AI Model Selection for Channel Estimation in PRB Bundling

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

In traditional channel estimation methods, the introduction of Physical Resource Block bundling (PRB bundling) configuration leads to mismatched input data dimensions for Demodulation Reference Signals (DMRS) of different numbers of Resource Blocks (RBs), causing AI models to fail in adapting to changes in channel estimation.

Innovation Solution

An information processing method where a base station sends configuration information including RB indication information to a User Equipment (UE), allowing the UE to determine an appropriate AI model for channel estimation based on the number of RBs with the same precoding.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If PRB bundling configuration is introduced to improve channel estimation adaptability, then the ability to handle different RB scenarios is improved, but the input data dimension mismatch causes AI model failure

Engineering Contradiction:

Improveadaptability to different RB scenariosVSAvoidAI model execution reliability

Core Design Contradiction:

Adaptability or versatilityVSReliability

Solution Approach 1:

The patent introduces dynamic selection of AI models based on the actual number of RBs. The UE determines which AI model to use by comparing the actual RB count against stored dimension information, allowing the system to adapt dynamically to different PRB bundling configurations while maintaining reliable channel estimation

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The patent changes the parameter being monitored from fixed model input dimensions to variable RB counts. By storing dimension information associated with each AI model and selecting based on actual RB count, the system accommodates parameter changes in PRB bundling configurations without causing model failure

Inventive Principle:

Principle #35Parameter changes

2Device complexity

If a single AI model is used to simplify the system, then device complexity is reduced, but the model cannot adapt to DMRS with different numbers of RBs

Engineering Contradiction:

ImproveAI model management complexityVSAvoidadaptability to different DMRS configurations

Core Design Contradiction:

Device complexityVSAdaptability or versatility

Solution Approach 1:

The patent segments the AI model set into multiple models, each associated with specific input dimension information. The UE stores multiple AI models with their corresponding dimension characteristics and selects the appropriate segment based on the actual RB count, achieving adaptability without requiring a single complex universal model

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent creates a universal selection mechanism that works across different RB configurations. The dimension information storage and comparison approach provides a universal framework that can handle any RB count scenario using the same selection logic, making the system multi-functional across different PRB bundling configurations

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS20250048137A1Information processing method and apparatus, and communication device and storage medium

Publication Date: 2025.02.06 BEIJING XIAOMI MOBILE SOFTWARE CO LTD

AI summary

An information processing method is performed by a base station and includes: sending configuration information, wherein the configuration information includes Resource Block (RB) indication information indicating a number of RBs with same precoding for a User Equipment (UE), and the number of RBs is used for the UE to determine an Artificial Intelligence (AI) model to perform channel estimation.