An experimental study of the channel switching cost in multi-radio wireless mesh networks

Abstract

A large body of research has recently addressed the channel assignment problem in multi-radio wireless mesh networks. In order to reduce interference, many proposals require radio interfaces to (more or less frequently) switch channels to exploit the availability of multiple orthogonal channels. However, such proposals have been almost exclusively evaluated by means of simulations and the impact of switching channels on the network performance has not been experimentally evaluated so far. In this article, we aim to fill such a gap and present the results of a thorough experimental campaign we conducted on the ORBIT wireless testbed to study the effects of switching channels in a multiradio wireless mesh network. Our experiments show that in common scenarios a channel switch causes a non-negligible interruption in the connectivity among nodes, which is on the order of 10 seconds. Also, our tests reveal that a rapid recovery from link failures caused by channel switches is prevented by a rather slow update of the advertised link quality. Inspired by the analysis of the experiments we conducted, we also propose and evaluate a preliminary technique to reduce the interruption in the delivery of packets caused by a channel switch.