Communication Quality Deterioration Prediction Using Dual Learning Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems are inadequate in predicting communication quality deterioration with high accuracy, which is crucial for timely intervention in communication devices, as they primarily focus on malfunction prediction rather than quality degradation.
Innovation Solution
A communication quality deterioration prediction system that generates two learning models, one focused on internal causes within the communication section and another on external factors, using machine learning to predict packet loss and integrate results for accurate future quality assessment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single learning model is used to predict communication quality deterioration, then the system complexity is low, but the prediction accuracy is insufficient
Solution Approach 1:
The patent divides the prediction system into multiple specialized learning models: a first learning model that focuses on communication section-specific factors (weather, line quality) and a second learning model that focuses on external factors (device performance, traffic conditions). Each model is trained on specific types of data relevant to its domain, allowing for more accurate predictions while keeping each individual model relatively simple and manageable.
2Reliability
If prediction is made based on general malfunction occurrence, then the system is simple to implement, but the prediction timing is too late for effective intervention
Solution Approach 1:
The system performs preliminary prediction of communication quality deterioration by analyzing current trends in communication section conditions and external factors. By predicting quality deterioration before actual malfunction occurs, the system enables advance intervention and preventive maintenance, allowing operators to address issues before they cause complete system failures.
3Measurement precision
If only internal communication section factors are considered for prediction, then the model is simple to train, but the prediction accuracy is limited
Solution Approach 1:
The patent segments the prediction factors into two distinct categories: communication section-specific factors (handled by the first learning model) and external factors (handled by the second learning model). This segmentation allows each model to be trained on specialized data with appropriate features and algorithms, improving overall prediction accuracy while maintaining model simplicity and interpretability.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A first prediction unit 83 predicts the amount of future communication quality deterioration or the presence or absence of occurrence of future communication quality deterioration in a communication section between one communication device and another communication device communicably connected to the one communication device by using a first learning model generated on the basis of first attributes being attributes related to a cause of communication quality deterioration in the communication section. A second prediction unit 84 predicts the amount of future communication quality deterioration or the presence or absence of occurrence of future communication quality deterioration outside the communication section regarding one communication device by using a second learning model generated on the basis of second attributes being attributes related to a cause of communication quality deterioration outside the communication section regarding the one communication device. A determination unit 85 determines the amount of future communication quality deterioration or the presence or absence of occurrence of future communication quality deterioration regarding one communication device on the basis of a prediction result of the first prediction unit and a prediction result of the second prediction unit.