Anomaly Detection of 5G Control Plane Based on Hidden Semi-Markov Model

This research delves into the security concerns surrounding 5th generation mobile networks(5G) control plane (CP) protocols. With intrusion detection playing a pivotal role in safeguarding the security of 5G CP, it is essential to overcome the limitations and incredibility issues found in existing machine learning and deep learning methods. These methods not only overlook important network functions responsible for data transmission and their corresponding states but also risk hindering the effective detection of CP regulation violations. To tackle this issue, we propose an anomaly detection algorithm based on the Hidden Semi-Markov Model (HSMM). We use HSMM to represent the 5G CP protocol by mapping unknown states and known signals to hidden and observable states, taking into account the duration of hidden states. Leveraging this HSMM-based 5G CP model, our algorithm design involves an anomaly detection methodology that calculates probabilities for observed signals, detecting abnormal behaviors within the CP by comparing these probabilities against predefined thresholds. We validate the effectiveness of our proposed algorithm through proposed scenarios including normal scenario, DoS scenario, and intercept scenario of the 5G CP. The experimental results show the capabilities of the HSMM algorithm in accurately detecting abnormal behaviors, making it well-suited for diverse types of attack scenarios when compared to two existing anomaly detection algorithms.