Channel Classification Training With On-Demand Labeling Triggers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing channel classification models in wireless communication networks face inefficiencies in resource utilization and accuracy due to the lack of effective methods for determining when and where additional training data is needed, leading to unnecessary resource expenditure and suboptimal performance in non-line-of-sight (NLOS) propagation scenarios.
Innovation Solution
A system for on-demand labelling of channel classification training, where devices assess the importance of measurement data and trigger labeling only when necessary, using importance assessment information to determine if a third device should perform classification labeling, optimizing the collection of training data for model improvement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If classification labeling is performed for all channel measurement data, then model training data completeness is improved, but labeling overhead and resource expenditure increase
Solution Approach 1:
The system performs classification labeling only on a subset of channel measurement data that is deemed important for model training, rather than labeling all available data. The importance assessment mechanism identifies and selects only the most valuable samples for labeling, achieving sufficient model training data completeness while significantly reducing labeling overhead and resource expenditure.
2Measurement precision
If more training data is collected for channel classification, then model accuracy is improved, but resource utilization efficiency deteriorates
Solution Approach 1:
The system applies different treatment to different portions of training data based on their local importance characteristics. By assessing the importance of individual channel measurement samples and selectively labeling only those with high importance values, the system achieves high model accuracy while avoiding the resource waste associated with uniformly processing all training data.
3Measurement precision
If classification labeling is performed frequently to improve model performance, then channel classification accuracy is improved, but time consumption and overhead increase
Solution Approach 1:
The system implements a feedback mechanism where the importance assessment information guides whether classification labeling should be performed. By continuously monitoring the importance of channel measurement data and only triggering labeling when importance thresholds are exceeded, the system achieves high channel classification accuracy while minimizing time consumption and labeling overhead.
Data Source
AI summary
Example embodiments of the present disclosure relate to on-demand labelling for channel classification training. A first device determines, using a classification model, a classification result of a communication channel based at least in part on channel measurement information about the communication channel, and determines, based at least in part on a type of the classification model, importance assessment information to indicate an importance level of the channel measurement information in updating the classification model. The first device transmits the importance assessment information to a second device. In accordance with a determination that the importance level of the channel measurement information exceeds the importance threshold, the second device causes a third device to perform classification labeling for at least the communication channel at a location associated with the first device.