4th International ICST Conference on Testbeds and Research Infrastructures for the Development of Networks & Communities

Research Article

Model validation through experimental testbed: the fluid flow behavior example

Download524 downloads
  • @INPROCEEDINGS{10.4108/tridentcom.2008.10411,
        author={Adel Aziz and Thomas Huehn and Roger Karrer and Patrick Thiran},
        title={Model validation through experimental testbed: the fluid flow behavior example},
        proceedings={4th International ICST Conference on Testbeds and Research Infrastructures for the Development of Networks \& Communities},
        proceedings_a={TRIDENTCOM},
        year={2010},
        month={5},
        keywords={},
        doi={10.4108/tridentcom.2008.10411}
    }
    
  • Adel Aziz
    Thomas Huehn
    Roger Karrer
    Patrick Thiran
    Year: 2010
    Model validation through experimental testbed: the fluid flow behavior example
    TRIDENTCOM
    ICST
    DOI: 10.4108/tridentcom.2008.10411
Adel Aziz1,*, Thomas Huehn2,*, Roger Karrer3,*, Patrick Thiran4,*
  • 1: EPFL School of Computer and Communication Sciences Lausanne, Switzerland
  • 2: Deutsche Telekom Laboratories TU Berlin 10587 Berlin (Germany)
  • 3: Deutsche Telekom Laboratories TU Berlin 10587 Berlin (Germany
  • 4: EPFL School of Computer and Communication SciencesLausanne, Switzerland
*Contact email: adel.aziz@epfl.ch, thomas@net.t-labs.tu-berlin.de, roger.karrer@telekom.de, patrick.thiran@epfl.ch

Abstract

This testbed practice paper presents our efforts to validate an analytical model for fluid flow behavior in wireless mesh networks with an experimental evaluation. We have developed a fluid model for multihop communication in wireless mesh networks and analyzed it with simulations. Now, we describe our efforts to reproduce the modeled and simulated network with an indoor WiFi mesh network and to measure flow parameters that allow us to verify that the underlying assumptions and the flow behavior can be matched in real networks. Our experiences emphasize the need to gap the bridge between simulations and experimental validation as well as the lack of tools to efficiently validate results. These findings are particularly true in wireless mesh networks where interference is beyond the control of the experiment and where nodes are distributed such that an easy coordination and monitoring of the nodes is not possible.