Learning from Errors: Detecting ZigBeeInterference in WiFi Networks

—In this work we show how to detect ZigBee inter-ference on commodity WiFi cards by monitoring the receptionerrors, such as synchronization errors, invalid header formats,too long frames, etc., caused by ZigBee transmissions. Indeed, inpresence of non-WiFi modulated signals, the occurrence of thesetypes of errors follows statistics that can be easily recognized.Moreover, the duration of the error bursts depends on thetransmission interval of the interference source, while the errorspacing depends on the receiver implementation.On the basis of these considerations, we propose the adoptionof hidden Markov chains for characterizing the behavior of WiFireceivers in presence of controlled interference sources (trainingphase) and then run-time recognizing the most likely cause oferror patterns. Experimental results prove the effectiveness ofour approach for detecting ZigBee interference.