Beam Reporting Split Feedback for AI/ML Beam Management

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing AI/ML-based beam management systems face poor performance due to high feedback overhead and inefficient reporting of beam information by user equipment (UE), leading to suboptimal model training and inference.

Innovation Solution

The UE splits reporting information into a first part and a second part, where the first part occupies a fixed number of bits, and the second part's bit number is determined by the first part's content, ensuring efficient feedback and accurate model training and inference.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If the UE reports L1-RSRPs of all beams or all optimal beams to ensure AI/ML model performance, then the model performance is improved, but the feedback overhead becomes excessively large

Engineering Contradiction:
ImproveAI/ML model performanceVSAvoidfeedback overhead
Core Design Contradiction:
ReliabilityVSQuantity of substance

Solution Approach 1:

The patent segments the beam reporting process into two stages: first reporting a subset of beams (e.g., top-K beams or beams above a threshold) with full L1-RSRP information, then reporting additional beams only when necessary. This segmentation allows the UE to provide sufficient data for AI/ML model training while avoiding the overhead of reporting all possible beams.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies partial action by having the UE report only a portion of the available beam information (e.g., top-K beams or beams exceeding a threshold) rather than all beams. This partial reporting approach provides enough data for effective model training while significantly reducing feedback overhead compared to reporting all beams.

Inventive Principle:
Principle #16Partial or excessive action

2Quantity of substance

If the UE reports only a part of the beams or reduces the number of optimal beams to reduce feedback overhead, then the feedback overhead is reduced, but the AI/ML model performance deteriorates

Engineering Contradiction:
Improvefeedback overheadVSAvoidAI/ML model performance
Core Design Contradiction:
Quantity of substanceVSReliability

Solution Approach 1:

The patent introduces dynamic reporting mechanisms where the UE can adjust the number and selection of beams reported based on channel conditions and model requirements. The reporting configuration can be dynamically adapted to balance between overhead reduction and performance maintenance, allowing the system to respond to varying operational conditions.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The patent changes parameters such as the threshold for beam selection, the value of K for top-K reporting, and the reporting configuration based on channel conditions and model requirements. By dynamically adjusting these parameters, the system optimizes the balance between feedback overhead and AI/ML model performance.

Inventive Principle:
Principle #35Parameter changes

3Device complexity

If a fixed number of beams are reported to simplify the reporting process, then the implementation complexity is reduced, but the ability to adapt to different channel conditions and model requirements is limited

Engineering Contradiction:
Improvereporting process complexityVSAvoidadaptability to channel conditions
Core Design Contradiction:
Device complexityVSAdaptability or versatility

Solution Approach 1:

The patent implements preliminary configuration where the network device pre-configures parameters such as the threshold for beam selection and the value of K for top-K reporting. This preliminary setup simplifies the UE's processing by providing clear reporting criteria while maintaining adaptability to different channel conditions through configurable parameters.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent creates a universal reporting mechanism that can handle multiple scenarios (different numbers of beams, various threshold conditions, top-K reporting) through a single flexible framework. The same reporting structure adapts to different channel conditions and model requirements by adjusting parameters rather than requiring separate processing paths.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentEP4730685A1Information transmission method and apparatus, and storage medium
Publication Date: 2026.04.22 DATANG MOBILE COMM EQUIP CO LTD
  • EP4730685A1 patent drawingFigure 1~2
  • EP4730685A1 patent drawingFigure 3~4
  • EP4730685A1 patent drawingFigure 5~6

AI summary

The present application provides an information transmission method and apparatus, and a storage medium. The method comprises: determining reporting information meeting a reporting condition, the reporting information being used for training an artificial intelligence model or being used for determining a beam/beam pair having an optimal transmission signal; and splitting the reporting information into a first part of information and a second part of information and sending same to a network device.