Predictive Network Traffic Management via SDN Bandwidth Provisioning
Solution Overview
Problem
Current network traffic management in software-defined networks (SDNs) is reactive, leading to over-provisioning in some areas and under-provisioning in others due to sudden traffic density shifts and transient surges, as static deployment methods fail to meet dynamic demands, especially with the increasing use of mobile devices and events that generate high internet traffic.
Innovation Solution
A predictive network traffic management system that uses machine learning and data analytics to collect and analyze events from online sources, such as news and social media, to forecast potential network traffic impacts, identify best-matching traffic patterns, and proactively adjust bandwidth provisioning through an SDN controller, employing techniques like multi-labeler active learning and OpenFlow for dynamic network reconfiguration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If static deployment of network resources according to predefined workload patterns is used, then network configuration is simple and manageable, but it is insufficient to meet dynamic demands caused by sudden traffic density shifts and transient surges
Solution Approach 1:
The patent implements dynamic network resource allocation by continuously monitoring traffic patterns and automatically adjusting bandwidth provisioning in real-time. The system transitions from static predefined configurations to dynamic adaptive provisioning, where network resources are automatically scaled up or down based on actual traffic demands, thereby resolving the contradiction between adaptability and complexity.
Solution Approach 2:
The system employs feedback mechanisms by monitoring network traffic patterns and using this information to automatically adjust resource allocation. The closed-loop control system detects traffic density shifts and transient surges, then provisions bandwidth accordingly, enabling the network to adapt dynamically without manual intervention while maintaining manageable complexity through automation.
2Reliability
If reactive network traffic management is used, then response to traffic changes is straightforward, but over-provisioning in some areas and under-provisioning in others occurs due to sudden traffic density shifts
Solution Approach 1:
The system performs preliminary actions by proactively provisioning network bandwidth based on predicted traffic patterns before actual traffic surges occur. By analyzing historical data and traffic trends, the system anticipates demand increases and pre-allocates resources, preventing both over-provisioning and under-provisioning while improving reliability and reducing bandwidth waste.
Solution Approach 2:
The patent dynamically changes network parameters such as bandwidth allocation, routing paths, and resource provisioning levels based on real-time traffic conditions. This continuous parameter adjustment enables the system to optimize bandwidth utilization, preventing over-provisioning during low traffic periods and under-provisioning during surge periods, thereby improving both reliability and efficiency.
3Productivity
If machine learning and data analytics are used to predict network traffic, then proactive bandwidth provisioning is achieved, but system complexity and computational requirements increase
Solution Approach 1:
The system implements self-service by using machine learning algorithms that automatically learn from historical traffic data and make autonomous predictions about future traffic patterns. The system provisions bandwidth without human intervention, with the machine learning model continuously improving its predictions through self-learning, thereby improving productivity while managing complexity through automation rather than manual processes.
Data Source
AI summary
Predictive network traffic management in a computer network. Information relating to events likely to have an adverse impact on network traffic, based on predefined parameters of network performance, is collected from the Internet. Details that may be used to predict network traffic density is extracted from the collected events information, and a predicted traffic density, based on the details is generated. Based on the predicted network traffic density, a traffic pattern is generated. A best-matching traffic pattern template to the generated traffic pattern is identified in a store of traffic pattern templates. Based on the best-matching traffic pattern template, bandwidth on the network is provisioned via an SDN controller.