Cross-Node AI Session Handover in RAN

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing wireless communication systems face challenges in efficiently managing cross-node machine learning sessions during user equipment (UE) mobility, leading to computational resource bottlenecks and increased costs for radio access network (RAN) operators.

Innovation Solution

The implementation of signaling techniques to handle UE mobility for cross-node AI/ML sessions in a cloud-based RAN architecture, allowing for the preservation, deactivation, and reactivation of AI/ML sessions across different network entities during handovers.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If cross-node machine learning sessions are maintained during UE handover between network entities, then machine learning processing continuity is improved, but network signaling complexity and resource management overhead increase

Engineering Contradiction:

Improvemachine learning session continuityVSAvoidnetwork signaling complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The patent applies preliminary action by pre-establishing machine learning session context information at the target network entity before the UE actually hands over. The source network entity forwards ML session information to the target entity in advance, so when handover occurs, the target entity can immediately continue the ML session without interruption or complex signaling during the transition.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent uses the source network entity as an intermediary to transfer ML session context information to the target network entity. This intermediary approach simplifies the handover signaling by having the source entity package and forward the necessary ML session data, reducing the burden on the target entity and core network during the handover process.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Loss of time

If AI/ML computations are processed at RAN-side network entities, then processing latency is reduced, but computational resource requirements and operational costs increase

Engineering Contradiction:

Improveprocessing latencyVSAvoidcomputational resource consumption

Core Design Contradiction:

Loss of timeVSUse of energy by stationary object

Solution Approach 1:

The patent applies local quality by enabling RAN-side network entities to perform AI/ML computations locally for UEs in their coverage area. Each RAN entity maintains and processes ML sessions for its local UEs, reducing the need to transmit data to distant core network entities. This local processing reduces latency while distributing computational load across multiple RAN entities rather than concentrating it all in the core network.

Inventive Principle:

Principle #3Local quality

Data Source

PatentUS20250126526A1Mobility handling of cross-node machine learning session in a radio access network

Publication Date: 2025.04.17 QUALCOMM INC

AI summary

Certain aspects of the present disclosure provide techniques for handling user equipment (UE) mobility for a cross-node artificial intelligence (AI) and/or machine learning (ML) session in a radio access network (RAN). A method of wireless communication by a first network entity includes obtaining, from a second network entity, an indication of a handover for a UE; obtaining, from a third network entity, an indication of first cross-node machine learning information associated with a cross-node machine learning session between the UE and the third network entity; providing, to the second network entity, an indication acknowledging the handover; and relaying communications between the UE and the third network entity for the cross-node machine learning session.