Special issue: Analysis of Internet traffic and usage traces

Vol. 62, n° 3-4, March-April 2007
Content available on Springerlink

Guest editors
Houssem Assadi, France Telecom R&D, France
Philippe Owezarski, LAAS/CNRS, France
Ben Anderson, University of Essex, UK


Houssem Assadi, Philippe Owezarski, Ben Anderson

Internet 1.0: early users, early uses


France Télécom R&D – 38-40, rue du Général Leclerc, 92794 Issy-les-Moulineaux, France

Abstract We are presenting here a set of tools that enabled us to build a detailed picture of Internet uses, connecting it to users with precisely defined socio-demographic profiles. On the one hand, our work is based on representative panels of the Internet-connected population monitored over time; this allows us to get a sociologically sound vision of Internet uses. On the other, the data we analysed is collected by a software probe installed on users’ PCs, thus allowing us to implement a user-centric approach. We firstly justify this methodological choice with respect to other possibilities (especially server-centric approaches). Then, we present the software modules we developed in order to collect usage data and qualify them. Finally, a few examples of our platform applications are presented: uses typology and Web paths according to the topics of visited Web. These results represent a unique insight into Internet uses in the 2000-2002 period.

Keywords Internet, User behavior, Telecommunication service usage, Traffic metering, Panel, Data analysis, Segmentation, Probe.

Turn-taking patterns in human discourse and their impact on group communication service design


* Communication Systems Group, Swiss Federal Institute of Technology (ETH Zurich) – Gloriastrasse 35, 8092 Zürich, Switzerland;
** Network Services Laboratory, Reykjavik University – Ofanleiti 2, 103 Reykjavík, Iceland;

Abstract Recent studies demonstrated the benefit of integrating speaker prediction features into the design of group-communication services supporting multiparty online discourse. This paper aims at delivering a more elaborate analysis of speaker prediction by analyzing a larger volume of data. Moreover, it tests the existence of speakers dominating speaking time. Towards this end, we analyze tens of hours of recorded meeting and lecture sessions. Our principal results for meeting-like interaction manifest that the next speaker is one of the last four speakers with over 90% probability. This is seen consistently across our data with little variance (standard deviation of 8.71%) independent of the total number of potential speakers. Furthermore, lecture time is in most cases significantly dominated by the tutor. In meetings, although a single dominating speaker is always evident, domination exhibited high variability. Generally, our findings strengthen and further motivate the act of incorporating user-behavior awareness into group communication service design.

Keywords Group dynamics, Speech, Speaker, Linguistic analysis, Human communication, Oral communication, Teleconference.

Multi-level analysis of an interaction network between individuals in a mailing-list

Rémi DORAT *, **, ***, Matthieu LATAPY*, Bernard CONEIN**, Nicolas AURAY***

* LIAFA – CNRS and Université Paris 7 – 2, place Jussieu, 75005 Paris, France ; latapy@liafa.jussieu.fr
** GSPM – CNRS and EHESS – 10, rue Monsieur le Prince, 75006 Paris, France
*** GET/Télécom Paris – 46, rue Barrault, 75013 Paris, France.

Abstract It is well known now that most real-world complex networks have some properties which make them very different from random networks. In the case of interactions between authors of messages in a mailing-list, however, a multi-level structure may be responsible for some of these properties. We propose here a rigorous but simple formalism to investigate this question, and we apply it to an archive of the Debian user mailing-list. This leads to the identification of some properties which may indeed be explained this way, and of some properties which need deeper analysis.

Keywords Human communication, Social behavior, Interaction, Message handling, Complex system, Graph, Sociology, Newsgroup.

Characterizing ADSL customer behaviours by network traffic data-mining

Françoise FESSANT, Joël FRANÇOIS, Fabrice CLÉROT

*France Télécom, division R&D, 2 avenue Pierre Marzin, 22307 Lannion, France

Abstract The very rapid adoption of new applications by some segments of the ADSL customers may have a strong impact on the quality of service delivered to all customers. This makes the segmentation of ADSL customers according to their network usage a critical step both for a better understanding of the market and for the prediction and dimensioning of the network. Relying on a “bandwidth only” perspective to characterize network customer behaviour does not allow the discovery of usage patterns in terms of applications. In this paper, we shall describe how data mining techniques applied to network measurement data can help to extract qualitative and quantitative knowledge.

Keywords ADSL, Traffic metering, User behavior, Telecommunication service usage, Self organization, Segmentation, Data analysis, Kohonen map.

Distribution of traffic among applications as measured in the French METROPOLIS project

Philippe OWEZARSKI*, Nicolas LARRIEU*, Laurent BERNAILLE**, Walid SADDI***, Fabrice GUILLEMIN***, Augustin SOULE**, Kavé SALAMATIAN**

* LAAS-CNRS – 7, Avenue du Colonel Roche, 31077 Toulouse Cedex 4 ; France.
** LIP 6, Université Pierre et Marie Curie – 4 place Jussieu, 75005 Paris ; France.
*** France Télécom, division R&D – 2 Avenue Pierre Marzin, 22300 Lannion ; France.

Abstract We investigate in this paper the time evolution and the composition in terms of applications of traffic in two different networks, namely the Renater network, dedicated to the French academic and research community, and the France Télécom backbone network supporting commercial traffic. For each network, we present the time evolution of traffic in terms of applications, the associated pie charts for global results, as well as, for each detected application, its flow size distribution, that should have an impact on the traffic nature (self-similarity or long range dependence due to the heavy tail of flow size distribution). Based on these results, this paper presents a discussion on the differences between academic and commercial traffic in terms of usage, as well as possible solutions against LRD and its associated degradation of network performance. For traffic analysis, we propose a new method of classifying traffic according to applications, which relies on applicative protocols recognition instead on the IANA ports numbers.

Keywords Telecommunication network, Teletraffic, Traffic metering, Internet, Scientific and technical research, France Telecom, Application, Classification.

Packet filter optimization techniques for high-speed network monitoring


Department of Information Technology (INTEC), Ghent University – Gaston Crommenlaan 8, bus 201, B-9050 Gent, Belgium.

AbstractEffective network monitoring is vital for a growing number of control and management applications typically found in present-day networks. The ever-increasing link speeds and the complexity of monitoring applications’ needs have exposed severe limitations of existing monitoring techniques. A majority of the current monitoring tasks require only a small subset of all observed packets, which share some common properties such as identical header fields or similar patterns in their data. In order to capture only these useful packets, a large set of expressions needs to be evaluated. This evaluation should be done as efficiently as possible when monitoring multi-gigabit networks. To speed up this packet classification process, this article presents different packet filter optimization techniques. Complementary to existing approaches, we propose an adaptive optimization algorithm which dynamically reconfigures the filter expressions based on the currently observed traffic pattern. The performance of the algorithms is validated both analytically and by means of the implementation in a network monitoring framework. The various characteristics of the algorithms are investigated, including their performance in an operational network intrusion detection system.

Keywords Internet, Intranet, Traffic metering, Communication security, Intranet security, Optimization, Filtering, Stochastic method, Performance evaluation, Classification.

Local and dynamic analysis of Internet multicast router topology

Jean-Jacques PANSIOT

LSIIT Pôle API, Université Louis Pasteur – Boulevard Sébastien Brant, BP 10413, 67412 Illkirch cedex, France

Abstract We study data on routers topology gathered using IGMP messages. Although this mechanism is limited to multicast routers we show that it allows getting precise information on the local topology of routers. This information is difficult to obtain with classical tools based on traceroute. Moreover its low cost allows to frequently collect data and to study the dynamics of this topology.

Keywords Internet, Network router, Multicast, Topology, Data analysis, Static characteristic, Dynamic Characteristic.

Open Topics

Joint source-channel coding of the FS1016 coder LSF parameters: application of the channel optimized trellis coded vector quantization

Merouane BOUZID

Laboratoire Communication Parlée et Traitement du Signal (LCPTS), Faculté d’Électronique et d’Informatique, Université des Sciences et de la Technologie Houri Boumediene (USTHB) – BP. 32, El-Alia, Bab-Ezzouar, Alger, 16111, Algérie

Abstract In this paper, we present an optimized trellis coded vector quantization (TCVQ) coding system designed for the effective and robust coding of LSF spectral parameters at low bit rate. The aim of this system, called at the beginning « LSF-OTCVQ Encoder », is to achieve a low bit rate transparent quantization of LSF parameters of the US Federal Standard (FS1016) speech coder. Once the effectiveness of our LSF-OTCVQ encoders (with weighted distance) was proven in the case of ideal transmissions over noiseless channel, we were interested after in the improvement of their robustnesses for real transmissions over noisy channel. To implicitly protect the transmission indices of our LSF-OTCVQ coders incorporated in the FS1016, we used a joint source-channel coding carried out by the channel optimized vector quantization (COVQ) method. In the case of transmissions over noisy channel, we will show that our new encoding system, called “COVQ-LSF-OTCVQ Encoder”, would be able to contribute significantly in the improvement of the FS1016 performances by ensuring a good coding robustness of its LSF spectral parameters.

Keywords Signal theory, Block quantization, Treillis coding, Source coding, Channel coding, Background noise, System robustness, Speech coding, Spectrum line, Coding circuit, Low rate, Digital transmission.

MobileJMS: a global adaptation solution for nomadic applications


GET/Télécom Paris, 46 rue barrault, 75634 Paris Cedex 13

Abstract Middleware technologies have emerged as a key element of future mobile systems. In this paper, we argue that Message Oriented Middleware Systems (MOMs) provides a valuable solution to nomadic environment issues. They are characterized by a communication style allowing a strong decoupling between distributed entities. The first part describes MobileJMS, a MOM system characterized by communication adaptability and context-awareness. Its communication module is based on components which can be plugged and configured dynamically to deal with various events such as bandwidth variation or frequent disconnections. The second part is related to an overall adaptation model that makes adaptation decisions by considering user preferences, application requirements and execution context state. The key function of the model is the selection of the most appropriate policy to the current context.

Keywords Mobile radiocommunication, Middleware, Adaptive system, Variability, Java language, System architecture, Theoretical study, Case study, Experimentation.

Super-orthogonal space-time trellis codes with eight dimensional phase-shift keying signal constellations

Corneliu Eugen D. STERIAN, Andrei Alexandru ENESCU, Ion BANICA

Universitatea Politehnica Bucuresti – Splaiul Independentei 313, 060032 Bucuresti, Romania.

Abstract We propose a new method of designing super-orthogonal space-time trellis codes (SOSTTCs) by labeling each state transition of the trellis diagram with an ordered pair of 2_2 orthogonal matrices instead of only a single 2_2 orthogonal matrix. An 8D instead of a 4D signal constellation is thus used for modulation and the delay elements of the trellis encoder are 4T instead of 2T, where T is the 2D symbol duration. Therefore, at the receiver, the sequence decoding using Viterbi algorithm requires two times less frequent updating of the branch metrics. We investigate by computer simulation the performance of the new SOSTTCs compared to SOSTTCs designed by the known method. The frame error rate (FER) and the bit error rate (BER) simulation results show that the performance of the new SOSTTCs is practically as good as that of the SOSTTCs of similar complexity designed along classical lines, with the advantage that the number of arithmetic and logic operations performed by the decoder per time unit is smaller, which is crucial in high data-rate applications.

Keywords Treillis coding, Phase shift keying, Signal set, Time space, Space diversity, Coded modulation, Orthogonal signal, Matrix method, Viterbi decoding.