The aim of the INTERMON project is to enhance the inter ...
Integrated Information System for Inter-Domain QoS
Monitoring, Modelling and Verification
D. Hetzer, I. Miloucheva**, U. Hofmann**, J. Quittek***, F. Saluta****
T-Systems Nova, Deutsche Telekom, Berlin, Germany,
**SalzburgResearch, Salzburg, Austria,
*** NEC Europe, Heidelberg, Germany,
****TILAB Turin, Italy
The paper is focused on the INTERMON toolkit – a visual data mining system for integrated end-to-end and
inter-domain QoS (Quality of Service) monitoring, analysis and modelling of application traffic - developed in
the framework of INTERMON European IST project .The INTERMON toolkit is aimed at automated
measurement and modelling of end-to-end QoS, inter-domain performance and border router traffic in an inter-
domain environment for different time scales as well as visual data mining based on relating of statistics and
models describing end-to-end QoS, inter-domain performance and border router traffic .
Tools for distributed measurement, modelling and visual data mining using relational data base are integrated in
the INTERMON architecture. Concepts like ”spatial composition” of inter-domain QoS and “policy-based”
performance measurement and traffic collection at border routers are considered. The paper compares the
INTERMON inter-domain toolkit with current existing systems in the area of QoS analysis and verification. The
key points of INTERMON architecture discussion address:
- Integration of tools for automated Internet structure analysis, monitoring, modelling and visual data
mining using common data base for the purpose of QoS monitoring and verification in an inter-domain
- Data base design and data mining to support requirements for spatial composition of inter-domain QoS
and automated producing of monitoring, modelling and analysis reports considering different
- Open architecture concept with flexible import/export interfaces for measurement and modelling data
Scenario of INTERMON toolkit usage for integrated QoS measurement, modelling and verification in an inter-
domain environment based on long and short term measurements and modelling is explained.
QoS measurement architectures with different objectives and technologies are developed in research activities
, , , , , , , .
 compares different QoS measurement infrastructures with special focus on monitoring techniques (passive
and active techniques). Example for the active QoS monitoring in an interconnected DiffServ inter-domain
environment is the Qbone Internet2 architecture . An automated event detection for active measurement
systems is discussed in . As a regular service to ISP’s based on active measurements and traffic generation,
the RIPE NCC Test Traffic Measurement Service is proposed  .
Research efforts on advanced QoS monitoring architecture today are addressing the problems of information and
data mining techniques to relate large amounts of measurement data for modelling, simulation and visualization
purposes. Integration of monitoring and modelling data for the purpose of macroscopic QoS data mining is
discussed by CAIDA Macroscopic Internet data collection and analysis project . The CAIDA macroscopic
data mining uses combination of active and passive monitoring tools, such as skitter tool, based on active
probing monitors on different locations to discover and measure global Internet topology , , as well as
passive monitoring tools for relating the workload, , .
Considering the state-of-the-art of QoS analysis, modelling and verification in an inter-domain Internet
environment, there is a challenge for QoS data mining systems which allow:
- Relating of different kinds of measurement and modelling information for inter-domain QoS provision
(such as path resources, traffic and performance, application QoS and traffic)
- Automated inferring of monitoring and modelling information for different kind of QoS parameters and
traffic flows based on measurement scenarios
- Complex specification of measurement scenarios in order to obtain QoS prediction models and
automated producing of QoS measurement and modelling reports for different time scales
- Mapping of application traffic to QoS parameter results by using of application class models (VoIP,
streaming audio and video) in order to study inter-domain admission and resource control
- Obtaining of spatial composition of inter-domain performance metrics and traffic volume
- Data mining techniques and data reduction concepts that support visual mining of large amounts of
inter-domain QoS and traffic measurement data
- Complex performance analysis and problem detection based on automated inferring of passive and
active measurement data obtained by different tools using time dependencies.
To meet these requirements, the INTERMON toolkit is proposed. INTERMON is an inter-domain QoS
information and visual data mining system focused on “policy based” collection of different kind of QoS
measurement and modeling information for the purpose of inter-domain QoS analysis, verification and traffic
flow optimization . Considering inter-domain topology and geographical structures, INTERMON integrates
in a relational data base following kinds of measurement and modeling information:
- Application end-to-end QoS measurement data and statistical models
- Inter-domain performance measurements and statistical models
- Border router traffic flow monitoring data and statistical models.
INTERMON “policy based monitoring” of inter-domain QoS and Service Level Agreement (SLA) is aimed at
integrated monitoring, modelling and verification in different time scales (short and long term).
For this purpose, the INTERMON data mining includes novel concepts, such as:
- Integration of tools for inter-domain Internet structure analysis, QoS monitoring, modelling and visual
data mining using common data base
- Data base design for spatial composition of inter-domain QoS and automated obtaining of
measurement, analysis and modelling reports considering different time scales
- Open architecture concept with flexible import/export interfaces of measurement and modeling data.
The “policy based” monitoring in INTERMON is enhancing the concept of the distributed measurement
architecture (DMA) developed within the AQUILA IST project . AQUILA DMA is aimed at verification and
evaluation of the operation of Internet QoS enabled architectures (DiffServ, MPLS and IntServ) , .
Different measurement techniques based on common data base are integrated in AQUILA DMA in order to
study and verify QoS of applications and to relate them to network performance, admission control and traffic
- active network probing for study of network performance parameters
- passive monitoring of application and traffic flows to verify QoS of applications.
This paper is structured in the following sections. Section 2 describes the functional components of INTERMON
toolkit. Section 3 discusses the concept of INTERMON data base for obtaining of spatial composition of inter-
domain performance parameters to end-to-end QoS and automated QoS analysis in different time scales. Section
4 is addressing the open architecture design with flexible import/export measurement and modelling interfaces.
Section 5 is focussed on scenarios for QoS/SLA analysis and verification in an inter-domain environment.
2. INTERMON integrated toolkit for inter-domain QoS analysis
The functional components of the INTERMON toolkit are integrated based on common data base relating
topological, measurement and modelling information for different kind of parameters (end-to-end QoS, inter-
domain performance metrics, traffic) and Graphical User Interface (GUI). The following figure shows the
functional components of the INTERMON toolkit and their interfaces to the data base and management
operations (i.e. toolkit interaction control).
Inte rmon Modular User Interface for Toolit Integration and Interaction
Mechanisms for Interface Data base operations
Automated Tool to Intermon Tools - policy control
Interaction Control - data min ing functions
Inter- QoS Moni toring QoS Modelling Traffic Visual data
domain Toolkit Monitori ng & mining
End to end Inter-domain End to end Inter-domain Monitoring modelling
- QoS / SLA Visual
Border Router QoS monitoring Performa- QoS performance
data mining scenarios
level and AS dependent on nce prediction models - ISP Border - ISP border
for obtaining spatial
Connectivity application monitoring models router Traffic router Traffic
QoS and SLA - long Measurement Flow Models
different kinds of
requirements -short / - Border router
term - Application traffic matrix
- ISP Visualisation of
Traffic End system
Intermon Data Base – integration of different kinds of topology, QoS and traffic monitoring and
Figure 1: Intermon toolkit – integration of functional components
The following functional components are defined in the INTERMON architecture in order to provide
measurement and modelling information for inter-domain QoS analysis of applications and border router traffic
- Inter-domain structure discovery is aimed to obtain border router level and Autonomous System (AS)
level inter-domain connectivity and topology considering characteristics of ASs and border routers of
inter-domain connections. The inter-domain topology is derived from a given source (AS or user end
system) to specific destination(s) (one or many ASs or user end systems) by geographical and
- QoS monitoring. End-to-end QoS monitoring is intended to obtain QoS parameter values between end-
systems based on specified end-to-end measurement scenario using active or/end passive QoS
measurement techniques. Inter-domain performance monitoring is aimed at measurement of inter-
domain performance metrics, such as delay and loss, based on inter-domain measurement scenario.
Inter-domain performance metrics are defined between two border routers of the inter-domain path.
- QoS modelling components. INTERMON modelling functions are used to obtain statistical modelling
data per measured end-to-end QoS and/or inter-domain performance parameters. Long and short term
modelling data requirements are derived dependent on the measurement scenario specification
(measurement duration and aggregation intervals, measurement sampling frequency). Statistical
modelling data, such as accumulative distribution, autocorrelation function, prediction models, e.g.
Auto Regressive Integrated Moving Average (ARIMA), are obtained per end-to-end QoS and/or inter-
domain performance parameters for a given time interval. The dependencies between the metrics are
quantified by cross correlation analysis. Statistical models are stored in the INTERMON data base in
order to be used for simulation of end-to-end QoS of applications in an inter-domain environment.
Measurement based simulation concept of INTERMON means integration of prediction models derived
from INTERMON measurement scenarios into inter-domain simulation environment including specific
inter-domain and QoS simulation facilities, such as inter-domain routing protocol (BGP-4) and QoS
technologies for inter-domain traffic engineering (DiffServ, MPLS).
- Border router traffic monitoring and modelling. To infer inter-domain QoS and traffic analysis,
INTERMON includes border router traffic monitoring (by ISP provider) derived from the requirements
for IP Flow Import/Export (IPFIX) architecture . Based on the border router traffic measurements
obtained for different scenarios and aggregation intervals, border router traffic flow models and traffic
matrix are obtained. The border router traffic flows are measured considering different granularities
(total traffic volume, traffic per source and destination end systems, traffic per source, destination and
transit autonomous systems, traffic per application class and protocol). Border router traffic
measurements and models are used to provide visual data mining for ISP inter-domain traffic
- Visual data mining. The objective of visual data mining concerning inter-domain QoS/SLA and traffic
engineering is to “translate” the huge amount of monitored and modeled data into useful views
supporting operational decisions of ISP operators and end users. Such views are aimed at topological
and geographical presentation of spatial composition of inter-domain QoS (i.e. mapping of inter-domain
performance parameters to end-to-end QoS), inferring of different kinds of end-to-end application QoS
and inter-domain performance parameters, as well as border router traffic flow and matrix visualization
for inter-domain traffic engineering (considering different flow granularities such as protocol and
The INTERMON users may request the execution of the mentioned functions for inter-domain structure
analysis, monitoring, modelling and visual data mining through the GUI menus and parameters. The user
interface includes management mechanisms for “policy based” data collection, interaction control, and
specification of configuration and interaction parameters.
3. Inter-domain QoS data base for measurement and modelling information
The INTERMON data base defines relationships between different kinds of entities which are stored and
processed with the INTERMON integrated tools for:
- Inter-domain structure discovery
- QoS monitoring (end-to-end QoS and inter-domain performance, border router traffic)
- Modelling of inter-domain and end-to-end QoS as well as border router traffic.
The “IM-user” entity is intended to identify the user of the INTERMON system. Based on the “IM-user” the
rules for sharing of “export/import” inter-domain QoS and traffic information are defined. Each user of the
INTERMON toolkit (ISP operator or application user) builds its “individual” INTERMON data base. Entities
describing inter-domain topology, end-to-end QoS, inter-domain performance and border router traffic data are
included in the INTERMON data base.
“Policy” controlled data collection defines rules for sharing of inter-domain QoS base data repositories
(monitoring, modelling and topology data) between different INTERMON users. Based on this, a “global inter-
domain QoS view” could be obtained combining the “individual inter-domain QoS views”. The concept of
policy controlled data collection allows inter-domain performance measurements and models for QoS / SLA
analysis and verification as viewed from one ISP operator or application end-user to be reused by other ISP
operators or application end-users.
The design of the INTERMON data base is optimised to provide information for:
- inter-domain QoS and SLA verification, especially to obtain spatial composition of inter-domain
performance to the end-to-end QoS in an inter-domain environment
- inter-domain traffic engineering, focussing on border router traffic flow measurements and modelling
considering different flow granularities.
The spatial composition of inter-domain performance (QoS) is base for QoS/SLA verification of applications in
an inter-domain environment and focus of this paper. It is defined by the mapping of the inter-domain
performance measurement and modelling data to the end-to-end QoS measurement and modelling data
considering topological information. Obtaining of spatial composition is supported by definition of entities and
relationships describing the inter-domain topology of the end-to-end connections, i.e. inter-domain connection
entities are linked to inter-domain paths describing the specific end-to-end connection.
The integrated measurement and modelling concept is based on relating of the modelling entities (derived per
end-to-end QoS parameter, inter-domain performance metric and border router traffic flow) to the corresponding
measurement statistics and result entities (describing aggregated statistics results) of the specific measurement
scenario for a given time aggregate.
Entity groups are differentiated in order to explain the complex data base design:
- Topological structure entities describing inter-domain usage and topology
- Border router traffic measurement and modelling entities focussed on border router traffic measurement
scenarios and flow models
- Inter-domain performance measurement and modelling entities addressing measurement scenarios and
models of inter-domain connection
- End-to-end QoS measurement and modelling entities defining end-to-end measurement scenarios and
models belonging to specific end-to-end connection and corresponding inter-domain path. Each end-to-
end connection could have different inter-domain paths.
- Traffic description entities used to specify the generated traffic for the inter-domain and end-to-end
Using the “entity group” approach, the INTERMON data base is divided into data base “parts”.
The following figure illustrates the logical design of the INTERMON data base including entities, groups of
entities and their relationships.
ISP Description entity IM-User entity End system description entity
(ISPident , border router number) (User Id, DB access, password) (end system id, IP name,
End-to-end connection entity
BR description entity Inter-domain link entity (end path id, end source id, end
(BR, BR IP name, BR (BR link id, BR sender id, BR receiver id, BR number, end
configuration) receiver id, link statistics,..) service descr,…)
Interface description entity
(Interface id, BR IP name) End-to-end QoS
Inter-domain performance measurement and
measurement and modelling modelling entity
Border router traffic entitity group group
measurement and (ID-IM-DB data base part) (ES-IM-DB data
modelling entity base part )
Traffic description entity group
Figure 2: INTERMON logical data base model - entity groups and relationships
The INTERMON data base stores the results and the measurement statistics for the measurement scenarios in a
time dependent manner dependent on the specified “measurement aggregation interval”. The measurement
scenario is the main entity, based on which measured parameter, statistics and corresponding statistical models
are related. Dependent on the kind of measurement (end-to-end QoS, inter-domain performance parameter,
border router traffic), there are different types of measurement scenarios. The results of measurement scenarios
are inferred based on time parameters such as measurement aggregation intervals. On this way, spatial
composition of inter-domain performance metrics or border router traffic can be related to the end-to-end QoS of
In the next figure, the end-to-end measurement scenario data base part (entities and relationships) are illustrated:
End-to-end connec tion
(ES sender id, ES receiver id,...)
End-to-end QoS parame ter me asureme nt scenari o
(flow ident, flow group, flo w state, flo w options, mu ltiple x option, start time, end time, measurement
aggregate interval description, measurement frequency, E-QoSP measure ment options, E-QoSP statistics and
modelling options, time aggregate option…)
End-to-end Aggregate d E-
packet QoSP Flow
(packet number, (Aggregate interval E-QoSP E-QoSP E-QoSP
snd time, ident, Accumul ati ve Autoc orrelati on Predicti on
recv time, statistics options, Distribution func tion model
E-QoSP- statistics data description
values....) structures…) (E-QoSP ident (E-QoSP ident
accumulat ive Autocorrelation (E-QoSP ident
distribution data model option,
data structures…) model data
Emul ate d tr affic description data base part
Figure 3: End- to- end QoS parameter measurement and modelling entities with
Each measurement scenario is described with measurement duration, measurement aggregation interval for the
results and statistics type, as well as the measurement frequency. Considering these parameters of the
measurement scenario specification, data mining functions for automated measurement and modelling reports for
different aggregation intervals are defined to interpret the measurement and modelling data (per hour , day,
week, month, year).
The modelling entities, such as accumulative distribution, autocorrelation function, and Auto Regressive
Integrated Moving Average (ARIMA) prediction models, are linked to the measurement results and statistics for
a given measurement aggregation interval. INTERMON data mining functions are specified in order to obtain
automated generation of modelling reports for a different kind of aggregation intervals (e.g. short term : inter-
domain routing, long-term : network planning ).
The relationships between measurement results, statistics and models belonging to different measurement
scenarios are derived from their “time states”, which are calculated from parameters describing measurement
scenarios and their results such as measurement aggregation interval, duration, and measurement frequency.
4. INTERMON open and flexible policy based data collection
The INTERMON data base includes measurement results and statistics obtained from the policy controlled data
collection. The INTERMON monitoring tools use remote meters and adapters for execution of
measurement/monitoring scenarios specified with the monitoring tools.
The remote meters are configured by the monitoring tools to filter their results and to parameterize adapters for
the specified measurement scenarios in order to interact with the INTERMON data base.
The policy based data collection in INTERMON is open for interoperation with other QoS architectures and
flexible to integrate different kind of measurements and statistics into the relational data base.
INTERMON open architecture design is intended to support import/ export measurement and modelling
interfaces between different INTERMON users and towards other QoS monitoring and modelling systems.
The policy based collection of measured data in INTERMON uses meters belonging to the INTERMON QoS
monitoring components, which obtain measurement results and statistics at different collection points: border
routers, end-systems and measurement points selected for inter-domain performance metric collection.
While the structure discovery and modelling toolkits are designed to operate directly with the INTERMON data
base, the monitoring toolkits could use so called adapters for flexible filtering of the measured data and
integrating it to the INTERMON data base.
Adapters allows different interfaces to the INTERMON data base entities considering the requirements for inter-
domain data mining. Adapters could be also used for integration of different kind of external measurement and
modelling data into the INTERMON data base.
Examples for adapters used for integration of external measurement and modelling data into the INTERMON
data base are:
- Traffic trace filter transforming ISP border router traffic traces into INTERMON data base entities (i.e.
border router traffic flow measurement scenarios and results)
- Translator of external QoS measurement data repositories to INTERMON data base entities (i.e.
measurement scenarios and their corresponding results). For instance, such adapter could be written to
transform AQUILA data base measurement data to INTERMON measurement data presentation.
Adapters as well as remote meter are configured by the corresponding tools integrated in the INTERMON GUI
considering the measurement scenarios.
The adapter concept allows to reuse the great amount of QoS monitoring data obtained by other QoS monitoring
architectures developed in European projects , , .
The following figure presents the open and flexible approach of INTERMON data collection architecture:
INTERMON Toolkit GUI , Management and Configurati on
Integrated INTERMON tools :
- Internet structure analysis
- QoS monitoring (end-to-end QoS, inter-domain performance metrics) and QoS modelling
- border router traffic monitoring and modelling
- visual data mining (QoS/SLA verification, spatial composition, inter-do main t raffic engineering)
INTERMON Data Base (IM-DB)
Adapters- Other Adapters Adapters
Fro m IPFIX adapters
Border Router End-to-end QoS Inter-domain Interface
Traffic parameter performance - Adapter to
Measurements measurements metrics other
measurements INTERM ON
user data base
Border router traffic flow End-to-end QoS Inter-domain external
meter meter performance meter
Figure 4: Open and flexible architecture design with adapters for filtering of source
measurement data and export/import interfaces
5. INTERMON toolkit scenario for QoS/SLA verification of applications in an inter-
There are different benefits in using the INTERMON toolkit for QoS monitoring, modelling and forecasting as
well as SLA planning and verification in an inter-domain environment:
- The QoS/SLA verification could be studied automatically for different aggregation intervals (i.e.
INTERMON automated measurement reports per hour, day, week, month, year) to provide insight in the
impact of different factors on the QoS perceived by an application in different time scales.
- Based on spatial composition of inter-domain performance and traffic to end-to-end QoS, the inter-
domain connection and the border router(s) of the inter-domain path, could be identified, which are
responsible for the degradation of end-to-end QoS of applications.
- Integrated measurement and modelling provides the possibility to predict the end-to-end and inter-domain
QoS, to specify and plan the application QoS and SLA for different inter-domain routes. Especially in
case of QoS degradation, there is a need to find an alternative inter-domain path for the provision of end-
to-end services. For this purpose, the monitoring of application QoS for different aggregation intervals is
complemented with measurement based short and long term QoS modelling for the specific inter-domain
- The usage of INTERMON emulated traffic generation allows to obtain QoS measurements for different
load conditions (i.e. multiplexing of application traffic flows) and to plan the SLA in the inter-domain
environment dependent on the border router traffic loads.
- Predictive long and short term models of inter-domain parameters derived from the monitoring data in the
INTERMON data base allow to take decisions on application traffic mapping to inter-domain routing
paths, and border router resources.
A simple scenario is used to illustrate the INTERMON QoS/SLA analysis and verification concept for the Voice
over IP traffic flows considering different inter-domain paths for the end-to-end connection in an inter-domain
environment. A multi homed transit domain (domain C) for VoIP traffic based on multiple border egress nodes
leading to the destination AS of the VoIP traffic is assumed.
The Voice over IP QoS/SLA inter-domain scenario is given in the following figure:
end-to-end QoS Application Client
Application Ser er
Do main A
Transit Domain C
Transit Domain B
Border Domain Router Inter-domain QoS analys is with policy controlled data
colle ction (s ome obje ctives)
Edge Domain Router
Measure d inter-domain parameter
Inter Domain Traffic
Ingress / egress QoS of domain as black box
End acces dev ice monitor
egress/ingress QoS interf ace
monitoring inf ormation
inter-domain QoS considerin g at least one transit domain
Figure 5: Scenario for inter- domain QoS and SLA verification of VoIP application
The following diagram shows the usage of INTERMON functional components for the inter-domain SLA
analysis and verification:
VoIP QoS monitoring for given
Degradation check of end-to-end QoS for a given
aggregation interval and inter-domain route
discovery to find
Obtaining of spatial composition for end-to-end QoS other possible
- inter-domain parameter analysis for aggregation interval in optimal inter-
order to detect the cause of QoS degradation domain route for
Found Inter-domain yes
connection or border
router as cause for
QoS degradation ?
Integrated measurement and predictive modelling
of end-to-end QoS and inter-domain performance
QoS predictions are not
satisfactory SLA ?
Inter-domain structure discovery to find other possible
optimal inter-domain route for VoIP traffic
Figure 6 Usage of INTERMON components for inter- domain QoS and SLA verification
for VoIP application traffic
The scenario designed to support “optimal” inter-domain QoS / SLA analysis and verification could have more
variants if inter-domain QoS interfaces (DiffServ, MPLS) are considered which allow different traffic mapping
strategies at the egress border router.
The integrated inter-domain QoS monitoring, modelling and simulation concept developed in the framework of
the INTERMON project, was presented.
The focus was the novel concept of inter-domain data mining in INTERMON and the data base design for
integrated measurement and modelling tools. The use of the INTERMON data base for the spatial composition
of inter-domain performance to the end-to-end QoS was discussed and scenario for inter-domain QoS/SLA
verification based on INTERMON toolkit was proposed.
Another main usage of INTERMON toolkit is the inter-domain traffic engineering  based on border router
measurement and modelling using IPFIX . The border router traffic measurement and modelling as well as
visual data mining for inter-domain ISP traffic engineering in the framework of INTERMON project is topic of
 Advanced architecture for INTER-domain quality of service MONitoring, modelling and
visualisation, INTERMON-IST-2001-3412 http://www.ist-intermon.org
 Co-operative Association for Internet Data Analysis (CAIDA) http://www.caida.org
 Surveyor: An Infrastructure for Internet Performance Measurements
 B. Cheswick. Internet mapping project. http://www.cs.belllabs.com/who/ches/map/
 RIPE Test Traffic project http://www.ripe.net/ripencc/mem-services/ttm
 The Skitter project. http://www.caida.org/Tools/Skitter.
 T. Hansen, J. Otero, T. McGregor, H.-W. Braun. Active Measurement Data Analysis Techniques
 Internet 2 Qbone http://www.internet2.edu/qos
 F Strohmeier, H. Dörken, B. Hechenleitner. AQUILA Distributed QoS Measurement. In Proc. of
COMOCON8 Conference, Crete, Greece, pages 177-185, June 2001.
 AQUILA project page http://www.ist-aquila.org
 I. Miloucheva, U. Hofmann, F. Strohmeier, T. Pfeiffenberger, Integrated toolkit for performance
analysis in intra- and inter-domain environment, The 10th International Conference on
Telecommunication Systems, Modeling and Analysis Monterey, CA, USA, October 3-6, 2002
 M. Murray, KC Claffy, Measuring the Immeasurable: Global Internet Measurement Infrastructure,
Proceedings PAM 2001, p. 159-167, Amsterdam, April 2001
 A.J. McGregor, H.-W. Braun, Automated Event Detection for Active Measurement Systems,
Proceedings PAM 2001, Amsterdam, April 2001
 CAIDA skitter monitor locations. http://www.caida.org/tools/measurement/skitter/monitors.xml.
 Requirements for IP Flow Export, J. Quittek, T. Zseby, B. Claise, G. Carle, S. Zander, K.C.
Norseth, August 2002 (work in progress), Internet-Draft http://www.ietf.org/internet-drafts/draft-
 The CADENUS Project (1999) Creation and Deployment of End-User Services in Premium IP
Networks, http://www.cadenus.org/ (March 10, 2002)
 Traffic Engineering for Quality of Service in the Internet, at Large Scale http://www.ist-tequila.org
 A. Nassri (ed.) Requirements and Analysis Report, INTERMON Deliverable 1, July 2002,