Lecture Notes in Computer Science 5233 Commenced Publication in 1973 Founding and Former Series Editors: Gerhard Goos, Juris Hartmanis, and Jan van Leeuwen Editorial Board David Hutchison Lancaster University, UK Takeo Kanade Carnegie Mellon University, Pittsburgh, PA, USA Josef Kittler University of Surrey, Guildford, UK Jon M. Kleinberg Cornell University, Ithaca, NY, USA Alfred Kobsa University of California, Irvine, CA, USA Friedemann Mattern ETH Zurich, Switzerland John C. Mitchell Stanford University, CA, USA Moni Naor Weizmann Institute of Science, Rehovot, Israel Oscar Nierstrasz University of Bern, Switzerland C. Pandu Rangan Indian Institute of Technology, Madras, India Bernhard Steffen TU Dortmund University, Germany Madhu Sudan Microsoft Research, Cambridge, MA, USA Demetri Terzopoulos University of California, Los Angeles, CA, USA Doug Tygar University of California, Berkeley, CA, USA Gerhard Weikum Max Planck Institute for Informatics, Saarbruecken, Germany
16
Embed
Lecture Notes in Computer Science 5233 - Home - Springer978-3-642-02742-0/1.pdf · Lecture Notes in Computer Science 5233 ... ‘All IP Convergence and Networking’, ... and carry
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Lecture Notes in Computer Science 5233Commenced Publication in 1973Founding and Former Series Editors:Gerhard Goos, Juris Hartmanis, and Jan van Leeuwen
Editorial Board
David HutchisonLancaster University, UK
Takeo KanadeCarnegie Mellon University, Pittsburgh, PA, USA
Josef KittlerUniversity of Surrey, Guildford, UK
Jon M. KleinbergCornell University, Ithaca, NY, USA
Alfred KobsaUniversity of California, Irvine, CA, USA
Friedemann MatternETH Zurich, Switzerland
John C. MitchellStanford University, CA, USA
Moni NaorWeizmann Institute of Science, Rehovot, Israel
Oscar NierstraszUniversity of Bern, Switzerland
C. Pandu RanganIndian Institute of Technology, Madras, India
Bernhard SteffenTU Dortmund University, Germany
Madhu SudanMicrosoft Research, Cambridge, MA, USA
Demetri TerzopoulosUniversity of California, Los Angeles, CA, USA
Doug TygarUniversity of California, Berkeley, CA, USA
Gerhard WeikumMax Planck Institute for Informatics, Saarbruecken, Germany
Demetres D. Kouvatsos (Ed.)
NetworkPerformanceEngineering
A Handbook on ConvergentMulti-Service Networks andNext Generation Internet
13
Volume Editor
Demetres D. KouvatsosUniversity of BradfordSchool of ComputingInformatics and MediaRichmond Road, Bradford, BD7 1DP, UKE-mail: [email protected]
ISSN 0302-9743 e-ISSN 1611-3349ISBN 978-3-642-02741-3 e-ISBN 978-3-642-02742-0DOI 10.1007/978-3-642-02742-0Springer Heidelberg Dordrecht London New York
Typesetting: Camera-ready by author, data conversion by Scientific Publishing Services, Chennai, India
Printed on acid-free paper
Springer is part of Springer Science+Business Media (www.springer.com)
Preface
Over recent years a great deal of progress has been made in the performancemodelling and evaluation of the Internet, towards the convergence of multiser-vice heterogeneous networks, supported by internetworking and the evolutionof diverse access and switching technologies. Performance modelling, evalua-tion and prediction of such networks are of crucial importance in view of theirever-expanding usage and the multiplicity of their component parts and thecomplexity of their functioning.
However, many important and challenging performance-engineering issuesneed to be addressed and resolved, such as those involving heterogeneous networkarchitectures and technology integration, traffic modelling and characterization,management, congestion control, routing and quality-of-service (QoS). The ulti-mate goal is the establishment of a global and wide-scale integrated broadbandnetwork infrastructure for the efficient support of multimedia applications withdifferent QoS guarantees. Of particular interest and challenge is the design andengineering of the next- and future-generation Internets, such as those based onthe convergence of heterogeneous wireless networks enabled by internetworkingand wireless mesh networking technologies. Moreover, of vital interest is the cre-ation of generic evaluation platforms capable of measuring and validating theperformance of networks of diverse technology and multi-service interoperability.In this context, robust quantitative methodologies and performance modellingtools are needed, such as those based on queueing network models (QNMs), inorder to provide a sound theoretical underpinning of application-driven researchleading to credible and cost-effective algorithms for the performance evaluationand prediction of convergent heterogeneous networks under various traffic han-dling protocols.
The principal objective of the Network Performance Engineering handbookis to bring together technical contributions and future research directions in theperformance engineering of heterogeneous networks and the Internet by eminentresearchers and practitioners from industry and academia worldwide. The hand-book consists of 44 extended and revised chapters, which were selected followinga rigorous international peer review. These chapters were drawn from selectedlectures and tutorials of the recent six HET-NETs International Working Con-ferences and associated EU PhD courses on the ‘Performance Modelling andEvaluation of Heterogeneous Networks’. These events took place in Ilkley, UK(July 2003 – 2005, September 2006), Karlskrona, Sweden (February 2008) andZakopane, Poland (January 2010) and were staged under the auspices of theEU Networks of Excellence (NoE) Euro-NGI and Euro-FGI (c.f., Work-packageWP.SEA.6.1) and with the collaboration of EU academic and industrial consor-tia and other international organizations. Moreover, this handbook is part of the
VI Preface
Final Deliverables of NoE Euro-NGI and Euro-FGI to the European Commission(c.f., Deliverable D.SEA.6.1.6c).
The chapters of the handbook constitute essential introductory material forfurther research and development in the performance modelling, analysis, designand engineering of heterogeneous networks and of next- and future-generationInternets. They aim to unify relevant material already known but dispersed inthe literature, introduce the readers to unfamiliar and unexposed research areasand, generally, illustrate the diversity of research found in the high-growth fieldof convergent multiservice heterogeneous networks and the Internet. Moreover,the theoretical themes of the handbook, such as those focusing on traffic mod-elling, quantitative network methodologies and associated performance engineer-ing tools, are also of relevance to the design and development of other types ofdiscrete flow systems such as flexible manufacturing systems and transportationnetworks.
The chapters of the Networks Performance Engineering handbook are broadlyclassified into 12 parts covering the following topics: ‘Measurement Techniques’,‘Traffic Modelling and Engineering’, ‘Queueing Systems and Networks’, ‘Ana-lytic Methodologies’, ‘Simulation Techniques’, ‘Performance Evaluation Studies’,‘Mobile, Wireless and Ad Hoc Networks’, ‘Optical Networks’, ‘QoS Metrics andAlgorithms’, ‘All IP Convergence and Networking’, ‘Network Management andServices’ and ‘Overlay Networks’.
In Part 1, ‘Measurement Techniques’, Popescu and Constantinescu investi-gate Kleinrock’s independent assumption by carrying out measurements, per-formance modelling and analysis of end-to-end delay in a chain of IP (Inter-net Protocol) routers represented by a tandem queueing system with correlatedtraffic flows. Arlos discusses some of the fundamental aspects of performancemeasurements with particular focus on the application-level measurements forthe estimation of the network’s performance properties. Fiedler et al. assess theimpact of application-perceived throughput on the performance of networkedapplications and focus on the process of user-perceived throughput in GPRS(General Packet Radio Service) and UMTS (Universal Mobile Telephone Sys-tem) systems over small averaging intervals and active measurements of stream-ing applications. In Part 2, ‘Traffic Modelling and Engineering’, Markovich andKrieger provide a common methodology for the statistical characterization ofpeer-to-peer packet traffic arising from passive VoIP (voice over IP) and videomeasurements, and consider applications using individual Skype flows and theaggregated flow of video packets exchanged with a mobile peer in an overlay net-work. Nogueira et al. discuss the suitability of MMPP (Markov Modulated Pois-son Processes), and evaluate the credibility of parameter-fitting procedures forthe characterization of Internet packet traffic flows incorporating self-similarityand long-range dependence over multiple time scales. Iovanna et al. propose anadaptive traffic management system in MPLS (Multi-Protocol Label Switching)networks operating on short timescales, and employ an economics-based figureof merit for the relocation of bandwidth. Fretwell and Kouvatsos present thebatch renewal process for modelling both LRD (long range dependent) and SRD
Preface VII
(short range dependent) traffic flows in both discrete and continuous time do-mains, and present applications in the analysis of simple queues and queueingnetwork models. Larijani reviews LAN (Local Area Network) technologies withself-similar and long-range dependent traffic processes, and highlights some mod-elling methods with particular emphasis on pseudo self-similar models. Lianget al. undertake measurements and analytic studies of IP traffic in a WLAN(Wireless Local Area Network) environment, and carry out an investigation intothe characterization on protocol distribution and the modelling of IP packetinter-arrival times. In Part 3, ‘Queueing Systems and Networks’, Walraevens etal. present analytical techniques for the study of discrete-time two-class trafficqueueing systems with priority scheduling disciplines, and determine related per-formance measures via the probability generating functions approach. Balsamohighlights exact and approximate algorithms for the quantitative evaluation ofopen and closed queueing networks with finite capacity, and reviews equiva-lence properties amongst different blocking mechanisms as well as applicationsinto communication networks and distributed computer systems. Anisimov re-ports on asymptotic investigations and a new methodology for the analysis ofqueueing systems and networks with heavy traffic, based on the limit theoremsof the averaging principle and diffusion approximation types. Levy et al. dis-cuss the fundamental principles and properties related to queue fairness fromthe perspective of the relevant applications and carry out a comparative studywith some emphasis on computer communications networks. In Part 4, ‘AnalyticMethodologies’, Pagano provides a heuristic interpretation of basic concepts andtheorems of LDT (Large Deviation Theory) and highlights its applications intothe analysis of single queues and network dimensioning as well as rare eventsimulation. Thomas and Bradley use the Markovian process algebra PEPA tospecify and analyze non-product form parallel queues, which are decomposed intotheir components to obtain, with some degree of confidence, a scalable solution.Harrison and Thomas use the reversed process, based on the RCAT (ReversedCompound Agent Theorem), to derive expressions for the steady state proba-bility distribution of a class of product-form solutions in PEPA for generalizedclosed, queueing networks with multiple servers, competing services and func-tional rates within actions. Kouvatsos and Assi undertake an exposition of the‘classical’ EME (Extensive Maximum Entropy) formalism and generalised NME(Non-extensive Maximum Entropy) formalism in conjunction with their applica-bility to the analysis of queues with bursty and/or heavy tails often observed inperformance evaluation studies of heterogeneous networks and the Internet ex-hibiting traffic burstiness, self-similarity and LRD. Ferreira and Pacheco addressthe usual and level-crossing stochastic ordering of semi-Markov processes, andcarry out comparisons against simulation of processes with a given distributionby employing the sample-path approach. Mitrani applies the spectral expansionmethod to obtain exact solutions for a large class of state-dependent queueingmodels, and illustrates their applicability in the fields of computing, communi-cation and manufacturing systems. Czachorski and Pekergin present the methodof diffusion approximation for the modelling and analysis of single queues and
VIII Preface
networks of stations with general service times and transient states in the pres-ence of general bursty traffic streams such as multimedia transfers in moderncommunication networks. In Part 5, ‘Simulation Techniques’, Dhaou et al. adoptthe ‘Cross Layer’ concept to implement a dynamic simulation tool for the perfor-mance optimisation of network models composed of evolving MANETS (MobileAd Hoc Networks) and Satellites, which can be often reduced to smaller sub-models by decomposition or aggregation methods. M. Vill’en-Altamirano andJ. Vill’en-Altamirano present RESTART, an accelerated simulation techniquefor estimating rare-event probabilities in queueing networks and ultra-reliablesystems, based on the choice of the importance function of the system stateused for determining when simulation retrials are made. In Part 6, ‘PerformanceEvaluation Studies’, Krieger considers a hierarchical algebraic description of aWeb graph with host-oriented clustering of pages, and proposes a computationof the stationary distribution of the underlying Markov chain of a random surfer,based on aggregation/disaggregation procedures and algebraic multigrid meth-ods. Pagano and Secchii address the widespread diffusion of TCP (TransmissionControl Protocol) over the Internet, and introduce simple approaches to describethe dynamics of an individual source over a simplified network model and de-tailed techniques for modelling the behaviour of a set of TCP connections overan arbitrary complex network. Mkwawa et al. analyze an open queueing net-work model (QNM) representing the functional units and application servers ofan IMS (IP Multimedia Subsystem)-based testbed architecture implemented byNokia-Siemens as part of the EU IST (Information Society Technologies) VITALproject and assess the handover process of SIP (Session Initiation Protocol) mes-sages between WLAN and GSM (Global System for Mobile Communications)access networks. Do and Chakka suggest Markovian queueing models, based ongeneralizations of QBD (Quasi Birth and Death) processes and devise steady-state solutions assessing the impact of burstiness and autocorrelation of trafficflows of packets for the performance evaluation of next-generation networks.Chakka and Do present an analytic methodology for the steady state solutionof a complex multi-server Sigma-type queue and its applicability to the perfor-mance evaluation of an optical burst switching multiplexer. Wang et al. presenta detailed review of various handover schemes proposed in the literature andfocus on an analytic model developed for a DGCS (Dynamic Guard ChannelScheme), which manages adaptively the channels reserved for handover calls.Shah et al. employ simulation and analytic methodologies for the performancemodelling and optimisation of DOCSIS (Data-over-cable service interface speci-fication) 1.1/2.0 HFC (hybrid fibre coax) networks with particular focus on thecontention resolution algorithm, upstream bandwidth allocation strategies, flow-priority scheduling disciplines, QoS provisioning and TCP applications. In Part7, ‘Mobile, Wireless and Ad Hoc Networks’, Casares-Giner et al. deal with mo-bility aspects of wireless mobile telecommunication systems and provide somebasic frameworks for mobility models as applied to the performance evaluationof relevant mobility management procedures, such as handover and location up-date. Remondo reviews some of the main enabling technologies of wireless Ad
Preface IX
Hoc networks, including physical and medium access control layers, networking,transport issues and dynamic routing protocols, and discusses proposals thataim at maintaining service level agreements in ad hoc networks either in iso-lation or connected to fixed networks. Popescu et al. present ROMA, a newmiddleware architecture implemented at the application layer, enabling seam-less handover in wireless networks with dynamic combinations of services andunderlying transport substrates (overlay and underlying networks). Mkwawaand Kouvatsos review some current broadcasting methods in MANETs for thecontrol and routing information of multicast and point-to-point communicationprotocols in conjunction with recommendations on how to improve the efficiencyand performance of tree- and cluster-based methods. Popescu et al. report onsome recent developments and challenges focusing on seamless handover, sup-ported by several components such as mobility and connectivity management aswell as Internet mobility, under the auspices of the recent EU research projectsMOBICOME and PERIMETER, based on IMS technology standards. In Part8 ‘Optical Networks’, Atmaca and Nguyen review the infrastructure and evo-lution of MANs (Metropolitan Area Networks) towards OPS (Optical PacketSwitching) networks and highlight performance issues in optical networking inmetropolitan areas in terms of optical packet format, MAC (medium access con-trol) protocol, QoS and traffic engineering issues. Castel-Taleb et al. focus onpacket aggregation mechanisms on the edge router of an optical network andpresent an efficient aggregation mechanism supporting QoS requirements of IPflows. Moreover, analytical models based on Markov chains are devised in or-der to assess the packetization efficiency (filling ratio) and determine its meantime for each class. Mouchos et al. review the traffic characteristics of an opticalcarrier’s OC-192 link, based on the IP packet size distribution, traffic bursti-ness and self-similarity. Under optical burst switching (OBS), a performanceevaluation is undertaken involving the dynamic offset control (DOC) and JustEnough Time (JET) allocation protocols. Moreover, parallel generators of opti-cal bursts are implemented and simulated using the Graphics Processing Unit(GPU) and the Compute Unified Device Architecture. In Part 9, ‘QoS Metricsand Algorithms’, Belzarena and Simon address the estimation of QoS parametersin the Internet using traffic traces, end-to-end active measurements and statis-tical learning tools, and determine the admission control problem from resultsof the many sources and small buffer asymptotics. De Vuyst et al. employ thesupplementary variables approach in the transform domain to analyze a discrete-time single-server queue at equilibrium with a new type of scheduling mechanismfor the control of delay-sensitive and delay-tolerant classes of packet arrivals. InPart 10, ‘All IP Convergence and Networking’, Sun presents an overview of thefundamental concepts and research issues of IP networking with particular em-phasis on the current and next-generation Internets, including some importantmechanisms to control and manage diversity and network resources towards theenhancement of network performance and QoS. Popescu et al. report, in thecontext of the NoE Euro-NGI project ROVER, on some recent advances inthe research and development of multimedia distribution over IP and suggest
X Preface
routing in overlay routing as an alternative solution for content distribution. InPart 11, ‘Network Management and Services’, Akhlaq et al. carry out a per-formance evaluation study of NIDS (Network Intrusion Detection Systems) oncommodity hardware by employing evasive and avoidance strategies on a suit-able test bench, and they implement techniques to simulate real-life normal andattack-like traffic flows. In Part 12, ‘Overlay Networks’, Dragos and Popescuintroduce fundamental concepts of unicast QoS in overlay networks and reportperformance evaluation results by adopting the ORP (Overlay Routing Protocol)developed at BTH (Blekinge Institute of Technology) in Karlskrona, Sweden. Fi-nally, Mkwawa and Kouvatsos review some graph-theoretic-based methods forthe selection of a set of topologically diverse routers towards the provision ofindependent paths, and propose a graph decomposition-based approach for themaximization of path diversity without degrading network performance in termsof latency.
I would like to end this preface by expressing my deepest thanks to the fol-lowing organizations for their support of the HET-NETs International WorkingConferences over the recent years: NoE Euro-NGI and Euro-FGI (EU Com-mission), INFORMS - The Applied Probability Society (USA), EPSRC – TheEngineering and Physical Sciences Research Council (UK), The British Com-puter Society (UK), IEE – The Institute of Electrical and Electronic Engineers(UK), The ACATS Forum - ATS Network Consortium Proprietary (EU) andthe EU IST Consortium ‘VITAL’ consisting of Telekom Austria AG (Austria),Siemens AG, Solinet GmbH & Alcatel SEL (Germany), Teletel SA, Voiceglobesprl, Keletron & University of Patras (Greece), Telefonica I+D (Spain) andthe University of Bradford (UK). Thanks are also extended to the membersof the Advisory Boards and Program Committees of the HET-NETs Interna-tional Working Conferences, as well as to the expert referees worldwide for theirinvaluable and timely peer reviews. Thanks are also due to Is-Haka Mkwawa,University of Plymouth (UK), for his expert technical support and kind up-loading of the Network Performance Modelling handbook to the Springer FTPServer.
Broadcasting Methods in MANETS: An Overview . . . . . . . . . . . . . . . . . . . 767Is-Haka M. Mkwawa and Demetres D. Kouvatsos
ROMA: A Middleware Framework for Seamless Handover . . . . . . . . . . . . . 784Adrian Popescu, David Erman, Karel de Vogeleer,Alexandru Popescu, and Markus Fiedler
Seamless Roaming: Developments and Challenges . . . . . . . . . . . . . . . . . . . . 795Adrian Popescu, David Erman, Dragos Ilie, Markus Fiedler,Alexandru Popescu, and Karel de Vogeleer