Mobile Network ML Model Validation for Radio Resource Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
There is no guarantee that predictive machine learning models in mobile networks are working properly, leading to potential poor network performance due to issues like corrupted models or lack of experience in certain situations.
Innovation Solution
Implement measures for evaluating and controlling predictive machine learning models by receiving information on radio resource management functions, obtaining behavior information, measuring network conditions, determining prediction and behavior results, and evaluating model validity based on these factors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If predictive machine learning models are deployed in mobile networks, then network performance and mobility efficiency are improved, but model validity and reliability cannot be guaranteed due to corruption or lack of experience in certain situations
Solution Approach 1:
The patent implements a feedback mechanism where the mobile terminal evaluates the predictive model's predictions against actual network behavior. The terminal compares predicted handover targets with actual handover outcomes and network conditions, then reports evaluation results back to the network. This feedback loop enables continuous validation and refinement of model validity, ensuring reliability while maintaining productivity benefits.
Solution Approach 2:
The mobile terminal autonomously performs evaluation of the predictive model's validity by comparing predictions with actual network behavior. The terminal independently determines whether the model is working properly based on local measurements and observations, without requiring constant network intervention. This self-service approach ensures continuous reliability monitoring while maintaining network performance optimization.
2Reliability
If predictive models are continuously evaluated and refined, then model reliability is improved, but signaling overhead and system complexity increase
Solution Approach 1:
The patent implements partial evaluation by the mobile terminal, where the terminal performs local validation of model predictions against actual network behavior. Instead of comprehensive continuous evaluation, the terminal selectively evaluates models based on specific conditions and triggers, reducing signaling overhead while maintaining sufficient reliability. The terminal reports only when evaluation results indicate model validity issues or when configured thresholds are exceeded.
3Speed
If predictive models are downloaded to mobile terminals for local inference, then processing speed is improved, but device complexity and memory requirements increase
Solution Approach 1:
The patent segments the machine learning system into two parts: the predictive model is downloaded to the mobile terminal for fast local inference, while the model evaluation and validation functionality is distributed to the terminal. This segmentation allows speed improvement through local prediction while managing device complexity by offloading complex evaluation tasks to the network based on terminal reports.
Solution Approach 2:
The predictive model is downloaded to the mobile terminal in advance before it is needed for inference. This preliminary action enables fast local predictions without real-time network communication delays, improving speed. The model is prepared and stored on the terminal beforehand, reducing processing latency when predictions are required for handover decisions.
Data Source
AI summary
There are provided measures for evaluation and control of predictive machine learning models in mobile networks. Such measures exemplarily comprise receiving information on a predictive model related to a radio resource management function, obtaining behavior information on an intended behavior of said predicted model, obtaining difference determination information on difference determination with respect to a predictive model prediction and said intended behavior, measuring a network condition, determining a prediction result based on said network condition and said information on said predictive model, determining a behavior result based on said network condition and said behavior information, and evaluating validity of said predictive model based on said prediction result, said behavior result, and said difference determination information.