Dynamic Reference Signal Selection for Beam Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional wireless communication networks face challenges in accurately predicting optimal receive beams due to the use of fixed beam patterns, which can result in blocked beams and reduced measurement accuracy, especially during cell handovers and beam management.
Innovation Solution
A communication method and apparatus that integrates AI to dynamically select a group of reference signal resources for model inference, allowing flexible beam selection and improving accuracy by avoiding blocked beams, using AI models trained on multiple groups of beams.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If fixed beam patterns are used for model inference, then the system structure is simple and easy to implement, but the measurement accuracy deteriorates due to blocked beams
Solution Approach 1:
The patent applies dynamics by transitioning from fixed beam patterns to dynamic beam selection. The system now adapts beam patterns based on real-time network conditions, terminal positions, and blockage states. Multiple beam patterns are maintained in the codebook, and the selected pattern changes dynamically to optimize measurement accuracy while avoiding blocked beams.
Solution Approach 2:
The patent changes the parameter of beam pattern selection from static to variable. Different beam patterns with different spatial characteristics are prepared in advance, and the system selects appropriate patterns based on current conditions. This parameter change allows the system to adapt to varying measurement scenarios and avoid blockages.
2Measurement precision
If multiple groups of reference signal resources are prepared for dynamic selection, then the accuracy of predicting optimal receive beams is improved, but the network complexity and resource overhead increase
Solution Approach 1:
The patent segments the beam resource space into multiple groups of reference signal resources, where each group contains beams with different spatial characteristics. This segmentation allows the system to select from diverse beam patterns rather than using a single fixed pattern, improving prediction accuracy while managing resource overhead through organized grouping.
Solution Approach 2:
The patent creates multi-functional reference signal resources that can serve different purposes depending on the selected group. The same infrastructure supports multiple beam patterns and selection strategies, making the system versatile in handling different measurement scenarios without proportionally increasing overhead.
3Measurement precision
If fixed beams at several locations are used as inputs for model inference, then the implementation is straightforward, but the predicted receive beam precision is low due to beam blockage
Solution Approach 1:
The patent makes the beam selection dynamic by allowing the system to choose from multiple groups of reference signal resources based on current conditions. This dynamic selection avoids fixed beam blockage issues while maintaining reasonable implementation complexity through predefined groups and selection criteria.
Solution Approach 2:
The patent performs preliminary action by pre-preparing multiple groups of reference signal resources with different spatial characteristics before actual beam prediction. This advance preparation allows the system to quickly select appropriate beams without complex real-time calculations, balancing accuracy with implementation simplicity.
Data Source
Figure 1
Figure 2~3
Figure 4~5
AI summary
A communication method and apparatus are provided. The method includes: A first communication apparatus determines a first group of reference signal resources from multiple groups of reference signal resources. The first communication apparatus determines a first reference signal resource based on the first group of reference signal resources and a model. According to the method in this application, a group of reference signal resources is selected from the multiple groups of reference signal resources, and model inference is performed by using the selected group of reference signal resources. In comparison with a manner of performing model inference by using a fixed pattern or a random pattern, this method can improve accuracy of an inference result.