Model Adaptation for Lifecycle Management in AI-native 5G/6G Networks

초록

As part of ongoing standardization efforts in 5G-Advanced for artificial intelligence (AI) and machine learning (ML)-based beam management (BM), model adaptation framework is proposed in this paper as a key lifecycle management (LCM) mechanism to ensure reliable model performance in dynamic wireless environments. The proposed approach enables the network to proactively respond to performance degradation by selecting and activating a more suitable model based on support information and user-side reporting. The superiority of the proposed method over non-adaptive or dataset-only approaches in maintaining beam prediction accuracy is demonstrated by simulation results, highlighting its effectiveness as a robust and scalable solution for future AI-native 6G systems. © 2025 IEEE.

키워드