Issue No. 11 - November (2009 vol. 20)
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/TPDS.2008.256
Pan Li , University of Florida, Gainesville
Nicola Scalabrino , University of Florida, Gainesville
Yuguang (Michael) Fang , University of Florida, Gainesville and Xidian University, Xi'an
Enrico Gregori , Italian National Research Council (CNR) - IIT, Pisa
Imrich Chlamtac , CREATE-NET, Trento
Operating on a frequency band occupying several nonoverlapping channels, IEEE 802.11 is now widely used in Wireless Mesh Networks (WMNs). Many multichannel MAC protocols are proposed to improve the spatial reuse in the network under the assumption that the transmissions on nonoverlapping channels do not interfere with each other. Some joint routing and channel assignment algorithms are also designed to increase the network throughput based on the premise that we can switch between different channels freely. Although simulations show that great improvements on network throughput can be observed in both cases, two fundamental questions remain: 1) Can we really use multiple nonoverlapping channels freely in WMNs? 2) If we can, what will be the cost when we switch channels dynamically and frequently? In this paper, by conducting extensive experiments on our testbed, we attempt to answer these questions. We find that in spite of interference between both overlapping and nonoverlapping channels, we can still use multiple channels in mesh networks under certain conditions but with care. We also show that the channel switching cost is actually very significant in WMNs. We recommend not to switch the channels too frequently when designing the channel assignment algorithms, and those channel assignment algorithms selecting one channel for each packet are not really beneficial.
Multiple channels, channel interference, channel switching cost, wireless mesh networks.
Pan Li, Nicola Scalabrino, Yuguang (Michael) Fang, Enrico Gregori, Imrich Chlamtac, "How to Effectively Use Multiple Channels in Wireless Mesh Networks", IEEE Transactions on Parallel & Distributed Systems, vol. 20, no. , pp. 1641-1652, November 2009, doi:10.1109/TPDS.2008.256