Communication Parameter Determination via ML Data Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for determining communication parameters in 5G networks, such as beamforming weights, are inefficient due to reliance on unreliable PMI and CQI measurements, which are affected by neighboring cell interference, environmental changes, and UE mobility, leading to inaccurate beam shaping and energy wastage.
Innovation Solution
A method using a machine learning algorithm, specifically unsupervised anomaly detection, to filter out error measurement data from PMI and CQI reports, allowing for more accurate determination of communication parameters like common beamforming weights by extracting stable UE trajectories and channel information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional manual cell shaping methods are used with pre-defined cell shape sets, then operators can perform driving tests to select the best configuration, but this process consumes significant time and effort for network deployment and optimization
Solution Approach 1:
The patent replaces the manual mechanical trial-and-adjustment process with an automated mathematical system based on SVD decomposition. The system automatically calculates optimal common beamforming weights by processing UE measurement data (PMI, CQI) through singular value decomposition, eliminating the need for operators to manually test pre-defined cell shape sets and perform repeated driving tests.
2Productivity
If SVD-based automatic calculation is used to generate common beamforming weights, then deployment time is reduced, but measurement precision deteriorates because PMI and CQI are sensitive to short-term channel disturbances such as neighboring cell interference and environmental changes
Solution Approach 1:
The patent applies preliminary filtering actions to the measurement data before SVD decomposition. The system pre-processes PMI and CQI reports by identifying and removing error measurements caused by short-term channel disturbances, ensuring that only reliable measurement data is used in the subsequent automatic calculation of common beamforming weights.
Solution Approach 2:
The patent introduces an intermediary filtering mechanism between the raw measurement data and the SVD calculation process. This intermediary layer processes the measurement data to eliminate errors from neighboring cell interference and environmental changes, providing clean input data to the SVD algorithm for accurate beamforming weight generation.
3Measurement precision
If all measurement data is collected without filtering for SVD calculation, then calculation accuracy may be improved, but error measurements from temporary strong reflectors and UE mobility cause beam energy to be wasted in unnecessary directions
Solution Approach 1:
The patent extracts and removes error measurements from the measurement data before SVD calculation. The system identifies measurements affected by temporary strong reflectors and UE mobility anomalies, and extracts only the reliable measurements for calculating common beamforming weights, preventing beam energy from being directed toward unnecessary directions.
Data Source
AI summary
Embodiments of the present disclosure provide method and apparatus for determining communication parameter. A method performed by a network node. The method includes obtaining measurement data for at least one terminal device. The method further includes filtering the measurement data to remove error measurement data by a machine learning algorithm. The method further includes determining at least one communication parameter based on the filtered measurement data.