NEF-Assisted Federated Learning for Multi-Access PDU Sessions

The Network Exposure Function is enhanced to provide multi-functional support, serving both traditional communication services and federated learning behaviors through a unified interface. The NEF exposes communication information that can be utilized by both conventional applications and AI/ML federated learning systems, eliminating the need for separate communication management mechanisms.

The Network Exposure Function acts as an intermediary between the core network and AI/ML federated learning applications. It translates and exposes relevant communication information in a format suitable for federated learning while maintaining the existing network architecture, thus enabling federated learning without direct modification of network core components.

The system proactively subscribes to communication information events through the SMF before federated learning operations are needed. By obtaining communication information in advance and caching it locally in the NEF, the system avoids real-time information retrieval delays, thus improving the efficiency of support information management for AI/ML federated learning.

The system establishes a feedback mechanism where the NEF continuously monitors communication information events through subscription to the SMF. This feedback loop ensures that the NEF has up-to-date communication information available when federated learning operations require support information, enabling efficient decision-making without manual intervention.

The NEF proactively retrieves and caches communication information by subscribing to events through the SMF in advance. This preliminary action stores communication information locally, eliminating the need for real-time retrieval during federated learning operations. The system thus achieves both information freshness (through continuous subscription) and reduced retrieval time (through local caching).

The system dynamically adapts its information retrieval strategy based on operational context. During federated learning operations, it uses pre-cached information for immediate access. The subscription mechanism ensures the cache is continuously updated, making the system dynamic in balancing information freshness with retrieval speed without requiring real-time network access for every information request.