Model-based fault Diagnosis for IEEE 802.11 wireless LANs

The increasingly deployed IEEE 802.11 wireless LANs (WLANs) challenge traditional network management systems because of the shared open medium and the varying channel conditions. There needs to be an automated tool that can help diagnosing both malicious security faults and benign performance faults. It is often difficult, however, to identify the root causes since the manifesting anomalies from network measurements are highly interrelated. In this paper we present a novel approach, called MOdel-based self-Diagnosis (MODI), for fault detection and localization. Our solution consists of Structural and Behavioral Model (SBM) that is constructed using both structural causality from wireless protocol specifications and behavioral statistics from network measurements. We use logic-based backward reasoning to automate fault detection and localization based on SBM, by comparing observed network measurements with expected network behaviors and by tracing back causality structures. The reasoning algorithm and the model description are decoupled so a SBM model can be easily updated for varying WLAN configurations and changing network conditions. Compared to previous work, the contribution of this paper is the architecture and the algorithm of the diagnosis core, rather than the WLAN measurement techniques. We built and deployed MODI-embedded wireless APs that can detect both security attacks and troubleshoot performance problems. These MODI-enabled APs can also cooperate to diagnose cross-AP problems, such as those caused by device mobility. The evaluation results demonstrate that the proposed model-based diagnosis is fast and effective with little overhead.