Tutorials (Sunday, June 17, 2007)
 

Tutorial 1 (9:00 - 12:30): WiMax System and Mobility Supports

Maode Ma, Nanyang Technological University, Singapore

Abstract:
Recent developments on wireless communication technology have resulted in tremendous innovations to make wireless access networks able to replace the wired access networks with much more bandwidth. As a wireless access network, WiMax has shown great potential to provide broadband transmission services to the residential houses. With the support of mesh networking, WiMax systems can be easily configured as a wireless metropolitan area networks (WMAN). The IEEE 802.16e, which was officially published in February 2006, has further enhanced the ability of WMANs with mobility support. The amended standard specifies the mobile WMANs for combined fixed and mobile broad bandwidth access supporting subscriber stations moving at vehicular speeds operating in licensed bands below 6 GHz. The IEEE 802.16e standard equips service providers with the ability to offer a wide range of new and revolutionary high-speed, mobile wireless applications and services that will greatly improve people’s way of life.

The WiMax system is becoming more attractive to the industry as well as research community. It is significant for this tutorial to provide an instructive overview on the WiMax system as well as the mobility supports in the networks based on the IEEE 802.16e standard. This tutorial is aimed at the researchers from academia and industry, engineers, and technical managers who want to have the knowledge of both of the technical aspect and academic research of WiMax systems and its mobility supports.

In this tutorial, the WiMax system will be explained according to the IEEE 802.16e standard. An introduction on the mobility support in the WiMax networks will be the focus. The tutorial will first provide an introduction and overview on the WiMax system. The physical layer will be introduced as a fundamental knowledge to understand the entire system. Then the basic functions of the MAC layer protocol will be introduced with emphasis on 4 types of service provided in WiMax systems to accommodate multimedia traffic, which will affect the operation of the MAC protocol. Besides the normal functions of the MAC layer protocol, functions of supporting mobility will be the major issue. At last, the potential research topics will be discussed and will be surveyed in order to attract the attentions from the research community.

Bio: Maode Ma (Maode_Ma-AT-pmail.ntu.edu.sg) received his Ph.D. degree in computer science from Hong Kong University of Science and Technology in Hong Kong in 1999. He started his professional career as an engineer in computer industry in 1982 after he has got BE degree from Tsinghua University. Starting from 1991, he was an assistant professor in the Department of Computer Engineering at Tianjin University in China. Since 2000, Dr. Ma has joined the School of Electrical and Electronic Engineering at Nanyang Technological University in Singapore. His current research interests are wireless networking and optical networks. He has delivered 6 tutorials on international conferences in recent years. He has more than 70 academic publications including journal papers, conference papers, and book chapters. He currently serves as an associate editor for the IEEE Communications Letters, IEEE Communications Surveys & Tutorials, International Journal of Vehicular Technology, and International Journal of Wireless Communications and Mobile Computing. He is also a Guest Editor for the special issue titled “Wireless Broadband Access–WIMAX and Beyond” for the IEEE Communications Magazine in May 2007. Due to his much contribution to the research, he has been profiled in the 2007 (24th) Edition of Marquis Who's Who in the World® and in the AcademicKeys Who's Who in Engineering Higher Education (WWEHE).

 

Tutorial 2 (9:00 - 12:30): Tools and Techniques for Corporate Network Measurements, Analysis and Remediation

Jean-Laurent Costeux, France Telecom R&D

Abstract:
The goal of this tutorial is to explore new directions in corporate network measurement, analysis techniques and tools for network monitoring, management, and remediation. We will also apply these techniques to define capacity planning methodologies in order to meet the requirements of corporate applications. In this age of introduction and convergence of new generation of enterprise networking and services solutions, we will especially focus on solutions suitable to mobility and performance problems.

The main topics of this course will include the following:

• Collection, storage & access infrastructure: localization of probes, collection techniques (e.g. event sampling, filtering, aggregation, etc.), passive and active measurements techniques, storage and access (e.g. retention policy, indexing techniques, etc.), frequency and granularity of measurements
• Network data analytics techniques & tools: QoS and performance indicators, clustering, temporal and statistical correlation, causality tracking, trends and prevision (traffic prediction using time series)
• Traffic Characterization (e.g. flow durations, volumes, idle periods…): statistical methods (e.g. density functions, quantile estimations), detection of applications using data mining techniques, link to traffic modeling and analytical models
• Applications to network operations & management: network problem determination, network reliability and performance, root-cause analysis of performance problems, localization of faulty elements, emerging phenomenon detection.

We will apply these topics to different networks, e.g. mobile networks as well as "classical" networks, or access networks as well as core networks. We will observe the different techniques and tools in these cases, and see how these techniques can converge.

Bio: Jean-Laurent Costeux (jeanlaurent.costeux-AT-orange-ftgroup.com) received his M.S. degree in 1996 and Ph.D. degree in Computer Science and Telecommunication Systems in 1999 from the Ecole Nationale Supérieure des Télécommunications (ENST, Paris, France). He is a member of France Telecom R&D's expert pool. His research interests at France Telecom R&D include Traffic Characterization, Capacity Planning, Network Problems Detection and Localization. He is in charge of carrying out a measurement platform to analyse the quality of service and the performance of a running public ADSL network. The relevant problems concern data collection and analysis on a high speed network, in order to obtain statistical results representative of the Internet traffic or of corporate traffic. He is the main author of several papers on P2P and VoIP traffic detection and characterization. He has led several studies on the characteristics of modern enterprise traffic of France Telecom's business clients, in terms of usage, traffic, performance and interactions between applications. He also works on tools and methods to detect and localize performance problems in the networks, based on a large amount of IP traces.

 

Tutorial 3 (9:00 - 12:30): Optimizing Traffic Performance in Converging Heterogeneous Optical-Based Networks Resource and Traffic Engineering

Tibor Cinkler, Budapest University of Technology and Economics, Hungary

Abstract:
The growth of the traffic in networks induced, among others, by new applications and the advance of optical technology have made clear that the networking of the forthcoming decades will definitively relay on optics. Not only the transmission links and the core but also the network nodes as well as the metro and access parts are expected to become optical.

This tutorial gives first an overview of optical networks in general, then it focuses onto the transport (backbone) part starting with a short overview of networking techniques ranging from SONET and SDH through ATM/MPLS, ngSDH/SONET, OTN/DigitalWrapper, 10GbE, MPLambdaS/ASON and GMPLS/ASTN to OBS/OPS. The trends will be presented as well as the advantages and drawbacks of certain networking solutions, with special emphasis on common aspects and solutions reused by different techniques. The focus will be on the traffic and resource engineering capabilities of these.

The tutorial will give an overview of heterogeneous (Multi-Service, Multi-Layer, Multi-Domain, Multi-Provider, Multi-Vendor) networks with emphasis on vertical and horizontal interconnection and integration including the aspects of the Data (User), Management and Control Planes. Then I will discuss and illustrate the problems of Routing, Traffic Engineering and Resilience (Robustness) in such heterogeneous networks, with special emphasis on performance, on automatic distributed operation and on fast adaptivity to changing traffic and networking conditions of the discussed methods (algorithms).

Problems introduced by the networking evolution and the solution alternatives to these problems will be presented trying to answer among others the following questions:

• What is better: static, dynamic or adaptive routing?
• Should it be centralized, or distributed?
• Should network domains be used to make the routing scalable? How?
• What information aggregation strategies and what flooding mechanisms are needed for Multi- Domain Routing where multiple carriers, providers operators are present?
• How to make this high capacity network more resistant to failures (Protection, restoration, fast reroute (FRR), Multi-Path Protection (MPP), p-cycle)?
• How can the performance be improved through traffic engineering (TE)?
• How can multilayer architectures be handled (overlay, peer and augmented interconnection models or the integrated MRN one)?
• How do the layers impact granularity (e.g., sub-lambda granularity)?
• How do Traffic, Wavelength (Lambda) and Waveband grooming improve throughput?
• What about QoS and transparency?
• What is the role of the User (Data), Control and Management Planes and through what interfaces and how do they cooperate?
• What are the most promising services over these networks, e.g., (o)VPN (Virtual Private Network), (o)VON (Virtual Overlay Network), Leased Bandwidth, Leased Lambda, Bandwidth on Demand, Lambda on Demand, multicast, peer-to-peer, GRID?
• What if we have multiple services, multiple layers and multiple domains within a network?

The tutorial is partly based on experience gained from projects NoE e-Photon/ONe, NoE e- Photon/ONe+, IP NOBEL, IP NOBEL II, CELTIC PROMISE, COST 291 and COST 293, funded by the European Union.

Bio: Tibor Cinkler (cinkler-AT-tmit.bme.hu) has received M.Sc.('94) and Ph.D.('99) degrees from the Budapest University of Technology and Economics, Hungary, where he is currently associate professor at the Department of Telecommunications and Media Informatics. His research interests focus on routing, TE, design, configuration, dimensioning and resilience of IP, MPLS, ATM, ngSDH and particularly of WR-DWDM based multi-layer multi-domain heterogeneous networks. He leads a research team of 28 researchers including undergraduate (MSc) and graduate (PhD) students and he is author or co-author of over 120 refereed scientific publications and of 3 patents. Involvement.

 

Tutorial 4 (14:00 - 17:30): Incorporating practical issues in the Design and Planning of Wireless Networks (Practitioners' Tutorial)

Shekhar Srivastava, Schema Inc.

Abstract:
Recent years have seen phenomenon growth and profitability in the wireless operations business. This can be ascribed to the aggressive growth in the customer count. However, the customer count seems to be flattening in the parts of North America, Japan, Europe, etc. At the same time, on one hand, customers are increasingly asking for better download speeds, newer technologies and broader coverage; on the other hand, fierce competition is preventing any possible increase in the cost of wireless access. This is pressing the wireless operators to take a hard look at cost cutting measures and better design alternatives. Attempts at minimizing the operational and capital expenditures are faced by concrete problems, which share resemblances with problems from the general area of "operations research." In this tutorial, we will introduce the audiences to the landscape of wireless network design and discuss some prominent optimization problems that are faced by wireless operators. Various sections of the tutorial are:

• Introduction and Background: We will provide an understanding of various elements in a typical wireless network and their role and relative locations. We will also acquaint the audiences with the problems that have been studied in the literature.
• Most wireless networks run on leased lines for backhaul (cellsite to switch connectivity), and hence we will also discuss the pricing and tariff related information, particularly in the North America (similar pricing models are used in other countries as well).
• BSC and MSC planning: Connections and configurations in a wireless network often lead to overloading of network elements due to incremental design decisions. We will study the problem of BSC and MSC planning, and discuss potential formulations.
• Hub Design: Due to non-linear pricing models for leased lines, lower speed circuits need to be hubbed in order to scale them to higher speed circuits. Such a problem is sometimes referred to as "Hub Design" problem. We will introduce the problem; discuss many variations, conflicting objectives and some possible formulations.
• Challenges for Next Generation Wireless networks: We will consider some of the strong candidates for the next generation wireless standards such as Wi-MAX, LTE, etc. and will discuss their impact on the BSC/MSC planning and the Hub design problems, if any.
• Impact of New services to the Wireless Network Design: In this section, we will discuss new services such as Voice (VoIP), Video and Data, and will consider the challenges that need to be considered for supporting these new and upcoming services.

In Summary, the tutorial will introduce the audiences to the wireless networks and identify and delve into the details for some interesting problems from the domain.

Bio: Shekhar Srivastava (shekhar-AT-schema.com) is currently working in the Transport Network Optimization division of Schema Inc. in New Jersey, USA. He has personally conducted and has been involved in wireless optimization studies for large nation-wide carriers. Before that he was a post-doctoral researcher at INRS-Telecommunication, Montreal, Canada. As a post-doc, he worked on design problems pertaining to IP networks supporting applications with average -case delay requirement. Before that he worked for the RF Engineering Group at Nortel Networks, Kansas City, KS where he focused on throughput estimation of the newly laid down EVDO network as compared to the older 1xRTT network. The estimates were based on actual field measurements of measures such as Ec/Io, handoff, and call drop using statistical tools. His interest are in development of techniques for achieving better performance of computer and communication networks; particularly towards development of algorithms for network optimization, network routing, simulation, performance evaluation and queueing. Shekhar received his PhD from School of Computing and Engineering, University of Missouri-Kansas City, MO, USA, Masters in Telecommunication from Electrical Communication Engineering department at Indian Institute of Science, Bangalore, India and Bachelors in Technology in Electronics and Communication Engg. from College of Technology, Pantnagar University, India. His doctoral dissertation was adjudged with an honorable mention at INFORMS Telecommunication Section Ph.D. Dissertation Award for Operations Research in Telecommunications for 2006. He has more than a dozen papers in peer-reviewed international conferences and journals, including papers in ITC18 and ITC19.

 

Tutorial 5 (14:00 - 17:30): A survey on disruptive QoS architectures providing controllable flow-level performance in IP networks

Sara Oueslati, France Telecom R&D

Abstract:
ISPs are nowadays under increasing pressure to introduce QoS to support voice, video and data services over a converged IP network. The failure of classical QoS models to provide manageable and controllable QoS (e.g. Diffserv, Intserv) calls for a new QoS approach, built on a sound understanding of the fundamental traffic relation: capacity-demand-performance. Controlling performance at flow-level is increasingly recognized as the right direction towards providing QoS in IP networks. The tutorial is divided into the following three parts.

The first part of the tutorial justifies the need for a flow oriented approach to realize QoS, based on lessons drawn from traffic modeling. In particular, we outline the importance of taking into account flow-level dynamics (i.e., flow arrivals and departures) and present fluid traffic models allowing performance prediction for both elastic and streaming applications. An emphasis will be put on flow characteristics (e.g. flow peak rates) that impact network performance, user perception of QoS, as well as the scalability of stateful flow control mechanisms.

The second part and main focus of the tutorial presents disruptive light-weight architectures enhancing the best-effort model, while preserving its simplicity and ubiquity. Flow-oblivious networking (F.P. Kelly et al.) and flow-aware networking (J.W. Roberts et al.) are two interesting approaches providing flow-level QoS, though using radically different control concepts. Flow-oblivious networking is based on enhanced e2e congestion control mechanisms yielding optimal performance and provable convergence, without requiring per-flow state maintenance, but vulnerable to user misbehaviour. Using a combination of flow-aware router mechanisms, namely fair queueing and admission control, flow-aware networking provides comparable performance without relying on user cooperation. Discussing pros and cons of each approach, and examining their possible extensions, leads to a vision of QoS based on an appropriate form of cooperation between traffic control mechanisms implemented by end-systems, offering them useful control options, and by the network, ensuring performance and stability.

The last part of the tutorial extends the survey to solutions that, while recognizing to various degrees the importance of controlling traffic at flow level, are not attached to preserving the simplicity of the best-effort network interface. In these proposals, flow control mechanisms are coupled with either signalling protocols (e.g. recent ITU-T recommendations proposed by BT and Anagran) or Diffserv marking techniques (e.g. IETF Pre-Congestion Notification WG).

Bio: Sara Oueslati (sara.oueslati-AT-orange-ftgroup.com) graduated in Computer Science from the "Ecole Nationale des Science de l'Informatique", Tunis (Tunisia) in 1997. She gained her PhD from the "Ecole Nationale Supérieure des Télécommunications", Paris, in 2000. Her thesis work on the subject of QoS routing for elastic flows in multiservice networks was performed in France Telecom R&D. She received the best student paper award for the conference ITC 16. She joined France Telecom R&D as a research engineer in 2000. Her research has been mainly in the field of performance evaluation and design of traffic control mechanisms for multiservice networks. She received the title of Senior Expert in 2003 and is presently in charge of the "Traffic and Resource Management" research team. She has published around 20 papers and is a member of many conference programme committees in the networking field including INFOCOM, HPSR, ICC and QoS-IP. In the last 3 years she has been active in the ITU-T "Service and Network Operation" group.
 

Tutorial 6 (14:00 - 17:30): Overlay Networks and Teletraffic Issues

Phuoc Tran-Gia, Simon Oechsner, Tobias Hoßfeld, Andreas Binzenhöfer, Universität Würzburg, Germany

Abstract:
Overlays have recently gained much attention due to the wide-spread use and evaluation of Peer-to-Peer services. One significant advantage of overlay networks is that they are independent of the underlying physical network structure. This enables end-users to deploy applications and services at the edge of the network. The network providers have to cope with the fact that these edge-based applications dynamically determine the amount of consumed resources, e.g., bandwidth. A popular example for this is Skype, a VoIP application which performs its own network quality measurements and reacts to quality changes in order to keep its users satisfied. This edge-based intelligence is established via traffic control in the overlay. The new paradigm implicates that teletraffic issues, like dimensioning or load distribution, have to take the overlay architecture and its performance indicators into consideration.

Overlays are logical networks which run on top of the physical network structure. Thus, one logical overlay link, i.e. a connection between two nodes in the overlay, usually spans several physical links. This potential source of inefficiency might be overcome by different optimization approaches as proposed in literature. However, there are still some issues related to overlay performance which must be kept in mind when considering the use of an overlay for a given service. In this tutorial, we will cover some of these aspects and present concepts to solve these problems for selected examples. The main focus of the tutorial will be on the performance evaluation of these exemplary systems. Since self-organization is an important feature of overlay networks we will also show how it can be supported by measurements and the estimation of key characteristics of an overlay. In addition, we will outline the different types of overlay techniques as used today ranging from DHT (distributed hash table) architectures to publish/subscribe systems. The basic architecture and functionality will be explained for current implementations and new developments.

The tutorial is divided into five parts. The first part motivates the usage of overlays and the resulting edge-based intelligence illustrated by the Skype application. An overview of overlay networks and emerging applications as well as the differences to P2P are given. In the second part, we briefly discuss the history of overlay networks and show in detail different overlay techniques applied in today's telecommunication systems. In particular, we focus on distributed hash tables like Pastry and Chord, as well as on publish-subscribe systems like the Java messaging service. The important influence factors on the system performance and the need for quality of service mechanisms at the edge are outlined. The third part is devoted to a selected teletraffic issue. We investigate congestion in overlay networks and demonstrate how to model these overlays and which performance indicators have to be taken into account. Two different examples are shown in order to see how congestion in overlays can be coped with. In the first example, an overlay is used to replace a central database, which requires network dimensioning. In the second example, we consider a vertical handover support system based on overlay techniques. Here, congestion appears due to unbalanced load in the system and self-organization is used to avoid this. The fourth part of this tutorial discusses how the key characteristics for self-organization in overlays can be measured and estimated. In this context, the concept of stochastic scalability and the estimation of the churn rate in a structured overlay are depicted. Finally, the fifth part concludes this tutorial.

Audience: The tutorial addresses PhD students/researchers/engineers which so far have a basic knowledge of overlay techniques and want to design and evaluate services as well as applications based on overlays. The participants need to have moderate knowledge about queuing theory.

Bios:
Phuoc Tran-Gia (trangia-AT-informatik.uni-wuerzburg.de) is professor and director of the Institute of Computer Science at the University of Würzburg, Germany. He is also Chairman of the Advisory Board of Infosim. Previously he was at academia in Stuttgart, Siegen (Germany) as well as industries at Alcatel and IBM Zurich Research Laboratory. Professor Tran-Gia was chairperson of the management committee of the COST 257 action of the European Union entitled "Impact of new services on the performance and architecture of broadband networks". He is also founding director of the multi-university Nortel's "Center of Network Optimization". He is consultant and cooperation project leader with Siemens (ICN Board, Munich, ICM Berlin), Nortel (Texas), T-Mobile International (Bonn), France Telecom (Belfort), European Union (European Science Foundation, Brussels). His current research areas include architecture and performance analysis of communication systems, and planning and optimization of communication networks.

Simon Oechsner (oechsner-AT-informatik.uni-wuerzburg.de) studied at the University of Würzburg and received his Masters Degree in 2004. Since then, he has been a research assistant at the department of Distributed Systems, University of Würzburg. His main area of work is the simulative and analytical performance evaluation. He is active in the European Network of Excellence for the Next Generation Internet (EuroNGI) and in the German Information Technology Society (ITG). His current research interests are the adaptation, modeling and performance evaluation of overlays and P2P networks in an industrial environment, especially in mobile operator core networks. His field of work also includes scheduling and industrial engineering. He has worked in several industry projects with Siemens AG, Berlin, and supervised a project on optimizing conductor board tests with ATG Test Systems.

Tobias Hoßfeld (hossfeld-AT-informatik.uni-wuerzburg.de) is a research assistant at the department of Distributed Systems at the Institute of Computer Science, University of Würzburg since 2003. His research interests include the performance evaluation and optimization of distributed and cooperative systems and overlay architectures, the modeling and analysis of future services especially in mobile networks, and methods for the estimation and characterization of traffic profiles in telecommunication networks. His current focus covers the application, the modeling, and the performance evaluation of overlay techniques in mobile networks, in particular 3G and 4G. Tobias Hoßfeld is involved in several industrial projects (Siemens AG, T-Mobile, France Telecom, Bertelsmann AG) and national, also funded by the German research council, and international research projects. He takes an active part in the German Information Technology Society (ITG), European Cost Actions (COST 279 and COST 290), as well as in network of excellence projects EuroNGI/FGI.

Andreas Binzenhöfer (binzenhoefer-AT-informatik.uni-wuerzburg.de) studied Computer Science in Würzburg, Germany, and at the UCLA in California, USA. Since 2002 he is a research assistant at the department of Distributed Systems at the University of Würzburg. He co-organized the EuroView2006 workshop on "Visions of Future Generation Networks" and currently assists in the process of building a nationwide testbed facility in Germany. He takes an active part in European Cost Actions as well as in different Network-of-Excellence projects including EuroNGI/FGI, REDLARF and AutoMon. He supervises national industrial cooperations with Siemens, AOK, and BMW and additionally acts as a consultant and advisor for Bosch and Datev in the Trend-Scout Project. His current research focus is performance analysis of structured P2P systems and their application to distributed network management, with a particular interest in modeling and understanding the dynamics of deployed P2P overlay networks.