User Equipment Model Updating for Resource-Efficient State Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing user equipment systems face challenges in achieving accurate state estimates while minimizing resource requirements, as prediction models may become invalid and state measurements may be unreliable due to limited availability and accuracy.
Innovation Solution
Implement a model improvement module that alternates between a training mode and an idle mode to enhance the prediction model's accuracy and reliability, adjusting resource usage accordingly to ensure consistent and efficient state estimation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the prediction model is continuously trained with high resource consumption, then the accuracy of state estimates is improved, but the resource requirements increase
Solution Approach 1:
The system switches between training mode and idle mode in periodic cycles. During training mode, the prediction model is updated using recent state measurements with high resource consumption. During idle mode, the trained model is used for state estimation with reduced resource consumption. This periodic alternation resolves the contradiction by concentrating resource usage in time windows when model improvement is needed, rather than continuously consuming resources.
Solution Approach 2:
The system performs preliminary training of the prediction model during training mode before using it for state estimation in idle mode. By preparing the model in advance with the most accurate and recent data, the system ensures high estimation accuracy without needing to continuously train, thus reducing overall resource requirements.
2Measurement precision
If state measurements are continuously collected with high reliability, then the accuracy of state estimates is improved, but the resource consumption increases
Solution Approach 1:
The system periodically switches between training mode (where state measurements are collected with high reliability for model improvement) and idle mode (where the trained model is used for estimation). This periodic collection of high-quality measurements resolves the contradiction by limiting the duration of high-resource measurement collection to only when needed for model training, rather than continuously collecting measurements.
3Reliability
If the prediction model is updated frequently, then the reliability of state predictions is improved, but the device complexity increases
Solution Approach 1:
The system segments the operation into distinct training mode and idle mode phases. The model improvement module is activated only during training mode to update the prediction model, while remaining inactive during idle mode. This segmentation reduces the apparent complexity by limiting the active computational burden to specific time periods, making the system more manageable while maintaining frequent updates when needed.
Data Source
Figure 1
Figure 2~4
Figure 3
AI summary
A user equipment (1) for implementing a control in a set of user equipments is provided with a model improvement module (18). Said model improvement module is dedicated to updating a prediction model that is used for inferring state predictions. The model improvement module alternates between a training mode where it determines a new prediction model to be implemented next, and an idle mode. A trade-off is obtained between limited resource requirement and accuracy of a state knowledge. In the idle mode, the model improvement module may assess a validity of the prediction model currently used.