Channel State Information Prediction for Low-Overhead MIMO Precoding

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current wireless communication systems face challenges in obtaining accurate channel state information (CSI) for efficient downlink transmissions, particularly in 5G networks, due to the overhead and precision issues in CSI feedback methods, which affect the performance of multiple-input multiple-output (MIMO) precoding.

Innovation Solution

Implementing a channel predictor function at both the network node and terminal device using machine learning techniques, such as Kalman filters or artificial intelligence-based methods, to predict future CSI values, reducing the need for extensive feedback by using past CSI estimations and minimizing over-the-air overhead.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional CSI feedback methods are used to obtain accurate channel state information, then downlink transmission performance is improved, but feedback overhead increases

Engineering Contradiction:
ImproveCSI accuracyVSAvoidfeedback overhead
Core Design Contradiction:
Measurement precisionVSLoss of information

Solution Approach 1:

The network node performs preliminary CSI prediction using machine learning models before actual downlink transmission. By predicting future CSI values based on historical channel state data, the system prepares accurate CSI information in advance, reducing the need for extensive real-time feedback while maintaining transmission performance

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

A machine learning-based prediction model acts as an intermediary between historical CSI data and current channel state requirements. The model processes past channel measurements and generates predicted CSI values, serving as a bridge that reduces direct feedback requirements while preserving accuracy

Inventive Principle:
Principle #24Intermediary (Mediator)

2Measurement precision

If extensive CSI feedback is collected to ensure accuracy, then MIMO precoding performance is improved, but system complexity increases

Engineering Contradiction:
ImproveMIMO precoding accuracyVSAvoidfeedback processing complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent replaces traditional mechanical feedback collection and processing systems with machine learning-based prediction models. Instead of collecting and processing extensive raw CSI feedback through complex algorithms, the system uses trained ML models to predict CSI values, significantly reducing processing complexity while maintaining precoding accuracy

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

3Reliability

If real-time CSI feedback is transmitted to maintain connection quality, then quality of service is improved, but transmission delay increases

Engineering Contradiction:
Improveconnection qualityVSAvoidfeedback transmission delay
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The system performs CSI prediction in advance using machine learning models, generating predicted channel state information before it is actually needed for downlink transmission. This preliminary action eliminates the need for real-time feedback transmission, reducing delay while maintaining connection quality through accurate predicted CSI values

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP3989459B1Channel state information reporting
Publication Date: 2025.11.05 NOKIA TECHNOLOGIES OY
  • EP3989459B1 patent drawingFigure 1~2
  • EP3989459B1 patent drawingFigure 3~5
  • EP3989459B1 patent drawingFigure 6

AI summary

Apparatuses and methods in a communication system are disclosed. Reference data is transmitted (600) to terminal device. Information describing state of the channel between the apparatus and the terminal device is received (604), for a given time period (602), from the terminal device. A predicted state of the channel is calculated (606) utilising a given channel predictor function known by the terminal device. Update information to the predicted state of the channel is received (608) from the terminal device.