A Data Plane Approach for Detecting Control Plane Anomalies in Mobile Networks

This paper proposes an anomaly detection framework that utilizes key performance indicators (KPIs) and traffic measurements to identify in real-time misbehaving mobile devices that contribute to signaling overloads in cellular networks. The detection algorithm selects the devices to monitor and adjusts its own parameters based on KPIs, then computes various features from Internet traffic that capture both sudden and long term changes in behavior, and finally combines the information gathered from the individual features using a random neural network in order to detect anomalous users. The approach is validated using data generated by a detailed mobile network simulator.