ML-Based CSI Reporting for Future Downlink Quality Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing communication systems struggle to accurately predict future downlink channel quality in non-terrestrial networks (NTN) due to insufficient channel models and limited information sharing between user equipment (UE) and base stations (gNB), leading to suboptimal modulation and coding scheme adjustments.
Innovation Solution
A method and apparatus that utilize machine learning to fuse ego-sensor information and sidelink data at the UE to predict future downlink channel quality, enabling accurate estimation and reporting of channel quality indicators (CQI) to the gNB, allowing for adaptive transmission parameters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine learning-based prediction is implemented at the UE to improve future channel quality prediction accuracy, then prediction accuracy is improved, but device complexity and energy consumption at the UE increase
Solution Approach 1:
The patent introduces an estimation quality indicator as an intermediary element that mediates between the UE's prediction capabilities and the gNB's transmission decisions. This indicator allows the UE to communicate the reliability of its predictions without requiring the gNB to fully understand or process the underlying machine learning models, thus reducing complexity while maintaining prediction accuracy benefits
Solution Approach 2:
The patent changes the parameter being reported from raw channel quality measurements to estimated future channel quality indicators accompanied by estimation quality metrics. This parameter transformation enables more accurate future-oriented predictions while the estimation quality parameter provides the gNB with sufficient information to make reliable transmission decisions without requiring complex UE-side processing
2Measurement precision
If processing effort is shifted to the UE for channel quality prediction, then prediction accuracy improves, but energy consumption at the UE increases
Solution Approach 1:
The patent applies preliminary action by having the UE perform machine learning-based predictions in advance for future channel quality conditions. By predicting future channel states before actual transmissions occur, the system enables proactive optimization of transmission parameters, improving accuracy while allowing the UE to complete processing work during idle periods rather than during active communication when energy is more critically consumed
3Productivity
If traditional channel quality reporting is used in NTN, then system compatibility is maintained, but spectrum efficiency decreases due to suboptimal transmission parameter adaptation
Solution Approach 1:
The patent implements feedback by introducing estimation quality indicators that provide the gNB with information about the reliability of UE predictions. This feedback mechanism enables the gNB to adaptively adjust transmission parameters based on both the predicted channel quality and the confidence level of those predictions, thereby improving spectrum efficiency while maintaining compatibility with existing systems through the use of standardized reporting formats
Data Source
AI summary
Methods and apparatuses for radio communication. The method includes transmitting first information that characterizes at least one quality associated with a downlink channel received by a radio terminal and transmitting second information that characterizes a present or future environmental situation of a physical entity that is associated with the radio terminal.