CORE: An accurate congestion detection method in IEEE 802.11 wireless mesh networks

Although IEEE 802.11 wireless mesh networks (WMNs) provide low-cost extension of wireless Internet access coverage, the wide adoption of WMNs is limited due to the bandwidth scarcity. One of the important mechanisms to fully utilize the scarce bandwidth is congestion control. Especially, congestion detection, a module of congestion control, plays an essential role since the source rates should be adapted according to the detected network congestion. Due to the various dynamics of IEEE 802.11 WMNs, wireless congestion control mechanisms usually use complicated combination of various metrics, such as channel-idle time, interface queue length and average link-layer retransmission in intermediate nodes. However, this metric combination requires the sophisticated design of congestion control mechanism and eventually leads to the optimization difficulty. Thus, we propose a simple-yet-effective congestion detection metric, CORE (COngestion Residual timE). CORE measures the expected time of congestion-loss occurrence on the interface queue between IP and MAC layers. We obtain CORE by measuring the residual queue length and the slope of the queue length and then refining them with signal processing schemes (e.g., triangular moving average and linear regression). Moreover CORE provides over-generated rate information that is essential to the congestion control mechanisms. By taking a series of experiments, we verify that CORE accurately predicts the time of congestion-loss occurrence as well as the over-generated rate.