John Heidemann / Papers / MAC Stability in Sensor Networks at High Network Densities

MAC Stability in Sensor Networks at High Network Densities
Tyler McHenry and John Heidemann
USC/Information Sciences Institute

Citation

Tyler McHenry and John Heidemann. MAC Stability in Sensor Networks at High Network Densities. Technical Report ISI-TR-2007-628. USC/Information Sciences Institute. [PDF] [alt PDF]

Abstract

Although MAC protocols have been the subject of extensive study, there has been little study of MAC operation as network density (number of neighbors per node) increases. Although network densities are often low (2–12 neighbors), density can rise in special situations (such as hundreds of people attending a conference in the same room) and new deployments (such as dense instrumentation of a structure with a sensor network). In anticipation of these applications, this paper studies the stability of S-MAC as network density increases to densities of 50 to 150 neighbors, well beyond its current design parameters. We present a mathematical model of expected behavior, then use experiments to show the importance of accounting for clock offset. Although offset cannot easily be modeled, we show that a simulation closely matches the experimental data. Finally, we describe how an offset-aware MAC can correct for hardware variation to allow operations at twice the density of current S-MAC. Although the details are specific to S-MAC, the results apply more generally.

Bibtex Citation

@techreport{McHenry07a,
  author = {McHenry, Tyler and Heidemann, John},
  title = {MAC Stability in Sensor Networks at High Network Densities},
  institution = {USC/Information Sciences Institute},
  year = {2007},
  sortdate = {2007-01-01},
  project = {ilense},
  jsubject = {sensornet_subtransport},
  number = {ISI-TR-2007-628},
  month = jan,
  jlocation = {johnh: pafile},
  keywords = {dense network operation},
  url = {https://ant.isi.edu/%7ejohnh/PAPERS/McHenry07a.html},
  pdfurl = {https://ant.isi.edu/%7ejohnh/PAPERS/McHenry07a.pdf},
  myorganization = {USC/Information Sciences Institute},
  copyrightholder = {authors}
}
Copyright © by John Heidemann