Campus Wireless Access Point Anomaly Detection using Unsupervised Method at Multiple Time Granularities

As wireless networks play an increasingly important role in campus communications, identifying anomalous APs(wireless access points) is becoming increasingly important. This paper proposes an unsupervised AP anomaly detection method at multiple time granularities, which does not depend on the labeling quality of training samples. In addition, the proposed method considers the APs’ behavior under different time granularities and can eliminate the influence of improper selection of the time granularity on anomaly detection. Experimental results show that the proposed method can detect anomalous APs and help network administrators locate wireless network problems.