Real-Time Scene-Sensitive Anomaly Detection of 5G Control Plane Based on Markov Chain
Lulu Dai, Qian Lai Sun, Lin Tian · 2024
Despite security enhancements in the Control Plane (CP) of the fifth-generation (5G) communication network, such as the introduction of Public Key Infrastructure (PKI), security vulnerabilities still exist. Exploiting these vulnerabilities can lead to severe impacts, resulting in 5G CP process anomalies. Although anomaly detection techniques can help detect such attacks, existing research often needs to pay more attention to scene features and provide prevention capabilities, rendering it insufficient for ensuring the security of 5G CP. This paper addresses these challenges by proposing a real-time scene-sensitive anomaly detection scheme for the 5G CP. The paper introduces the design of a CP process model using the Markov Chain, in which the transition probability changes during an attack compared to regular network operation. The change in likelihood serves as an indicator of the attack's influence. We propose a real-time method for calculating the attack influence, considering adversary capabilities such as cost, motivation, and feasibility of attacking the CP process. Based on the attack influence, a scene-sensitive anomaly detection scheme is designed, which calculates the attack probability and constructs a threshold function by considering the varying transition probabilities across different scenarios. When the attack influence exceeds the threshold in a specific scenario, the network is alerted to abnormal signaling and thwarts the attack. Simulation results effectively demonstrate the efficacy of the proposed scheme.