UE Capability Hash Reporting for AI/ML Signaling Overhead

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

The increasing complexity and volume of UE capability information, particularly due to AI/ML features, lead to significant signaling overhead and delays in wireless communication networks, and the handling of AI/ML models requires efficient updates to avoid interruptions.

Innovation Solution

Implementing a UE capability reporting mechanism using identification values, such as hash values, to represent similar capabilities collectively, and caching AI/ML models for local updates, reducing signaling overhead and enabling faster network integration.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If detailed UE capability information is reported for each UE, then network can accurately understand UE capabilities, but signaling overhead increases significantly

Engineering Contradiction:

Improvecapability information accuracyVSAvoidsignaling overhead

Core Design Contradiction:

Measurement precisionVSQuantity of substance

Solution Approach 1:

The patent groups UEs with similar capabilities into capability sets and reports them collectively using a single capability information message, rather than reporting individual capability details for each UE. This merging approach maintains accurate capability representation while significantly reducing the quantity of signaling data transmitted.

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The capability information message is designed to serve multiple UEs simultaneously by using a universal structure that can represent capabilities for a group of UEs. The message includes identification information that links the capability description to multiple specific UEs, allowing one message to fulfill the capability reporting function for several UEs.

Inventive Principle:

Principle #6Universality (Multi-functionality)

2Reliability

If complete UE capability information is transmitted, then network can make accurate decisions, but transmission time and delays increase

Engineering Contradiction:

Improvenetwork decision accuracyVSAvoidcapability reporting delay

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The patent extracts and transmits only the essential capability information that is necessary for network decisions, rather than transmitting complete detailed capability data for each UE. By identifying and reporting only the relevant capability characteristics, the system maintains decision-making accuracy while reducing transmission time and delays.

Inventive Principle:

Principle #2Taking out (Extraction)

3Reliability

If AI/ML models are updated through network signaling, then models can be kept current, but signaling overhead and interruptions increase

Engineering Contradiction:

Improvemodel currencyVSAvoidsignaling overhead

Core Design Contradiction:

ReliabilityVSQuantity of substance

Solution Approach 1:

The patent enables UEs to autonomously update their AI/ML models using locally stored capability information and self-service mechanisms. Instead of requiring continuous network signaling for model updates, UEs can independently retrieve and apply model updates based on their capability set, significantly reducing signaling overhead while maintaining model currency through local resource utilization.

Inventive Principle:

Principle #25Self-service

Data Source

PatentUS20260075408A1UE capability reporting and handling of artificial intelligence/machine learning models

Publication Date: 2026.03.12 FRAUNHOFER GESELLSCHAFT ZUR FORDERUNG DER ANGEWANDTEN FORSCHUNG EV

AI summary

A user device, UE, for a wireless communication network, like a 3rd Generation Partnership Project, 3GPP, network, is described. The UE has a plurality of UE capabilities, the UE capabilities indicating functions the UE is capable to perform. The UE is to inform on one or more parts of the plurality of UE capabilities using one or more identification values, e.g., hash values.