Direction des Relations Internationales (DRI)
|
EQUIPE ASSOCIEE |
GENESIM |
|
sélection |
2008 |
|
Equipe-Projet INRIA : Planète |
Organisme étranger partenaire : University of Washington (UW) |
|
Centre de recherche INRIA : Sophia Antipolis Méditerranée |
Pays : États-Unis |
|
|
Coordinateur français |
Coordinateur étranger |
|
Nom, prénom |
Walid Dabbous |
Sumit Roy |
|
Grade/statut |
DR2, responsable EPI Planète |
Professeur |
|
Organisme d'appartenance |
INRIA |
UW Electrical Engineering Department |
|
Adresse postale |
2004, route des Lucioles, BP 93 06902 Sophia Antipolis, Cedex, France |
University of Washington |
|
URL |
http://planete.inria.fr/dabbous |
http://www.ee.washington.edu/faculty/roy/ |
|
Téléphone |
+33492387718 |
+ 1 (206) 221-5261 |
|
Télécopie |
+33492387978 |
+ 1 (206) 543-3842 |
|
Courriel |
Walid.Dabbous@inria.fr |
roy@ee.washington.edu |
La proposition en bref
|
Titre de la thématique de collaboration: Global Environment for wireless Networking Experiments and SIMulations (Environnement global d’expérimentation et de simulation pour les réseaux sans fil). |
|
Description: Evaluation of
new network protocols and architectures is at the core of networking
research. This evaluation is usually performed using simulations, emulations,
or experimental platforms. Simulations allow a fast evaluation process, fully
controlled scenarios, and reproducibility. However, they lack realism and the
accuracy of the models implemented in the simulators is hard to assess.
Emulation allows controlled environment and reproducibility, but it also
suffers from some degree of realism. Experiments allow more realistic
environment and implementations, but they lack reproducibility and ease of
use. So, each evaluation technique has strengths and weaknesses and therefore
they complement one another. However, there is currently no way to combine
them in a scientific experimental workflow. On the other hand, wireless
network protocols are challenging to evaluate mainly due to the high
variability of the channel characteristics and their sensitivity to
interference. Indeed, as the wireless environment is very difficult to
control, repeatable experiments are complex to perform. In addition, a large number
of parameters impact the results of an experiment. It is therefore difficult
to find the subset of key parameters to be taken into account to characterise
a wireless experiment. The full implementation of all these requirements is a
long-term goal of both teams and is larger than the scope of this
collaboration. The objective of this Associated Team is to contribute toward
this goal. In this context, we propose to provide a prototype evaluation environment for wireless experiments. This evaluation environment is
based on a common programming interface between ns-3, Orbit and OneLab. This
prototype will allow running basic wireless networking scenarios on these
three environments and to compare the simulations and experiments’ results.
Based on |
Sumit Roy received the B. Tech.
degree from the Indian Institute of Technology (
M. Philipose, J. R. Smith, B. Jiang, A.
Mamishev, S. Roy and K. Sundara-Rajan, "Battery-free Wireless
Identification and Sensing," IEEE Pervasive Computing Magazine, Spl. Issue
on Energy Harvesting and Conservation, Jan-Mar. 2005, pp. 37-45.
K. Fishkin,
I. Ramachandran, A. Das and
I. Ramachandran and
2. Historique de la collaboration
ns-3 is the follow-up to the wildly
successful ns-2 project. ns-2 was, for many years, the reference network
simulator for IP networks to the point that more than 50% of all network
simulation-related papers published in many conferences and journals used ns-2
to validate their research. Despite (or because of) this success, ns-2 is
showing its age: its architecture suffers from a number of important problems
which could not be solved without a significant overhaul. This lead a number of
researchers to start the development of ns-3 from scratch with NSF funding
(project co-PIs are Tom Henderson from Boeing and
Planète contribution to ns-3
The Planète project-team is interested
in a new generation of simulation tools that support more heterogeneous, yet
closer to reality, models for links and access networks. Modelling the physical
characteristics of the actual transmission media, notably for wireless
networks, is required and now seems reachable for producing simulated results
that would constructively complement experimental results. The project-team is
interested in ns-
Therefore the project-team has been
involved in the ns-3 project from its very early stages. The Planète
project-team contributed to the architecture and the implementation of its core
facilities: most notably, the implementation of the event scheduler, the packet
data structure, the tracing subsystem, important aspects of the object model,
and the default network node programming interface. The project-team also
worked on the first version of the UDP/IPv4 stack. This work was based on YANS
("Yet Another Network Simulator" [2]), which was developed internally
just prior to starting work on ns-3 and from which it is planned to lift the
MAC+PHY 802.11 model for integration in ns-3. The development of MAC/PHY models
started at INRIA in 2005 with the development of an 802.11a/e MAC and PHY model
in ns-2. This model implementation was subsequently ported to the YANS
simulator to validate the architecture of this simulator. In 2006, the Planète
project-team started developing a WiMAX MAC model for ns-3. This effort is
currently focused on the implementation of a subset of components in the model
to perform the required application-level simulation.
UW contribution to ns-3
The UW personnel have a like-minded
interest in developing a simulation framework that allows a researcher to
easily migrate the workflow from simulation to emulation to
experimentation. Some key goals include
the support of lightweight virtual machines on ns-3 simulated wireless
channels, and the use of new ns-3-enabled research architectures, implemented
in the simulator, over testbeds and live networks. The basic architecture is now in place to
enable these capabilities, and UW plans to focus the next year on enabling such
combinations. UW has established ties
with the Emulab project [13] (
3. Impact
The success of long-term research that will
investigate new architectures and approaches to build the foundations for the
future Internet will depend on the availability of networking experimental
facilities that can support, validate and trigger potentially disruptive
research. The goal is to explore clean-slate designs and to test and experiment
at a very large scale user-centric services and concepts which are not
necessarily backward compatible with the existing Internet. The targeted
experimental facilities should allow addressing the challenges of mobility,
security, management, scalability; to test simultaneously and independently
different proposed architectures; to integrate real users for testing new
services; to interact with simulators and/or emulators; to perform effective
measurements; to realize controllable experiments and reproducible benchmarking
of new network architectures. Wired and wireless network virtualization is
important to enable full time sharing of resources while still ensuring
isolation between simultaneous experiments. Another important aspect is
federation: large networking testbeds
would consist in a federation of interconnectable testbeds. The key issue here
is the existence of different capabilities (wired/wireless) and different
needs. Such a global experimental facility should be defined in coordination
with European (e.g. FIRE) and
4. Connaissances
antérieures
The Planète project-team is deeply involved in the field of experimental
testbeds, both from a development and operation standpoint, and from a research
perspective, contributing to the OneLab project that extends the PlanetLab
architecture and operates the PlanetLab Europe platform. Our approach is to
contribute the existing, well-known experimental platform PlanetLab, rather
than trying to rebuild our own infrastructure from scratch. This activity is
done in collaboration with visible international teams involved in the domain:
(
We have also developed the Wireless
Statistical Monitoring tool (called WisMon) that generates real-time statistics
from a unified list of packets, coming from possible different probes. This
tool fulfils a gap on the wireless experimental field: it provides physical
parameters on real time for evaluation during the experiment, records the data
for further processing and builds a single view of the whole wireless
communication channel environment. WisMon is available as open source under the
Cecill license, via http://planete.inria.fr/software/WisMon/. WisMon will be integrated to
OneLab software as a building block toward a global experimental environment.
We also plan to build an experimental wireless networking platform in several
sites in Sophia Antipolis. This platform will be interconnected with the
European OneLab platform through INRIA. The goal is to study the performance in
terms of bandwidth and radio resources utilization in a heterogeneous radio
environment.
ORBIT [12] is, currently, the most
comprehensive wireless experimental testbed available, configured for
open-access experimentation by remote users, and funded over a period of years
by NSF. It seeks to strike a balance between the twin challenges of a) effort/cost
of reliable experimentation in the field (to gather data on network
performance), in the face of the inherent uncertainties in the wireless
environment and b) providing a controlled environment for repeatability. The
testbed consists of a large-scale radio grid of
approximately 400 nodes equipped with multiple radios (802.11,
Bluetooth, 2.5/3G ..) and allows users to download scripts and OS images of
planned experiments onto a subset of nodes in the grid, run experiments, and
gather desired outputs.
The work
proposed in the context of this Associated Team will build on the above
mentioned contributions and will be described in section II hereafter.
5.
References
[1] D.
Dujovne, T. Turletti and W. Dabbous. “Experimental Methodology for Real
Overlays”, ROADS Workshop,
[2] M. Lacage (INRIA), T. Henderson (UW), “Yet
another network simulator”, in Proceedings from the 2006 workshop on NS-2: the
IP network simulator,
[3] M.
Lacage, H. Manshaei, T. Turletti, “IEEE 802.11 Rate Adaptation: A Practical
Approach”, ACM International Symposium on Modelling, Analysis, and Simulation
of Wireless and
[4] J.
Villalón, P. Cuenca, L. Orozco-Barbosa, Y. Seok, T. Turletti, “Cross-Layer
Architecture for Adaptive Video Multicast Streaming over Multi-Rate Wireless
LANs”, in IEEE JSAC Special Issue on Cross-Layer Optimized Wireless Multimedia
Communications, Volume. 25, No 4, May 2007, pp. 699-711.
[5] P.
Ansel, Q. Ni, T. Turletti,”FHCF: An Efficient Scheduling Scheme for IEEE
802.11e”, in ACM/Kluwer MONET journal, Special issue devoted to WiOpt'04, Vol.
11, No. 3, pp. 391-403, June 2006.
[6] Q.
Ni, T. Li, T.Turletti, Y. Xiao, “Saturation Throughput Analysis of error-prone
802.11 Wireless Networks”, Wireless Communications and Mobile Computing
journal, Vol. 5, Issue 8, pp. 945-956, December 2005.
[7] I.
Aad, Q. Ni, C. Barakat, Thierry Turletti, “Enhancing IEEE 802.11 MAC in
Congested Environments”, Computer Communications journal, Vol. 28, Issue 14,
pp. 1605-1617, September 2005.
[8] M.H.
Manshaei, T. Turletti, T. Guionnet, “An Evaluation of Media-Oriented Rate
Selection Algorithm for Multimedia Transmission in MANETs”, EURASIP Journal on
Wireless Communications and Networking, Special Issue on Ad Hoc Networks:
Cross-Layer Issues, Vol. 2005, Issue 5, pp. 757-773, 2005.
[9] G.-R.
Cantieni, Q. Ni, C. Barakat, Thierry Turletti, “Performance Analysis of Finite
Load Sources in 802.11b Multi-rate Environments”, in Computer Communications
Journal, Special issue on Performance Issues of Wireless LANs, PANs, and Ad Hoc
Networks, Vol. 28, No 10, pp. 1095-1109, June 2005.
[10] PlanetLab
web page: http://www.planet-lab.org
[11] OneLab
web page: http://www.one-lab.org
[12] Orbit
web page: http http://www.orbit-lab.org
[13] Emulab
web page: http://www.emulab.net/
[14] Vini web page: http://www.vini-veritas.net/
[15] ns-3
web page: http://www.nsnam.org/
[16] Ettus
Research web page: http://www.ettus.com/
Evaluation of new network protocols
and architectures is at the core of networking research. This evaluation is
usually performed using simulations (e.g., ns-3 [15]), emulations (e.g., Orbit
[12] and Emulab [13]), or on experimental platforms (e.g., PlanetLab [10] and
OneLab [11]). Simulations allow a fast evaluation process, fully controlled
scenarios, and reproducibility. However, they lack realism and the accuracy of
the models implemented in the simulators is hard to assess. Emulation allows
controlled environment and reproducibility, but it also suffers from a lack of
realism. Experimentations allow more realistic environment and implementations,
but they lack reproducibility and ease of use. So, each evaluation technique
has strengths and weaknesses and therefore is complementary. However, there is
currently no way to combine them in a scientific experimental workflow. Typical
evaluation workflows are split into four steps: topology description and
construction, traffic pattern description and injection, trace instrumentation
description and configuration, and, analysis based on the result of the trace
events and the status of the environment during the experimentation. To achieve
the integration of experimental workflows among the various evaluation
platforms, the two following requirements must be verified:
·
Reproducibility: A
common interface for each platform must be defined so that the same script can
be run transparently on different platforms. This also implies a standard way
to describe scenarios including: the research objective of the scenario,
topology description and construction, the description of the traffic pattern
and how it is injected into the scenario, the description and configuration of
the instrumentation and the evolution of the environment during the
experimentation.
·
Comparability: As each
platform has different limitations, a way to compare the conclusions extracted
from experiments run on different platforms, or on the same platform but with
different conditions[1]
must be provided.
Wireless
network protocols are challenging to evaluate mainly due to the high
variability of the channel characteristics and their sensitivity to
interference. It is therefore important to develop a set of environmental
definitions for the evaluation of wireless systems focusing in particular on
the physical layer, in order to identify the key physical parameters that
influence the performance of different upper layer protocols in wireless
systems, such as radio propagation and interference. Based on these key parameters,
it will be possible to define a set a typical physical scenarios that will
enable realistic cross-layer studies of wireless systems. Note that it is not a
trivial task to develop a satisfactory evaluation approach for wireless
experiments in a global environment including experimental platforms such as
PlanetLab or OneLab. Indeed, as the wireless environment is very difficult to
control, repeatable experiments are complex to perform. In addition, a large
number of parameters impact the results of an experiment. It is therefore
difficult to find the subset of key parameters to be taken into account to
characterise a wireless experiment.
In order to obtain an efficient
chain of evaluation (simulation/emulation/experimental platforms), it is
important to integrate and automate some common functions. In particular, this
requires to:
·
Automate the definition
of proper scenario definition taking in consideration available infrastructure
to the experiment.
·
Automate the mapping of
the scenario topology on top of the experimental platform topology (simple
one-to-one node and link mapping is envisaged).
·
Define and provide
instrumentation within the experimental platform to allow users to monitor the
experiments, collect traces of selected events for offline analysis.
·
Measure and provide
access to state variables that characterize the state of the experimental
platform during an experiment.
·
Define an offline
analysis library to compare experiments based on collected traces and state
variables.
The full implementation of these
requirements is a long-term goal for both teams. The objective of this
Associated Team is to contribute to this ambitious project. In the context of
this collaboration we propose to provide a prototype
evaluation environment for wireless experiments. This evaluation
environment is based on a common programming interface between ns-3, Orbit and
OneLab. This prototype will allow running basic scenarios on the considered
environments (ns-3, Orbit and OneLab) and to compare the experiments results.
To complete this work successfully, we will leverage our expertise on wireless
protocols, ns-3 simulator, Onelab and Orbit described in section I.4 here
above.
In a first step toward such
integration, we will focus in 2008 on the enhancement of the ns-3 support for
wireless. The consistent criticism of wireless stacks available in current
releases of ns-2 has been that a) the
abstractions of the PHY and MAC layers
are poor (specifically, they do not support the new trend of cross-layer
inspired stack optimizations) and b) they do not scale and have not been
updated or maintained over many years. Within ns-3, there is renewed effort to
i) incorporate new protocol stacks (e.g. for 802.16, UWB), ii) significantly
upgrade the PHY/MAC for existing ones within ns-2, and iii) apply considerable
software re-engineering specifically targeted at improving run-time support for
larger scale experiments. These enhanced abstractions will allow more
sophisticated experimentation with cross-layer network management approaches to
individual networks, and in future, to definitions (architecture and protocol
stack modifications) of heterogeneous network management. This co-existence will be facilitated by the
emergence of software defined radio platforms such as the USRP [16].
Our goals for integration of ns-3
and Orbit is thus driven by the vision of generating a composite
software-simulation/hardware emulation environment that works to mutual
benefit. For example: a user can design and execute network simulation
experiments within ns-3 aimed at tuning network stack parameters for desired
optimization goals. If possible, the same experiment can then be run on the
Orbit testbed and the outputs compared; this would help to iteratively improve
the abstraction levels of the PHY/MAC in ns-3 code. A second use of such
integration would be to run `small’ scale experiments within ns-3 (for reasons
of complexity) to obtain optimal protocol stack settings; then these settings
are used on a larger scale ORBIT testbed experiment to determine how network
performance scales.
After
these functionalities are implemented, integration work with the OneLab and
Orbit platforms will be started later in 2009.
1. Co-financement
|
ESTIMATION PROSPECTIVE DES CO-FINANCEMENTS |
|
|
Organisme |
Montant |
|
NSF: It is possible to have a
supplement for |
13.000 Euros Funding for one or two students or
postdoctoral researchers to travel to INRIA (Planète) for a summer term, and
to fund Sumit Roy’s visit(s) to INRIA during 2008. |
|
CPER: The Planète project-team is
involved in the Plexus project whose aim is to set-up an experimental testbed
that will be helpful to realize the prototype evaluation environment for
wireless experiments. |
This
project funds only hardware that will be used in the context of this
collaboration. |
|
|
|
|
|
|
|
|
|
|
Total |
13.000 |
2. Echanges
|
ESTIMATION DES DÉPENSES |
Montant (2008) |
|||
|
|
Nombre |
Accueil |
Missions |
Total |
|
Chercheurs confirmés |
1 + 1 |
3.000 |
5.000 |
8.000 |
|
Post-doctorants |
1 |
5.000 |
|
|
|
Doctorants |
1 |
5.000 |
|
|
|
Stagiaires |
|
|
|
10.000 |
|
Ingénieurs |
1 |
|
15.000 |
15.000 |
|
Total |
5 |
|
|
|
|
|
|
- total des co-financements |
13.000 (for UW) |
|
|
|
Financement "Équipe Associée" demandé |
20.000 |
||
The following scientific
exchanges are envisaged:
Mathieu Lacage will visit
One PhD student and one
postdoctoral researcher from UW will
visit the Planète project-team to work on wireless modeling or cross layer
verification for wireless modules in the simulator.
Sumit Roy will visit the
Planète project-team in 2008.
Walid Dabbous and Thierry
Turletti will visit UW in 2008 or 2009.
© INRIA - mise à jour le 19/10/2007