Cloud-assisted channel selection for wireless mesh networks

This paper proposes a cloud-assisted channel selection method for multi-node wireless networks, in particular wireless mesh networks. Periodic channel idle airtime measurements of the access points (APs) that constitute the mesh network are accumulated at a remote controller that runs various forecasting methods on the aggregated data, and finds the best fitting method for each AP-channel pair. By using the selected forecasting method, for every channel and every AP, expected idle airtime in the near future is calculated. Channel assignment is carried out by the remote controller, based on the forecasted idle airtime of each AP. While performing the channel assignment, the remote controller takes into account the available frequency bands in the network, in particular, the band used by the wireless backbone of the mesh network, e.g. 5 GHz band, and the other frequency band, e.g., 2.4 GHz band. The proposed method is implemented and tested in a real test bed that comprises dual-band concurrent mesh APs, and a remote controller that runs in a cloud server. The results show that the proposed method accurately forecasts the idle airtime, and yields successful channel assignment for each frequency band the mesh network is operating in.