Bandwidth Utilization Prediction via Machine Learning Periodicity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for analyzing bandwidth utilization data in network customers fail to accurately predict future usage patterns, leading to potential network issues and capacity mismanagement.
Innovation Solution
A method using machine learning, specifically regression techniques with kernel functions, to train models on historical bandwidth data and predict future utilization based on periodic patterns, allowing for visual comparison of historical and predicted data to identify discrepancies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional bandwidth utilization analysis methods are used, then implementation is simple, but prediction accuracy of future usage patterns is poor
Solution Approach 1:
The system performs preliminary training of machine learning models using historical bandwidth utilization data before actual prediction is needed. This pre-processing step captures periodic patterns and establishes baseline models that can then be used for accurate real-time predictions without requiring complex calculations during operational phase.
Solution Approach 2:
The patent introduces machine learning models as intermediary components between historical data and future predictions. These models act as mediators that learn from periodic patterns in historical data and translate them into accurate predictions, bridging the gap between simple data collection and complex prediction requirements.
2Measurement precision
If machine learning models are trained on historical data, then prediction accuracy improves, but computational complexity and training time increase
Solution Approach 1:
The system performs model training in advance using historical bandwidth utilization data, completing the computationally intensive work before predictions are needed. This allows the system to establish accurate models during off-peak periods and then use these pre-trained models for rapid real-time predictions during operational phases.
Solution Approach 2:
The system leverages the periodic nature of bandwidth utilization patterns to train models on representative historical cycles. By focusing training on capturing periodic behaviors rather than processing every possible scenario, the system achieves accurate predictions with reduced training time and computational resources.
3Ease of operation
If visual display of historical and predicted bandwidth utilization is provided, then customer ability to detect issues improves, but information processing and display complexity increases
Solution Approach 1:
The visualization system segments bandwidth utilization data into distinct time periods (historical vs. predicted) and displays them separately but comparably. This segmentation allows customers to easily distinguish between actual past performance and forecasted future usage, making issue detection straightforward without overwhelming the user with undifferentiated data.
Solution Approach 2:
The system uses different visual representations (such as color coding or line styles) to distinguish between historical bandwidth utilization data and predicted future data. This visual differentiation enables customers to quickly identify discrepancies and potential issues by comparing the two datasets at a glance without complex analysis.
Data Source
AI summary
A method for providing a predicting bandwidth utilization for a customer of a connectivity service provider is disclosed. The method includes receiving bandwidth utilization data over time for a customer of a connectivity service provider. The method further includes training a machine learning model for the bandwidth utilization data during a time period based on the received data. In an embodiment, training the machine learning model includes determining a value for a parameter. The bandwidth utilization may be substantially periodic over the time period. Relying on the periodicity of the bandwidth utilization, the method further includes predicting, using the trained model, future bandwidth utilization over time.