LAUSR

In this paper, we present deep network analyzer (DNA), a big data analytics platform for anomaly detection (AD) and root cause analysis (RCA) in mobile wireless networks. DNA is motivated by the growing scale and complexity of cellular networks along with the lack of advanced big data analytics tools for effective network management. It abstracts the RCA process into two modules, namely rule (fingerprint) learning and the module of AD and fingerprint matching. We first develop a rare association rule mining method to learn the symptoms of network anomalies and to build a fingerprint knowledge database from the historic data. Then a statistical machine learning approach is employed to identify the anomalies within the incoming dataset collected via various probes in the network and map the fingerprints of the detected anomalies to the rules in the knowledge database. The DNA platform has been tested using the real production data from the field and has been shown to be a highly effective platform for AD and RCA for large-scale cellular systems serving tens of millions of mobile users.