RNN-Based Network Traffic Prediction and Bandwidth Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current network traffic prediction technologies face challenges in accurately forecasting network traffic volume and packet distribution due to the non-linear and self-similar nature of network data, which can lead to inefficiencies in bandwidth allocation and increased vulnerability to cyberattacks like DoS attacks.
Innovation Solution
A system employing a recurrent neural network (RNN) that captures network traffic data to generate predictive models of future traffic volume and packet distribution, utilizing LSTM or GRU units to account for short-term and long-term trends, and adjusts bandwidth limits and router configurations based on these predictions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional network traffic prediction methods are used, then the system is simple to implement, but the prediction accuracy is low due to non-linear and self-similar nature of network data
Solution Approach 1:
The patent replaces traditional mechanical/statistical prediction systems with a recurrent neural network (RNN) based system. The RNN uses LSTM or GRU units to process network traffic data, substituting conventional algorithms with a more sophisticated computational model that can capture non-linear and self-similar patterns in network traffic, thereby improving prediction accuracy despite increased complexity.
2Productivity
If bandwidth allocation is not adjusted proactively, then the network configuration remains stable, but network performance deteriorates during traffic surges
Solution Approach 1:
The patent implements preliminary action by using the RNN prediction model to forecast future network traffic patterns before they occur. Based on these predictions, the system proactively adjusts bandwidth allocation and network configuration in advance, ensuring optimal performance during traffic surges while maintaining stability during normal conditions. This predictive approach allows the system to prepare configurations before traffic demands change.
3Reliability
If fixed bandwidth limits are maintained, then the network router operates simply, but the network becomes vulnerable to DoS attacks and traffic anomalies
Solution Approach 1:
The patent implements a feedback mechanism where the RNN continuously monitors network traffic patterns, predicts future behavior, and feeds this information back to dynamically adjust bandwidth limits and router configurations. This closed-loop system enhances security by detecting anomalies and potential DoS attacks through predicted deviations from normal traffic patterns, while automatically adjusting parameters to maintain security without requiring complex manual router configurations.
Data Source
AI summary
A system can include a recurrent neural network operating on one or more computing devices that generates a model that outputs a predicted network traffic for a network infrastructure based on a sample of network traffic of the network infrastructure. The predicted network traffic characterizes a predicted traffic volume and a predicted packet type and distribution in the predicted network traffic. The system can also include a network manager operating on one or more computing devices that commands a network router logically coupled to a point of ingress of the network infrastructure to adjust bandwidth limits controlled by the network router based on the predicted network traffic of the network infrastructure.