Towards Efficient Mobility Management in 5G-Advanced: A Predictive Model for Network Slice Availability

초록

Network slicing in 5G-Advanced networks allows the creation of virtualized slices to support diverse services like IoT, autonomous driving, and entertainment. However, mobility management poses significant challenges, particularly as User Equipment (UE) moves across Tracking Areas (TAs) with non-uniform slice availability. Current systems restrict UEs from re-registering for slices within the same Registration Area (RA) if a slice is unavailable in just one TA, leading to service disruptions and inefficient resource utilization. To address this, we propose an LSTM-based prediction model that anticipates slice availability in different TAs. The model analyzes historical slice availability, UE mobility patterns, and current network conditions to predict future slice availability, allowing UEs to optimize their mobility and reduce the need for frequent Mobility Registration Updates (MRUs). Our simulation results show that the proposed model achieves high prediction accuracy, significantly reducing signaling overhead and improving both resource efficiency and service continuity.

키워드