AI-based intrusion detection and adaptive access control for enhancing security in 6G network slicing

As 6G networks become a reality, they will certainly usher in new security challenges that will need to be addressed, particularly due to network slicing in multi-tenant environments. In this work, we proposed an AI-based framework consisting of: (1) a deep learning-based intrusion detection system (AI-IDS), and (2) a reinforcement learning-based adaptive access control system (AACS) for slice-level security. The proposed system identifies threats (known and unknown), dynamically develops, and enforces access policies based on normal user behavior. The framework was evaluated using various benchmark datasets in a simulated 6G slicing environment. Overall, the proposed AI framework achieved ~93% detection accuracy, low false positive rates (~4%), and reasonably rapid response times (~75 ms). Results showed the proposed framework provided higher adaptability and accuracy over traditional fixed security mechanisms.