An Efficient Lightweight Anomaly Detection for 5G Core Network

Viettel’s 5GC network, which controls and manages mobile network services, is crucial for the operation of the its own telecommunications system. Any fault in the core can cause service interruptions, system crashes, etc., so it is necessary to indicate the cause of the fault whenever there is abnormal condition warning. Since Viettel has its own specific data, it requires us to design a solution specifically for the Viettel’s 5GC network. Our proposed solution detects anomaly by OCSVM (One-Class Support Vector Machine), and then the fault instance is analyzed by X-AI to localize the cause of the fault. The solution is shown to be able to achieve 98% of (F1-score) and at the same time indicate the corresponding fault components.