Wireless AI/ML Scheme Evaluation for Stable Network Switching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In AI/ML-based wireless communication systems, there is a challenge in reliably monitoring performance fluctuations due to factors like reporting errors, sample data, and temporary environmental changes, leading to unnecessary scheme switching and signaling overhead.
Innovation Solution
A method for evaluating AI/ML scheme stability at the network side using evaluation results reported by terminal devices, avoiding performance fluctuations and unnecessary scheme switching by determining conditions such as scheme availability and need for updates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If AI/ML-based schemes are used to enhance system performance, then system performance is improved, but performance fluctuations and instability occur due to reporting errors, sample data, and temporary environmental changes
Solution Approach 1:
The network device performs preliminary evaluation of AI/ML scheme performance by receiving multiple pieces of evaluation information before making a final determination. This preliminary assessment allows the system to account for temporary fluctuations and reporting errors before concluding that performance requirements are not met, thereby improving reliability while maintaining stability.
Solution Approach 2:
The system receives more than one piece of evaluation information (W pieces where W≥1) from terminal devices before making a determination. This excessive sampling approach ensures that temporary environmental changes or single-reporting errors do not trigger false performance degradation conclusions, thus balancing performance improvement with stability.
2Reliability
If performance monitoring is performed to detect scheme degradation, then reliability is improved, but unnecessary scheme switching and signaling overhead increase due to false detections
Solution Approach 1:
The network device performs preliminary evaluation by collecting multiple pieces of performance evaluation information before triggering scheme switching or updates. This preliminary filtering prevents false detections caused by temporary environmental changes or reporting errors, thereby improving monitoring accuracy while reducing unnecessary scheme switching operations.
Solution Approach 2:
The system implements a feedback mechanism where terminal devices report AI/ML scheme performance evaluation information to the network device. The network device uses this feedback to determine whether performance requirements are not met, allowing for reliable performance monitoring while avoiding unnecessary scheme switching through intelligent decision-making based on accumulated feedback data.
3Measurement precision
If multiple evaluation criteria are used to assess AI/ML scheme performance, then measurement precision is improved, but signaling overhead and procedure complexity increase
Solution Approach 1:
The network device uses a universal evaluation framework that can handle multiple pieces of performance evaluation information from different terminal devices using the same determination logic. This multi-functional approach allows accurate performance assessment through multiple criteria while avoiding redundant signaling procedures, as the same evaluation mechanism is applied universally to all received information.
Solution Approach 2:
The system merges multiple pieces of performance evaluation information from different terminal devices into a unified determination process. By combining these evaluations and applying a single determination logic, the system achieves high measurement precision through multiple criteria while reducing signaling overhead by consolidating the evaluation procedure rather than treating each criterion separately.
Data Source
Figure 1~2
Figure 3
Figure 4~5
AI summary
The embodiments of the present disclosure provide a method for wireless communication, a terminal device, and a network device. The method includes the following. A network device receives W pieces of information, where the W pieces of information indicate at least one of the following: relevant information of a first event, or relevant information of a second event. The first event include at least one of the following: a first scheme does not meet performance requirements, the first scheme is not recommended for use, the first scheme is unavailable, or a first metric associated with the first scheme does not meet performance requirements; and the second event includes at least one of the following: the first scheme meets performance requirements, the first scheme is recommended for use, the first scheme is available, or the first metric associated with the first scheme meets performance requirements. W is a positive integer. The network device determines whether a first condition is met based on the W pieces of information, where the first condition includes one of the following: a condition for determining that the first scheme does not meet performance requirements, a condition for determining that the first scheme is unavailable for further use, a condition for determining that the first scheme needs to be terminated, and a condition for determining that the first scheme needs to be updated.