MDU-CACS: A Coordinated Forecasting-Based Cloud-Assisted Dynamic Channel Assignment Mechanism for Wi-Fi Network Clusters

In this paper, we propose a cloud-assisted dynamic channel assignment system, MDU-CACS, for groups of Wi-Fi networks in a coordinated fashion to improve the performance and user experience in each network. MDU-CACS expands upon our previous mechanism, CACS, with the same design principle and framework. MDU-CACS utilizes periodic measurements of channel availability and neighbor information by the access points (AP) in all possible 2.4 GHz and 5 GHz channels. A cloud system gathers this measurement information and generates channel state predictions for each channel for each AP. Consequently, by using these predictions the cloud system searches for better a channel assignment for the whole Wi-Fi network group and sends channel change information to each AP if there is a better channel than their operating channels. We have conducted field trials of the two proposed channel assignment mechanisms over large scale actual residential Wi-Fi networks and compare their performance against widely deployed Least Congested Channel Search (LCCS) mechanism. Our results show CACS outperforms LCCS in terms of operating channel interference level with reduced number of channel changes; MDU-CACS further improves the operating channel interference over CACS albeit at a cost of somewhat higher number of channel changes. Both of our systems offer a complete framework for managing the channel assignment of Wi-Fi networks considering dynamically varying wireless channel states and organically changing Wi-Fi network topologies following realistic residential Wi-Fi network environments.