Machine-Learning Routing Value Selection for Variable Network Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing network management systems face inefficiencies in resource allocation due to high variability in the expense-to-efficiency ratio across data transmissions, leading to suboptimal utilization.
Innovation Solution
A computer-implemented method using machine-learning algorithms to analyze historical network traffic data, generate predictive outputs, and adjust routing values based on feedback to optimize data packet routing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If communication management tools allocate resources based on average expense-to-efficiency ratio, then resource allocation is simplified, but resource utilization efficiency deteriorates due to high variability in ETE ratio across different data transmissions
Solution Approach 1:
The patent implements dynamic routing value adjustment by training machine learning models on historical network traffic data to predict optimal routing values for different data transmissions. Instead of using a static average ETE ratio, the system continuously updates routing values based on learned patterns from historical data, allowing resource allocation to adapt to varying network conditions and transmission characteristics, thereby resolving the contradiction between allocation simplicity and utilization efficiency
Solution Approach 2:
The system employs feedback mechanisms by using historical network traffic data as training input for machine learning models. The models learn from past transmission outcomes (including actual ETE ratios achieved) and use this feedback to improve future routing decisions. This closed-loop approach enables the system to overcome the limitations of average-based allocation by continuously refining resource allocation based on actual performance feedback
2Measurement precision
If machine-learning models are trained on historical network traffic data to predict routing values, then resource allocation accuracy is improved, but system complexity increases
Solution Approach 1:
The patent implements self-service by enabling the system to automatically train and update its own machine learning models using its historical performance data. The network management system autonomously learns optimal routing strategies without requiring external intervention or manual configuration, thereby managing the increased complexity through self-optimization capabilities while maintaining high prediction accuracy
Solution Approach 2:
The system performs preliminary action by pre-training machine learning models on historical network traffic data before actual routing decisions are made. This advance preparation allows the models to learn patterns and relationships in the data, establishing predictive capabilities that improve routing accuracy while the complexity is managed through offline model training rather than real-time computation
Data Source
AI summary
The present disclosure generally relates to a method that involves in training a machine-learning model using historical data to generate predictive outputs for future routing performance. The method determines optimal routing values based on the model's predictions and uses a reinforcement learning algorithm to select these values from a set of candidates. A positive transmission feedback is generated when a data packet is successfully transmitted, which updates the selection model, thereby refining future routing decisions. The determined routing values are transmitted to a network management system to optimize data packet routing through the network.