AI-Driven Secure Network Slicing for Cloud-Native 6G Networks Using Zero-Trust Architecture and Intelligent Intrusion Detection

Abstract

6G will reshape digital systems by blending AI, cloud-native orchestration, edge computing, ultra-low latency, and smarter network automation. A key enabler is network slicing, which runs multiple virtual networks on the same hardware to meet diverse needs like enhanced mobile broadband, ultra-reliable low-latency links, and massive IoT connections. But slicing’s distributed, cloud-first design opens new attack paths—cross-slice breaches, compromised orchestration, DDoS, API abuse, lateral movement, and attacks on AI components. This work proposes an AI-based, zero-trust defense that ties together Kubernetes, SDN, NFV, Istio service-mesh segmentation, and ML-powered intrusion detection. It continuously checks identities, mTLS-protected channels, runtime telemetry, and adaptive trust scores. Multiple ML and deep models (Random Forest, XGBoost, LSTM, Autoencoders) were tested on telecom datasets including 5G-SliciNdd, 5G-NIDD, 5GCID, UNSW-NB15, and CICDDoS2019; XGBoost led in spotting DDoS, orchestration faults, and unauthorized slice traffic. The design boosts visibility, speeds response, and scales for AI-native 6G deployments.