

16.7. References

Chapter 17: Effect of Transmission Delay on Haptic Perception in Shared Virtual Environments

Hichem ARIOUI

17.1. Introduction

17.2. Haptic simulation in VR applications

17.3. Delayed force feedback systems

17.4. The Quality of Service for a good haptic rendering

17.5. References

List of Authors

Index

First published in France in 2007 by Hermes Science/Lavoisier entitled Mécanismes du contrôle de la qualité de service : applications temps réel et multimédia © LAVOISIER, 2007

First published in Great Britain and the United States in 2009 by ISTE Ltd and John Wiley & Sons, Inc.

Apart from any fair dealing for the purposes of research or private study, or criticism or review, as permitted under the Copyright, Designs and Patents Act 1988, this publication may only be reproduced, stored or transmitted, in any form or by any means, with the prior permission in writing of the publishers, or in the case of reprographic reproduction in accordance with the terms and licenses issued by the CLA. Enquiries concerning reproduction outside these terms should be sent to the publishers at the undermentioned address:

ISTE Ltd 27-37 St George’s RoadLondon SW19 4EUUKwww.iste.co.ukJohn Wiley & Sons, Inc.111 River StreetHoboken, NJ 07030USAwww.wiley.com

© ISTE Ltd, 2009

The rights of Abdelhamid Mellouk to be identified as the author of this work have been asserted by him in accordance with the Copyright, Designs and Patents Act 1988.

Library of Congress Cataloging-in-Publication Data

End-to-end quality of service : engineering in next generation heterogenous networks / edited by Abdelhamid Mellouk.   p. cm.

Includes bibliographical references.

ISBN 978-1-84821-061-5

1. Computer networks—Quality control. 2. Internetworking (Telecommunication)

I. Mellouk, Abdelhamid.

TK5105.5956.E53 2008

004.6#x2014;dc22

2008041895

British Library Cataloguing-in-Publication Data A CIP record for this book is available from the British Library ISBN: 978-1-84821-061-5

Chapter 1

Challenges for End-to-End Quality of Service over Heterogenous Networks

1.1. Introduction

With the advent of new services and the fact that some of these services will be charged per use, performance issues regarding the Internet are perhaps more important than before. Traditional wired/wireless networks can hardly meet the requirements of future integrated-service networks, and are expected to carry multimedia traffic with various Quality of Experience (QoE) and Quality of Service (QoS) requirements. Therefore, it is necessary to develop efficient global control mechanisms that can: a) satisfy QoS requirements while maximizing network resource utilization, and minimizing operational costs on all the types of wireless-mobile networks used to transport flow; b) mix the QoE associated with home, access and core networks to provide the level of QoS expected by users of new services; and c) balance network traffic loads.

The integration of QoE/QoS parameters will certainly increase the complexity of managing and operating heterogenous networks. It becomes a real challenge to address all relevant QoS issues and provide the QoE expected by end users in emerging heterogenous wireless/wired access networks. QoE does not replace QoS but improves end-to-end QoS by providing the quantitative link of user perception. With the emergence of high speed new services among heterogenous networks, major manufacturers and providers need to evaluate and control QoE in order to continue to generate differentiator and added-value for their product and/or service offerings.

For users, and also for operators and Internet service providers, end-to-end quality is one of the major factors to be achieved. In recent years, the new term of “QoE” introduced to end-to-end QoS more clearly captures the experience of the users. In this field, several new developments have occurred. QoS now encompasses a wide range of services, including wider bandwidth for audio and speech, and new multimedia services such as IPTV. Definition of new performance parameters and values, new assessment methodologies and new quality prediction models are needed.

From a number of angles it appears that we have reached a point in the impressive development cycle of the Internet that now requires a major change. However, research and development in these areas is still at an early stage and the space of potential solutions is far from being explored.

New solutions for dealing with current end-to-end transport challenges must be developed. Mainly focusing on the interoperation of schemes deployed between domains, the conception of a new paradigm facilitating the advance of an end-to-end QoS driven flow model and enriched traffic engineering mechanisms at a multi-domain level is necessary. These network models must include topics in their design such as mobility, stability, convergence, scalability, interoperability and interconnectivity of heterogenous broadband network technologies. The goal is a flexible, incrementally deployable solution, which takes into account differences between ISP policies, protocols, legacy network infrastructures, QoE, the use of wired or wireless access networks, etc. Key aspects are the interactions between intra- and inter-domain TE and QoS routing mechanisms.

The main focus of research activities actually will lead to the building of an adaptive end-to-end QoS model based on QoE and including interdomain QoS routing and the enriched traffic engineering (TE) model. True end-to-end QoS must be flexible enough to deal with issues arising inside heterogenous infrastructures across multiple management domains. The global solution must be able to handle both the different QoS provisioning mechanisms and the service specifications.

1.2. Research challenges in end-to-end QoS

Indeed the scientific and industrial research community has actively participated in many research projects providing proof about the existing problems of the current Internet model as well as supplying some partial solutions to them. In addition, current Internet mechanism weaknesses will be fueled by the expected network evolution. New client requirements in terms of safety, resilience, bandwidth, capacity, dynamism as well as new service applications (some of which are already coming up such as triple play, gaming, etc., and others not even envisaged right now) will lead to the whole network to crashing if no substantial changes are introduced in the current network model.

The problem of handling (routing, processing, releasing, dropping, preempting, queuing, etc.) Internet traffic (defined as an interconnection of multiple networks called domains) should be tackled as a complex problem including all issues to be considered when flowing data traffic over the network. The main research challenge lies in how to take the dynamic changes in the resources of communication networks into account to provide end-to-end QoS for individual flows.

However, monitoring any change across the Internet is simply not possible and even desirable, because not all changes are important. Developing an efficient and accurate QoS mechanism is of key importance and the focus of many challenges:

– Intra/interdomain routing management-related problems: iBGP is not flexible enough, and novel approaches will be needed. Understanding the interactions between intra-TE/QoSR and interdomain QoSR/TE and modeling these interactions is vital. In particular (as there is no model for exchanging TE information between domains) the target would be to come up with the first architecture and model for exchanging enriched path state information among domains, in a safe and scalable way.

– Complexity and stability: as end-to-end QoS with more than two non-correlated criteria is NP-complete, how should we reduce the complexity of the proposed approaches? A general problem of large-scale distributed systems such as networks is the ever-increasing complexity of their operation. This complexity is mainly driven by heterogenity. The wide variety of the technologies deployed within a network and their different, if not proprietary, operational paradigms produce a too complex equation for the network operators to solve. Moreover, network operations are typically handled by one or more human operators. Manual control is time-consuming, expensive and prone to errors. Nevertheless, both technologies and needs continue to develop and to grow. The risk is thus that complexity and cost become limiting factors in the future evolution of networks and in the enriched services they are expected to deliver. Moreover, management to network element interactions are mostly limited to configuration functions and/or bulk data retrieval; intelligence is most often outside the strict scope of such environments.

– QoE constraints: independently of the policies of specific organizations, how should we define a control plane with a set of parametric metrics that can be used as constraints on the path selection? From an end user perspective, QoE models aim to provide a better solution to deliver services with the greatest QoS in order to improve the QoE. Indeed, the end users are more interested by the perceived QoE rather than a simple QoS. The QoE covers a long term scale time frame compared to a simple QoS commitment. The QoE not only embraces the single QoS but also the reliability and the continuity of service.

– Scalability, robustness and reactivity: to take into account the network state and its dynamics in the case of irregular and differentiated flows, it should be necessary to design intelligent and adaptive optimizing interdomain routing algorithms. In an extremely large network such as the Internet, it is not effective (both in terms of cost and performance) for a router to measure, store and process all traversing traffic flows in order to provide guaranteed levels of QoS.

– Heterogenous ASes: how should we integrate the differences of technology used in wireless multi-hop ASes (Sensor, Ad-Hoc or Mesh ASes) to the wired ASes?

– End-to-end QoS: end-to-end QoS is meaningful if individual user flows can be distinguished at the user/network interface. However, end-to-end per-flow QoS is only feasible if it is scalable. As discussed earlier, this is difficult to achieve. Per-flow QoS, however, can still be achieved if it is derived from per-class QoS without introducing per-flow state information at the network core.

Regarding all of these challenges, the chapter-by-chapter description of the book is given in the next section.

1.3. Contents

1.3.1. Chapter 2: principles and mechanisms for Quality of Service in networks

With the emergence of new applications (such as teleconferencing, video-on-demand, voice-over-IP, games, etc.), the adaptation to serve applications with acceptable QoS is required. QoS is a general term for an abstraction covering aspects of the non-functional behavior of a system, as opposed to its functional behavior. QoS means so many things to so many people. In general, the users of a system will have requirements both for the functions that are to be performed and for the QoS with which they are performed.

Existing communication networks are very different from one another in terms of topology, complexity, physical media, protocols, covered distances, management architecture and policy, targeted applications, etc. Consequently, the deployment and management of QoS differs from one network class to another. In this chapter, QoS is presented in the context of packet-switched networks where nodes communicate with each other directly or in a multi-hop way.

This chapter presents an overview of the main issues and mechanisms for QoS specification and management. First, we present the terms and definitions relating to QoS. Then, we discuss the QoS parameters (i.e. transfer delay, jitter, bandwidth, reliability, security, cost, etc.) and application characteristics and their QoS requirements.

QoS provisioning and maintenance requires the deployment of various functions (including QoS establishment, admission control, QoS negotiation and renegotiation, resource management, signaling protocols, routing, traffic conformance control, packet classification and marking, traffic shaping, packet scheduling, congestion control, QoS control and monitoring) at different points of the network (particularly in routers). The aim of this chapter is also to provide an overview of these functions. Finally, the chapter gives an overview of the main standard architectures (i.e. IntServ, DiffServ and MPLS) for QoS provisioning in IP networks.

1.3.2. Chapter 3: different approaches to guarantee Quality of Service

In this chapter, we focus on different approaches used to meet the QoS requirements of user applications. The most widely accepted definition from the international community is given by CCITT’s E800 recommendation:

Definition 1: QoS is the collective effect of service performance determining the satisfaction degree of a service user.

The guarantees requested for QoS are generally quantitative. They concern delays or throughputs. They are probabilistic (i.e. the specified constraint on the delay or throughput is met with a given probability) or deterministic (i.e. the specified constraint must be imperatively met). QoS can be seen as a contract between the user who generates flows and the service provider. This contract is a double commitment:

– commitment of the user in the profile to the flows he/she will generate. These flows can be specified by means of a leaky bucket or a minimum inter-arrival time and a packet size in case of sporadic flows for instance;

– commitment of the service provider in the QoS granted to the accepted flows.

For flows having different QoS requirements, it turns out that a uniform processing of packets is inappropriate. A QoS support taking into account the different QoS requirements is needed. To provide the QoS requested by the user, the system considered should manage QoS using the different techniques we will detail in this chapter. The techniques we will present can be used to size the system considered (in the design phase of the system) or to perform an admission control of the flows (in the operational phase of the system).

With regard to guarantees expressed in terms of delay, we describe different techniques to ensure that an end-to-end time constraint is met; the constraint is generally called an end-to-end deadline or global deadline. We then examine how to evaluate a worst case end-to-end response time. The holistic approach, the network calculus and the trajectory approach will be presented with their respective advantages and illustrated by an example. These three approaches can also be applied in case of traffic shaping. Traffic shaping can act on the jitter (cancelation or limitation) or on the throughput (token bucket). We will then show how to provide a probabilistic guarantee with regard to the response time. We will discuss the merits of deterministic and probabilistic guarantees and study an illustrative example.

We will then consider networks more specifically and propose a QoS architecture allowing different layers to share QoS information in order to reach the given QoS target by optimizing resource utilization. In the particular case of mobile ad hoc networks, QoS support is made very complex (NP-hard) by radio interferences. We will propose a solution called QoS OLSR, extending the OLSR routing protocol with QoS support.

Finally, we conclude and present different perspectives leading to new research directions.

1.3.3. Chapter 4: Quality of Service-based adaptive routing approaches

Interest in QoS-based routing has been steadily growing in the networks, spurred by approaches like ATM PNNI, MPLS or GMPLS. A large amount of research has been conducted in a search for an alternative routing paradigm that would address the integration of dynamic criteria. The most popular formulation of the optimal distributed routing problem in a data network is based on a multicommodity flow optimization whereby a separable objective function is minimized with respect to the types of flow subject to multicommodity flow constraints. However, due to their complexity and increased processing burden, a few proposed routing schemes could be accepted for the Internet. For a network node to be able to make an optimal routing decision, according to relevant performance criteria, it requires not only up-to-date and complete knowledge of the state of the entire network but also an accurate prediction of the network dynamics during propagation of the message through the network. This, however, is impossible unless the routing algorithm is capable of adapting to network state changes in almost real time. So, it is necessary to develop a new intelligent and adaptive routing algorithm. This problem is naturally formulated as a dynamic programming problem, which, however, is too complex to be solved exactly. In these adaptive paradigms, the environment is modeled as stochastic (especially links, link costs, traffic, congestion, ASes), hence routing algorithms can take into account the dynamics of the network. However, no model of dynamics is assumed to be given. This means that these approaches have to sample, estimate and perhaps build models of pertinent aspects of the environment in order to support a framework which addresses the identified tradeoffs needed for an overall interdomain routing solution.

The goal of QoS routing is to find a network path which satisfies the given constraints and to simultaneously optimize the resource utilization. The integration of QoS parameters increases the complexity of current routing algorithms. In fact, the problem of determining a QoS route that satisfies two or more path constraints (for example, delay and cost) is known to be NP-complete. One major difficulty is that the time required to solve the multi-constrained optimal path problem cannot exactly be upper-bounded by a polynomial function. As such, much of the focus over the last few years has been on the development of pseudo-polynomial time algorithms, heuristics and approximation algorithms for multi-constrained QoS paths.

In this chapter, we focus upon the topic on QoS routing policies, and their effectiveness on heterogenous wired and wireless networks. Thus, we explore routing algorithms which take into account the dynamic’s change of communication networks and discuss their advantages and drawbacks when compared to traditional solutions. Compared to previous special issues on the integration of QoS, this issue focuses on a tutorial-based approach, and on the evolution of routing decisions applied in heterogenous networks, which may include the real-time adaptation of the routing policy according with the dynamicity of the load level traffic patterns and the topology of the network.

1.3.4. Chapter 5: optical networks: new challenges and paradigms for Quality of Service

New technologies and services such as HDTV and Triple-play are bringing an increasing need for bandwidth, which cannot be satisfied by the classical electronic-based network technologies. Wavelength Division Multiplexing (WDM) is currently the single available technology that is flexible enough to react adequately to the increasing dominance of data traffic. Nonetheless, one of the most important challenges that still need to be tackled is related to the architecture and the management of WDM-based optical networks. This is due to the numerous differences between optical networks and their electronic counterparts.

In this chapter, we provided first a brief introduction to optical networks. We then discussed the exact role that the optical networks are playing in the support of the deployment of a comprehensive network infrastructure. More specifically, we touched on several critical issues relating to QoS provisioning in optical networks. In this context, we presented in detail the GMPLS control architecture, which extends the MPLS technology. GMPLS is deemed a key component of QoS enforcement and effective Internet traffic handling in future optical networks. We also presented problems that are related to and/or induced by the optical networks, such as routing and wavelength assignment problem, and control plane architecture.

1.3.5. Chapter 6: pushing Quality of Service across interdomain boundaries

Pushing QoS across interdomain boundaries appears to be the bottleneck to provide end-to-end QoS guarantees across domains. In this chapter, we introduce the current techniques and the remaining challenges for establishing inter-AS LSPs with QoS guarantees. We describe the working of the current interdomain routing system. We discuss the consequences of path selection and distribution made by the current interdomain routing system on the visibility of the paths. We show that various paths exist today but their QoS is unknown. The current lack of knowledge about the QoS available from BGP routes makes the evaluation of different solutions to the interdomain QoS problem challenging.

We cover the existing signaling extensions to RSVP-TE that support the establishment of inter-AS LSPs, as well as the protection of those LSPs. We also detail the path computation techniques that have been proposed at the IETF. These computation techniques make it possible to calculate the LSP segments within each AS, in order to compose an end-to-end LSP with QoS guarantees when the sequence of ASs to be crossed is known. Finally, we put together these three components, i.e. interdomain routing, LSP signaling and path computation techniques. We show that inter-AS QoS is not out of reach, but that more work needs to be done in specific areas, especially concerning how to calculate feasible QoS paths. We believe that blind computation techniques without QoS hints or techniques coupled with local heuristics have significant limitations. A better approach would be for ASs to propagate QoS information to guide the search towards AS sequences across which feasible QoS paths can be found. Computing an end-to-end QoS path requires finding a trade-off between the amount of QoS information to be distributed across the Internet and the complexity of finding QoS paths. We believe that two subproblems need to be addressed by the community, namely the propagation of the QoS information and the selection of an AS sequence based on the propagated QoS information.

1.3.6. Chapter 7: Internet-based collaborative teleoperation: towards tailorable groupware for teleoperation

This chapter proposes an approach to design tailorable collaborative teleoperation systems via the World Wide Web. The problem is motivated by 1) the growing use of the Internet as a communication channel, 2) the need for a multi-user collaboration during complex teleoperation tasks, 3) the need for tailorable groupware to provide adaptive and flexible collaboration services.

Teleoperation consists of remotely controlling and manipulating robot systems to achieve complex tasks, some of them may be impossible for the human being. The application domains of teleoperation are numerous and present in most of research fields (medical, spatial, nuclear, etc.). Teleoperation systems are traditionally implemented using dedicated communication channels. The displacement of a slave site and/or the master site needs the displacement of materials as well as network reconfiguration linking the two sites. However, Internet-based teleoperation enables an easy delocalization of low-cost operators and interfaces. Section 7.2 presents a state of the art of the functional Internet-based teleoperation systems with time delay, by classifying them into two categories: non-collaborative systems and towards collaborative systems.

The third section presents a Collaborator Agent (CA) model to design collaborative teleoperation architecture via the Internet that takes into consideration the requirements of the teleoperation domain via the Internet and the requirements of the computer supported cooperative work (CSCW) domain. The CA model is based on a compromise between flexibility that groupware should offer, and the guarantee of realizing a teleoperation mission in which the operator needs assistance, and security in the task process. It is the result of applying the 3C (communication, coordination and cooperation) groupware model with multi-agent system concepts. This model is implemented to provide a collaborative Augmented Reality Interface for Teleoperation via the Internet (ARITI-C).

An Internet model enabling us to predict the time of delivery of a message still does not exist, even with the various works in this domain. In general, we talk about a network without QoS. As opposed, thus, to industrial materials that could guarantee the duration of delivery of a message, the Internet therefore has to be considered as a black box in which a message enters and can eventually exit. The lack of QoS can appear as incompatible with real time applications, and in particular applications that use the Internet for the collaborative teleoperation of robots.

Section 7.4 addresses the problem of integrating tailorability and QoS in the design process of groupware architectures, because the Internet-based collaborative teleoperation systems needs significant flows of data (images, video, audio, haptics, etc.) to be shared dynamically and distributed between distant users. In this section we propose an approach to design tailorable groupware architecture based on web service mechanisms and software agents. In fact, to design efficient collaborative teleoperation systems, we may take into account: 1) the design of tailorable software architectures to provide adaptive and flexible groupware; 2) the integration of QoS management in the designed tailorable architectures as a new service in the communication space of the 3C model of groupware.

1.3.7. Chapter 8: survivability-oriented Quality of Service in optical networks

Engineering networks for keeping service continuity and data integrity (i.e., survivability) plays an extremely important role in the design of robust transport networks. Optical WDM networks do not represent an exception to that respect. This is especially true since fiber cuts are prevalent. In fact, since a single fiber would be used to carry a huge amount of data, network survivability together with its impact on network design becomes a critical issue from the optical operator’s standpoint.

In this context, optical fault recovery techniques provide optical operators with the possibility to increase the operational time of their networks. Owing to the crucial need for such techniques, this chapter covers the main protection schemes that have been introduced in the literature to achieve survivability against failure. We start with an exploration of the failure characteristics of the different optical components. Then, we present a comprehensive study on the causes of fiber optic failures. Last but not least, we conclude with a detailed description of the fault recovery mechanisms (referred to the protection schemes) that are being used at the optical layer to improve the availability of the optical networks.

1.3.8. Chapter 9: MAC protocols for Quality of Service provisioning in mobile ad hoc networks

The support of multimedia and real-time applications in mobile ad hoc networks (MANETs) requires efficient and distributed medium access control (MAC) protocols. Unfortunately, characteristics of MANETs such as radio link vulnerability, mobility of stations and lack of centralized coordination and synchronization makes QoS provisioning in MANETs a very challenging task. In this chapter, we present an overview of the research work investigated in the field of MANETs with respect to MAC protocols. We discuss different approaches of QoS provisioning at the MAC level.

QoS provisioning schemes at MAC level are globally categorized as priority-based or reservation-based approaches. Priority-based schemes give higher priority to access the shared medium to stations that have real-time traffic to transmit. These schemes are extensions to the IEEE 802.11 DCF standard. They control the different values of inter-frame spacing and backoff timers to give priority to stations. The decreasing inter-frame and backoff timer, increase the station priority. The protocols in this category that we present are RT-MAC, DCF-PC, IEEE 802.11e, DPS, BB contention, ES-DCF and DB-DCF. In the reservation-based protocols, the channel time is segmented into contiguous frames. Each frame is composed of fixed length time-slots. The basis of these protocols is to give each real-time station a guaranteed periodic access to the wireless channel. The MAC protocol reserves some time slots for each real-time flow. The protocols in this category that we present are FPRP, D-PRMA, SRMA/PA, MACA/PR and RTMAC.

If the reservation procedures are initiated by a source, which needs to reserve some time slots, without any knowledge of the available bandwidth, there is a high probability (particularly when the network load is beyond a particular threshold) that the reservation will fail and, in addition, that bandwidth is wasted in attempting a blind reservation. It is highly recommended (in particular for MANETs where the bandwidth is limited) that reservations should be initiated based on information indicating the available bandwidth (or time slots) between neighbor stations. Such information is generally provided by available bandwidth estimation methods.

Recent years have seen a strong interest in methods for estimating available bandwidth along paths in networks. In MANETs the available bandwidth estimation is much more challenging than in wired networks, mainly due to station mobility, contention for channel access, unreliability of the transmissions, limited energy and the lack of central coordination. Available bandwidth estimation methods can be classified into two main categories: non-intrusive (also called passive) and intrusive (also called active) techniques. The aim of the chapter is to provide an overview of these methods.

1.3.9. Chapter 10: Quality of Service-based scheduling mechanisms in mobile networks

Wireless networks are an even more emergent technology allowing users access to information and electronic services regardless of their locations. The success of this type of network is due to the high interest shown by users, as well as the business and industrial world. Moreover, the rates currently reached with wireless networks make it possible to transfer multimedia flows subjected to significant constraints. Thus, it has become paramount to respect certain constraints such as bandwidth, delay or packet loss. However, the solutions that were introduced into hard-wired networks have become unsuitable for networks using a shred radio medium without a centralized administration.

In this respect, several propositions concerning the study of the QoS in wireless networks and in particular ad hoc networks were carried out in order to define QoS models, medium access protocols, QoS routing protocols and signaling protocols. Therefore, our prime objective was the study of the various mechanisms of QoS. Thus, this work is placed in the framework of the QOS concept and the proposal of mechanisms making it possible to offer optimal solutions to applications sensitive to certain QoS factors. In this chapter, we present a concept of traffic engineering in ad hoc networks. This concept has enabled us to propose some mechanisms offering differentiated services in order to manage the various types of flows traveling the MAC layer of the IEEE 802.11 standard with respect to the quantity of residual energy and the buffer overflow.

In this chapter we present a new mechanism to support the Quality of Service in ad hoc networks. It is based on the use of a QoS scheduler that takes into account two parameters of QoS: the state of the queue and the quantity of available energy. The IEEE 802.11b standard offers at MAC layer level a single First-In-First-Out queue in a Best-Effort (BE) manner. This means that all flows belong to the same class of service and are treated in the same manner as BE flows. Under this mechanism the information processing carried by flows requiring high priority, such as flows from multimedia applications, are similar to those demanding low priority.

The proposed mechanism solves this problem by introducing in its QoS scheduling a Weighted Round Robin (WRR) algorithm taking into account the state of the batteries and a multi-queue system. This latter is made up of two queues. The first one serves high priority packets and the second queue best effort packets. This technique manages the weight attributed to each queue as a function of two parameters: the state of the high priority queue and the rate of the residual energy. When a certain congestion threshold is reached or the battery is in a critical state, the proposed mechanism updates the weight of the queues, in order to serve high priority packets and avoid starvation for low priority packets.

1.3.10. Chapter 11: Quality of Service in wireless ad hoc and sensor networks

QoS enables a network to provide the underlining applications and users with a number of desired services such as resource reservation, different treatment for different types of traffic, and services with guarantees. The necessity of QoS was first noticed in resource constrained wired networks, especially with the increasing number of different types of traffic (e.g., multimedia). However, with the advent of wireless ad hoc networks, a whole new range of requirements, challenges and applications were imposed on QoS-aware protocols. More recently, new kinds of ad hoc networks such as wireless sensor networks (WSNs) are again imposing new challenges and stimulating the development of new interesting and novel solutions to provide QoS in these networks.

However, at the initial stage of development, a number of QoS protocols have been developed for ad hoc and sensor networks. In this chapter, we will take the approach of defining QoS under the viewpoint of these networks and, mainly, comparing and discussing each of the requirements, metrics, challenges and proposed solutions on different layers when applying QoS to these networks. After reading this chapter, the reader will have not only a good overview of QoS when applied to wireless ad hoc and sensor networks, but will also be able to understand and compare how the proposed solutions differ from each other when trying to provide QoS in different scenarios and with different challenges. This chapter assumes that the reader has some basic knowledge of the studied networks.

1.3.11. Chapter 12: Quality of Service challenges in WiMAX networks

In network communications it is essential to guarantee a minimum level of quality for data transferred over the network in terms of delay, latency, jitter or data loss. End-to-end QoS is still a challenging issue especially in wireless networks with unpredictable channel conditions like signal fading, path loss, radio frequency channel interference or the variable link capacity. In the last few years, wireless technology has developed rapidly and thus can be employed on a broadband scale. Used protocols are designed to support traffic from all types of applications like file transfer protocol FTP, video streaming, voice over IP, etc.

The Institute of Electrical and Electronics Engineers (IEEE) has normalized the IEEE 802.16 standard to address Wireless Metropolitan Area Networks (WMANs). The IEEE 802.16-2004 amendment of the standard, which is widely known as WiMAX, is designed with some built-in QoS features through both MAC and PHY layers. The MAC layer is connection-oriented, and thus signaling messages must be exchanged in order to establish a connection between the sender and the receiver. In addition, it classifies the incoming data traffic from different applications into different class of services, each with a predefined QoS parameters. Five types of scheduling services (UGS, rtPS, ertPS, nrtPS and BE) are defined where UGS is used for real time application like VoIP and is already scheduled by the standard. The PHY layer assures QoS through the implementation of either the FDD or TDD duplexing mode. It applies mainly the OFDM technology which reduces interference and facilitates data recovery since it uses multi-subcarriers.

Since many aspects and features of QoS in the IEEE 802.16 standard are still to be modified and enhanced, many researchers propose new QoS architectures and mechanisms which complete that supplied by the standard. For instance, admission control and scheduling of all service classes are not supported and thus many authors propose new admission control policies or scheduling schemes in order to decrease transmission delay and or to maximize throughput.

These chapters provide a thorough study of QoS limitations in wireless networks, while focusing on QoS features supported by the WiMAX protocol. It also summarizes some proposed works in the literature that aim at enhancing the end-to-end QoS in WiMAX networks.

1.3.12. Chapter 13: Quality of Service support for MPLS-based wired-wireless domains

Wireless technologies have experienced an explosive growth in recent years. This trend is clear from the emergence of various wireless devices such as personal digital assistants (PDAs), wireless computers and cellular phones. Wireless networks have also been proliferating at a rapid pace in our society and are increasingly being used as extensions to the wired Internet infrastructure to allow ubiquitous service access anywhere, anytime. These networking developments have also paved the way for a plethora of applications (such as those involving audio and video) many of which have stringent QoS requirements (bandwidth, delay, loss) that must be satisfied. However, it still remains a significant challenge to provide QoS solutions that operate seamlessly over wired-wireless domains and maintain end-to-end QoS with user mobility. The advent of MPLS technology promises to address some of these QoS challenges. We present and discuss various MPLS-based approaches that have been proposed recently to address the support of QoS over wired networks connected to wireless networks (as the last hop). We highlight and discuss the effectiveness and benefits of each of the approaches in minimizing end-to-end QoS degradations (for example, due to large handover delays) when deployed over wired-wireless domains for mobile users. In addition, we also present the limitations associated with each of the proposed MPLS-based approaches.

Wireless technologies have become a fast growing industry. This trend is clear from the emergence of various wireless, portable devices such as PDAs, laptops and cellular phones. With the development of wireless technology, wireless networks have become an integral part of wired networks. To meet the increasing demands for mobile services, wireless providers are currently implementing 3rd Generation (3G) and 4th Generation (4G) networking technologies that are heavily based on the Internet Protocol (IP). To support mobile applications, Mobile IP (MIP) was developed. MIP provides seamless mobility when a Mobile Node (MN) moves across IP subnets. However, MIP was not designed to support fast handovers and seamless mobility in a handover-intensive environment. With the rapid proliferation of wireless networks, the cell radius continues to decrease. Smaller cells result in more handovers from one cell to another cell due to frequent registration updates. To reduce signaling load that results from frequent MIP registration messages when the MN is far away from the home agent (HA), hierarchical registration (otherwise called IP Micro-Mobility Protocols) has been proposed to enhance the basic MIP. Multimedia support over networks has been extensively studied over the last decade. Several QoS approaches and protocols have been proposed and implemented. However, very few of these proposed approaches have been incorporated into commercial products. Consequently, these techniques have not been widely deployed. However, one of the latest traffic engineering technologies capable of providing QoS that has gained wide acceptance over the last few years is Multi--Protocol Label Switching (MPLS) which is supported by many commercial switches and routers on the market. In high-speed wired networking environments, MPLS is being deployed in the Internet backbone to support service differentiation and traffic engineering. In MPLS, the packet forwarding process is performed by means of label swapping. Since labels are short and have fixed length, MPLS can achieve high efficiency compared to conventional IP routing where the longest prefix matching is used.

1.3.13. Chapter 14: Quality of Service control in VoIP applications

Transmission of digitized voice over IP-based packet switched networks emerged in the early 1990s. Some work carried out at the University of Berkeley studied the effects of network delay and delay jitter on real-time streaming applications. At the same period, several packet media tools were produced: VIC (video conferencing tool), VAT (voice audio tool) and WB (white board) at the University of Berkeley, NETVOT (network voice terminal) at the University of Massachusetts, FREEPHONE at the University of Sophia-Antipolis and RAT (robust audio tool) at the University of London. These works led to the standardized RTP (real time protocol) which was first published in 1996. Among the authors of this protocol, Henning Schulzrinne worked on the characterization and reduction of packet losses in IP networks for real-time services including voice traffic. In 1994, Van Jacobson proposed using a small buffer at the receiver side to absorb the network delay variations. Received voice packets are thus buffered, and their playout is delayed to restore the original timing of the voice packets. In the same year, Ramachandran Ramjee et al. proposed four adaptive buffering algorithms for dynamically adjusting the playout delay of voice packets. Between 1993 and 1996, Jean Bolot published a series of papers that characterized the end-to-end delay and loss behavior of voice packets in the Internet. He reported the phenomenon of delay spikes and proposed to combine with the de-jitter buffer mechanism a forward error correction mechanism for packet loss repair.

This chapter firstly describes the real-time constraints of voice applications over IP networks, i.e. over a packet switched network with random packet delay and loss. Later, this chapter provides a survey of the adaptive buffering techniques used at the receiver side for network jitter compensation and also about the forward error correction techniques used at the sender side.

1.3.14. Chapter 15: towards collaborative teleoperation based on human scale networked mixed reality environments

This chapter presents some research tracks to design human-scale collaborative teleoperation environments. The problem is motivated by: 1) the need for multimodal 3D interaction techniques to enhance human robot interaction performances; 2) the need for distributed software architecture to support human scale collaborative teleoperation.

Many applications (nuclear, spatial, medical, etc.), are set to drive the development of human-scale multi-modal mixed reality environments. For these applications to be successful, they must be as efficient as possible allowing the possibility for the user to naturally interact with the remote robots implied in the task. In addition, some complex missions require multi-user cooperation with virtual reality technologies to provide multimodal 3D interactions between users to share remote robots and objects. Recent advances in both VR systems and CSCW technologies have resulted in a convergence of the appearance of the Collaborative Virtual Environment (CVE) systems supporting different forms of collaboration and interaction between users. The collaboration in these systems refers to the simultaneous interactions (cooperative interaction) of multiple users on a virtual object in an immersive or semi-immersive virtual environment.

To study and understand the limitations of the existing teleoperation systems in the different application domains, section 15.2 presents an overview on teleoperation and telerobotic systems in some application domains such as nuclear, spatial, military, underwater and medical.

Section 15.3 shows how mixed reality techniques (virtual and augmented realities) have contributed to circumvent certain problems generally related to the distance separating the operators (users) from the robot that they want to control.

Section 15.4 presents some recent research and experiments to design a human-scale collaborative teleoperation software architecture based on networked mixed reality environments. The experimentation shows that the collaborative teleoperation system involves several media types, such as stereoscopic (video and audio), control commands, haptic information. Different types of media stream can have different QoS requirements.

1.3.15. Chapter 16: Quality of Service driven context awareness using semantic sensors infrastructure

Current and future networked sensors technologies involves distributed wired and wireless networks consisting of large number of sensors, including, for example, active and passive RFID tags, wireless enabled video sensors, etc. Sensor network technologies have recently received strong attention and it has been decided to build frameworks of context aware pervasive computing environments. The integration and exploitation of context sensors in large-scale pervasive computing environments introduces the need for a software infrastructure designed specifically to enable the rapid development, integration and deployment of services which are able to instrument and monitor the sensors and processing entities with high scalability and resolution. In this chapter, we provide a QoS-driven framework for context aware services using a semantic sensor network. In this framework, we try to guarantee a transparent access to contextual knowledge with high QoS. The main QoS parameter taken into account in this work is the availability of the closest context information services in the pervasive environment. In addition, instead of sending to the centralized server complicated knowledge retrieval requests in languages like SQL or SPARQL, the agents or end user applications requesting contextual knowledge will subscribe to the nearest context information service. This will provide the closest contextual knowledge to agent request. The architecture of our framework is based on loosely coupled semantic sensors where every sensor is a standalone entity, called a context service agent, which is in charge of capturing, processing and delivering context semantic information. Our framework is also based on a distributed architecture of services directory caches which guarantee on one hand the availability of services directories and optimizes the access to context services. On the other hand, it makes it possible to significantly reduce the traffic on the network by specializing directory agents. From a functional point of view, when a context service agent is instantiated in the ubiquitous space it sends its service description to its affiliated directory agent. The choice of the directory agent can be based on several QoS parameters. In order to enable semantic interoperability of context services, we adopt an ontology representation model, allowing different service agents to share common contextual knowledge semantics and content with respect to privacy issues. This semantic context knowledge base is a centric virtual entity build up by the composition of context service agents sharing their own knowledge. To validate the feasibility of our framework in pervasive computing environments, we have implemented three scenarios. The first scenario concerns the implementation from FIPA recommendations of a personal agent dedicated to assist salespeople when traveling by automatically discharging them from travel organization tasks. The second and third scenarios under development concern the implementation of two context sensor infrastructures. The first infrastructure is dedicated to monitor a health care mobile robot service while the second is dedicated to rural living lab cooperative services.

1.3.16. Chapter 17: effect of transmission delay on haptic perception in shared virtual environments

The coupling use of force feedback teleoperation systems with virtual reality (VR) techniques have made it possible to implement networked shared haptic interaction between multiple remote users in virtual environments. This operator remoteness, via a communication network, makes it possible to achieve uncommon tasks. Otherwise, this multi-user implementation is inevitably causing transmission delays related to the communication protocol connecting all operators at the remote site. These problems caused by the transmission latency in force feedback systems are difficult and became even more complex after the emergence of new communication technologies such as the Internet. This complexity results from fluctuations due to network congestion or information transmission protocol.

Various concepts have been proposed to overcome this problem. Two main solutions are discussed here, the first being a solution based on computer programming techniques, called remote programming. The other approach is dedicated to classical automatic control schemes.

Firstly, virtual environments have been used by teleoperation systems in various remote-programming and predictive visualization architectures when the transmission delay made it insurmountable to perform delayed tasks. Indeed, a remote-programming system is an architecture that consists of disabling classic control law based on bilateral direct coupling principle between master and slave sites. The coupling connecting both sites is replaced by two local control laws. From the other side, this concept suffered from a lack of force feedback transparency and poor coherence between the operator’s feeling and visual restitutions. On the other hand, the classic bilateral force reflecting controls (based on the passivity or prediction principles) offer interesting alternatives but there are still problems in some cases.

This chapter presents an overview of the effect of transmission time delay on haptic simulation, on the one hand, giving a round-up of the conventional automatic control of such systems to enhance fidelity of force feedback rendering, and on the other hand, tracks to overcome handicaps of these control laws using the QoS techniques in a transmission system protocol.

1.4. Conclusion

This book sets out to provide comprehensive coverage of QoS aspects for heterogenous wireless/wired networks and optical networks. It is clear that the integration of end-to-end QoS parameters will increase the complexity of the algorithms used in heterogenous networks. Thus, there will be QoS relevant technological challenges in today’s emerging heterogenous networks that include different types of networks (e.g., wired, wireless, mobile, etc.).

The book contains 17 chapters and covers a very broad variety of topics. There is a very extensive literature on end-to-end Quality of Service mechanisms, and to give a complete bibliography and a historical account of the research that led to the present form of the subject would have been impossible. It is thus inevitable that some topics have been covered in less detail than others. The choices made reflect in part personal taste and expertise, and in part a preference for very promising research and recent developments in the field of end-to-end QoS technologies.

Finally, I would to thank all contributors to this book for their research and effort.

____________________

Chapter written by Abdelhamid MELLOUK.

Chapter 2

Principles and Mechanisms for Quality of Service in Networks

2.1. Introduction

Any system is built to perform a set of functions for its users. These functions can be called the functional behavior of the system. The performance of each function will take time, will require system resources and is subject to occasional system errors or failures. These and other similar features are the non-functional behavior of the system. Quality of Service (QoS) is a general term for an abstraction covering aspects of the non-functional behavior of a system, as opposed to its functional behavior. In general, the users of a system will have requirements both for the functions that are to be performed and for the QoS with which they are performed [ISO 95, 98].

QoS means many things to many people. Hardware and software venders, consumers, researchers, telecommunications operators, etc., seem to have their own definitions of QoS, but there is no common definition. Of course, multiple objectives yield multiple definitions of the same acronym. To some, it means introducing an element of predictability and consistency into the existing variability of best-effort network delivery systems. To others, it means obtaining higher transport efficiency from the network and attempting to increase the volume of data delivery while maintaining characteristically consistent behavior. Furthermore, to others, QoS is simply a means of differentiating classes of data services, offering network resources to higher-precedence service classes at the expense of lower-precedence classes [FER 98].

Existing communication networks are very different from others in terms of topology, complexity, physical media, protocols, distances covered, management architecture and policy, targeted applications, etc. Consequently, the deployment and management of QoS differs from one network class to another. In this chapter, QoS is presented in the context of packet-switched networks where nodes (which may be hosts, routers, or switches) communicate with each other directly or in a multi-hop way (Figure 2.1). Physical links used by nodes for transmission may be cables (i.e. wired networks) or the air (i.e. wireless networks).

In general, when QoS is concerned, at least two entities are involved: the entity which requires the service and the entity which provides the service. The first entity is called the service (end)user, client or customer. The second entity is called the server or service provider (or simply the network).

Figure 2.1.Simplified model of networks

This chapter is structured as follows. In section 2.2, the main concepts, terms and definitions relating to QoS are presented. Then, in section 2.3, the characteristics of QoS parameters and applications and their QoS requirements are introduced. Section 2.4 briefly presents the main functions and mechanisms to

2.2. Concepts and definitions

2.2.1. Definitions of QoS in a networking context

There is no common or formal definition of QoS spanning the requirements and characteristics of all the networks and networked applications. However, there are a number of definitions proposed in the literature and adopted in multiple organizations and forms such as ISO, ITU-T, IETF, ATM forum and QoS forum. The following definitions are the most commonly cited:

– ISO and ITU-T definition [ISO 95]: “A set of quality requirements on the collective behavior of one or more objects”.

– IETF definition [SHE 97b]: “As the demand for networked real-time services grows, so does the need for shared networks to provide deterministic delivery services. Such deterministic delivery services demand that both the source application and the network infrastructure have the ability to request, set up, and enforce the delivery of the data. Collectively these services are referred to as bandwidth reservation and Quality of Service (QoS)”.

– QoS forum definition [QFO]: “A collective measure of the level of service delivered to the customer. QoS can be characterized by several basic performance criteria, including availability (low downtime), error performance, response time and throughput, lost calls or transmissions due to network congestion, connection set-up time, and speed of fault detection and correction”.

– ATM forum definition [ATF]: “Quality of Service is defined on an end-to-end basis in terms of the following attributes of the end-to-end ATM connection: cell loss ratio, cell transfer delay, cell delay variation”.

– Ferguson and Huston’s definition [FER 98]: “Quality can encompass many properties in networking, but people use quality to describe the process of delivering data in a reliable manner or even somehow in a manner better than normal. Quality can also mean a distinctive trait or distinguishing property, so people also use quality to define particular characteristics of specific networking applications or protocols”.

From the previous definitions and many others not cited here, different aspects are of concern when we talk about QoS, and probably the combination of all these aspects may provide a common definition (unfortunately the definition will be very long): deploy for QoS provisioning and management. Section 2.5 gives an overview of IntServ, DiffServ and MPLS for QoS provisioning in IP networks.

– QoS describes requirements on the behavior of service providers.

– QoS may be described according to multiple parameters (delay, etc.).

– QoS means different levels of user satisfaction.

– QoS involves the networks as well as the applications.

– QoS involves physical devices and terminals as well as software.

– QoS may be considered at different communication layers (physical, data link, network, transport, and application), middleware, etc.

– QoS requires the deployment of various mechanisms (negotiation, resource reservation, scheduling, routing, etc.).

2.2.2. End-to-end QoS

As far as the users are concerned, it is the QoS perceived by them that matters. Hence, all intermediary networks need to cooperate with each other and understand each other’s QoS levels. Thus, the starting point for considering the QoS associated with a system is meeting its user needs. Users will have requirements for the speed with which certain functions are performed, the capacity of the system to perform those functions, the stability and synchronization of video and audio, the accuracy of information, etc. These requirements are user end-to-end requirements in the sense that the users are concerned with the behavior of the system at the points they interact with the system, not with what happens inside. So, when an application specifies its QoS parameters, they are usually in the form of end-to-end QoS (end-to-end delay, end-to-end jitter, etc.). Meeting end-to-end QoS requirements usually requires the cooperation of all the system components that are used to support the end-to-end service.

The local QoS provided by each node (switch or router) on the path should take into account the end-to-end QoS. The latter is considered as the “summation” of the local QoSs. Notice that “summation” does not necessarily mean “addition” because some QoS parameters such as loss rate and availability cannot be added.

2.2.3. Classes (levels) of service

Different qualities of service guarantees are appropriate for different applications. The level of guarantee refers to the level of commitment provided by the guarantee [MIC 99]. To take into account the characteristics and needs of applications and networking environments, different levels of service guarantees have been proposed, ranging from the deterministic to best effort levels.

– Guaranteed (or deterministic) service: in the guaranteed level of agreement between the users and the network, the desired QoS must be guaranteed so that the requested level will be met, barring “rare” events such as equipment failure. This implies that the service will not be initiated unless it can be maintained within the specified limits. Guaranteed service is intended for real-time applications, such as time-critical applications. The QoS is usually achieved by analyzing the worst-case traffic and reserving sufficient resources to provide the requested QoS even under the worst network conditions.

– Best-effort service: all parties use their best efforts to meet the user requirements, but understand that there is no assurance that the QoS will in fact be provided, and no undertaking to monitor the QoS achieved or to take any remedial action should the desired QoS not be achieved in practice. Best-effort does not consider the specific requested QoS parameters of the applications. It is a basic connectivity with no guarantee. Best-effort is suitable for a wide range of networked applications such as general file transfer or e-mail.

It should be noted that on the one hand, it is difficult or very expensive (in practice) to guarantee QoS with 100% satisfaction, and on the other hand best-effort is not suitable for many real-time and multimedia applications. This is why different forms of QoS guarantee levels ranging between guaranteed and best-effort levels have been proposed:

– Predictive service