AI-Based Prediction Model Evaluation for Faster OTA Testing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing over-the-air (OTA) testing process for wireless communication devices is complex and costly, leading to long testing times and high equipment and labor costs, despite its importance in evaluating device performance and compliance with regulations.
Innovation Solution
Implementing an AI/ML model to predict received power of unmeasured signals or beams, reducing the need for comprehensive multi-angle testing by using a single angle of arrival (AoA) and adjusting transmission power to simulate various environments, thereby simplifying the test system architecture and procedures.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If comprehensive multi-angle OTA testing is conducted to ensure accurate device performance evaluation, then measurement precision is improved, but testing time and device complexity increase significantly
Solution Approach 1:
The patent applies preliminary action by training an AI/ML model in advance using comprehensive multi-angle measurement data. Once trained, the model can predict device performance for new test cases without requiring actual multi-angle measurements, thus reducing testing time while maintaining evaluation accuracy.
Solution Approach 2:
The patent uses copying by creating virtual measurement results through AI/ML model predictions. Instead of physically conducting time-consuming multi-angle OTA measurements for each test case, the system copies the essence of comprehensive testing through predicted values generated by the trained model.
2Measurement precision
If comprehensive multi-angle OTA testing is conducted to ensure accurate device performance evaluation, then measurement precision is improved, but device complexity and equipment requirements increase
Solution Approach 1:
The patent replaces the mechanical/physical OTA measurement system with an AI/ML-based computational system. Instead of requiring complex physical test setups with multiple antennas and rotation mechanisms, the system uses a trained model to predict performance, significantly simplifying the test equipment while maintaining measurement precision.
3Productivity
If AI/ML model prediction is used to reduce testing requirements, then productivity is improved, but measurement precision may be compromised
Solution Approach 1:
The patent ensures measurement precision is maintained by performing preliminary training of the AI/ML model using comprehensive, high-precision multi-angle measurement data. This preliminary action creates a robust model that can accurately predict performance while enabling fast testing for production evaluation.
Data Source
AI summary
Methods and apparatus pertaining to performance evaluation of a prediction model utilized in mobile communications are described. A test equipment transmits one or more reference signals to a communication apparatus. Then, the test equipment receives one or more prediction results from the communication apparatus. The one or more prediction results are output by a model of the communication apparatus based on the one or more reference signals. The test equipment further determines a performance indicator of the model according to the one or more prediction results.