AI-Assisted Base Station Positioning With LMF Model Alignment

The patent extracts the AI model training and updating operations from the base station side and concentrates them on the LMF device side. The base station only needs to obtain assistance information from the LMF device and perform straightforward positioning calculations, while the LMF device handles the complex AI model training, updates, and management. This extraction of complex functions to a dedicated management entity resolves the technical contradiction by maintaining positioning accuracy through centralized AI processing while significantly reducing base station complexity.

The patent applies preliminary action by having the LMF device pre-train AI models and prepare assistance information before actual positioning operations. The assistance information containing pre-processed AI model parameters, deployment scenario configurations, and equipment characteristics is prepared in advance and stored on the LMF device. During positioning operations, the base station simply retrieves this pre-prepared information rather than exchanging large volumes of raw data, thus improving positioning accuracy while minimizing channel occupancy.

The patent implements local quality by creating specialized AI models tailored to specific deployment scenarios (e.g., urban canyon, rural area, indoor) and UE characteristics (e.g., device type, antenna configuration). The LMF device maintains multiple scenario-specific AI models and selects the appropriate model based on the current deployment environment and UE properties. This approach improves positioning accuracy by using locally optimized models while the LMF device's centralized management prevents information loss by systematically organizing and tracking which models apply to which scenarios.