Real-time anomaly traffic monitoring based on dynamic k-NN cumulative-distance abnormal detection algorithm

In recent years, the scale of mobile Internet is rapidly increasing because of the explosive growing of smartphone users and applications. The traffic analysis and anomaly detection become critical for mobile operators. Up to now, there are a number of studies for detecting anomaly network traffic. However, the way of detecting anomalies on massive traffic data in real-time manner is not well studied. In this paper, we propose a real-time anomaly detection method based on dynamic k-NN cumulative-distance abnormal detection algorithm. We also present the design and implementation of the method by leveraging Strom, a distributed steam computing technology. Experimental results from evaluation by real-world dataset show that our system is a promised solution for real-time anomaly detection solution in high-speed network.