Dynamic Frequency Selection through Collaborative Reporting in WLANs

Authors: Giannis Kazdaridis, Stratos Keranidis, Adamantios Fiamegkos, Thanasis Korakis, Iordanis Koutsopoulos  and Leandros Tassiulas

Conference: MobiCom 2011, Las Vegas, Nevada, USA, September 2011

The tremendous growth of 802.11 WLANs has resulted in congestion of the limited unlicensed spectrum, especially in densely populated urban areas. The performance experienced by end-users in such deployments is significantly degraded due to contention and interference among adjacent cells. In order to address this challenge, various studies have proposed mechanisms for allocating the available number of channels to cells. In infrastructure 802.11 WLANs, channel selection is performed solely by the AP. Most implemented approaches result in static channel assignments, which is not consistent with the dynamic nature of the wireless medium. In this demo, we present a frequency selection algorithm implemented in the Mad-WiFi open source driver that dynamically switches the operational channel by taking into account both channel as well as traffic conditions. The proposed algorithm features a collaborative reporting mechanism, which enables nodes of the cell, as well as nodes belonging to different cells, to contribute to interference measurements. Based on this mechanism the APs decide on frequency selection more accurately and moreover manage to detect "hidden" APs. Compared to current approaches our algorithm manages to improve total network throughput, up to a factor of 7.5. For the purposes of this demo we use a graphical user interface to demonstrate an online frequency selection experiment that is remotely executed in NITOS wireless testbed.

Download paper: DFS_DEMO_Mobicom_2011.pdf