White Paper: Joint Embedding Predictive Architecture (JEPA) for 5G Network-Based Location

How the Location Management Function can move beyond measurement aggregation to contextual prediction.

Network-based positioning in 5G has reached a paradoxical moment. Operators have deployed standards-compliant Location Management Functions (LMFs), yet performance in dense urban environments, indoor transitions, and handover boundaries remains fragile.

The issue is not measurement volume. It is architectural posture.

This technical position paper introduces a different approach — one that explores how representation learning, and specifically Joint Embedding Predictive Architecture (JEPA), can evolve the LMF from a stateless measurement engine into a context-aware inference system.

This paper establishes a foundation for how positioning infrastructure may evolve as networks begin to learn from their own operational history.

What you’ll learn

• Why today’s LMF architecture struggles in dynamic RF environments
• How representation learning enables context-aware positioning
• The role of JEPA in predicting location through uncertainty
• A practical path for augmenting existing 3GPP-based positioning systems

Click to access the white paper.