AccessNets. Third International Conference on Access Networks, AccessNets 2008, Las Vegas, NV, USA, October 15-17, 2008. Revised Papers

Research Article

Robust Coverage and Performance Testing for Large-Area Wireless Networks

Download
558 downloads
  • @INPROCEEDINGS{10.1007/978-3-642-04648-3_34,
        author={Caleb Phillips and Russell Senior and Douglas Sicker and Dirk Grunwald},
        title={Robust Coverage and Performance Testing for Large-Area Wireless Networks},
        proceedings={AccessNets. Third International Conference on Access Networks, AccessNets 2008, Las Vegas, NV, USA, October 15-17, 2008. Revised Papers},
        proceedings_a={ACCESSNETS},
        year={2012},
        month={5},
        keywords={},
        doi={10.1007/978-3-642-04648-3_34}
    }
    
  • Caleb Phillips
    Russell Senior
    Douglas Sicker
    Dirk Grunwald
    Year: 2012
    Robust Coverage and Performance Testing for Large-Area Wireless Networks
    ACCESSNETS
    Springer
    DOI: 10.1007/978-3-642-04648-3_34
Caleb Phillips1,*, Russell Senior1,*, Douglas Sicker1,*, Dirk Grunwald1,*
  • 1: University of Colorado
*Contact email: caleb.phillips@colorado.edu, russell@unwirepdx-watch.org, sicker@cs.colorado.edu, grunwald@cs.colorado.edu

Abstract

As large-scale wireless networks continue to proliferate, a reliable way to test coverage and communicate requirements becomes increasingly important. In this paper we discuss concerns and provide guidelines to consider when developing a coverage testing methodology for large-scale wireless networks. We propose a method which complies with these guidelines and apply it to a large municipal mesh network in Portland, Oregon. This approach is at the same time simple, cost-effective, and rigorous. We use commodity hardware to perform both high and low-layer tests at a random sample of points. Our method provides insights into the “expected” performance and coverage of the network. We find that a greater density of nodes is required in Portland to provide the required level of coverage, but that at coverered areas, the performance is within specification. We are able to make these extrapolations with high statistical confidence, on a large network, using only 53 measurement points and a single measurement device which cost less than $200 USD to build.