Neural Network Traffic Projection for Change Window Optimization

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current network traffic management systems lack effective tools for visualizing historical traffic patterns and projecting future traffic to optimize change management activities, such as system maintenance, leading to potential disruptions and inefficiencies in downtime planning.

Innovation Solution

A network traffic projection and visualization system that uses a neural network to generate projections based on historical data, incorporating change management data to display potential impact on infrastructure assets through a graphical user interface, allowing administrators to identify optimal change windows and assess disruption factors.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If change management activities are scheduled without historical traffic data analysis, then change management process is simple and quick, but system disruptions increase and operational reliability deteriorates

Engineering Contradiction:

Improveoperational reliabilityVSAvoidchange management process complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The system performs preliminary analysis of historical network traffic data to identify patterns and predict future traffic before scheduling change management activities. This allows administrators to proactively select optimal downtime windows that minimize disruption to business operations, thereby improving operational reliability without adding significant complexity to the change management process.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system incorporates feedback loops that continuously monitor actual network traffic against predicted traffic patterns. This feedback mechanism allows the system to refine its predictions over time and provide increasingly accurate guidance for change management scheduling, improving reliability while maintaining process simplicity through automated learning.

Inventive Principle:

Principle #23Feedback

2Productivity

If detailed network traffic projection is generated using neural networks, then change window optimization improves, but computational resources and processing time increase

Engineering Contradiction:

Improvechange window optimization efficiencyVSAvoidcomputational resource consumption

Core Design Contradiction:

ProductivityVSUse of energy by moving object

Solution Approach 1:

The system dynamically adjusts the complexity of neural network models and the granularity of traffic projection based on the specific change management scenario. For critical systems requiring high-precision optimization, the system employs more complex models with finer time resolutions. For less critical changes, it uses simplified models with coarser projections, thereby optimizing productivity while controlling computational resource consumption according to actual needs.

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The system applies partial neural network analysis only to the most critical network segments and time periods relevant to the planned change. Rather than analyzing entire network traffic across all possible time windows, it focuses computational resources on the specific change windows being evaluated, improving optimization efficiency while reducing overall computational consumption through targeted analysis.

Inventive Principle:

Principle #16Partial or excessive action

3Measurement precision

If historical network traffic data is collected and analyzed, then future traffic projection accuracy improves, but data storage requirements and system complexity increase

Engineering Contradiction:

Improvetraffic projection accuracyVSAvoiddata storage volume

Core Design Contradiction:

Measurement precisionVSQuantity of substance

Solution Approach 1:

The system extracts and stores only the most relevant features from raw network traffic data, such as traffic patterns by time of day, day of week, and seasonal variations. It discards redundant raw packet data and keeps only the processed metrics needed for accurate projection, thereby improving measurement precision while significantly reducing data storage requirements compared to storing complete raw traffic records.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The system segments historical traffic data into meaningful categories such as business hours, weekends, holidays, and seasonal patterns. This segmentation allows the neural network to process and store compressed representations of traffic behavior rather than raw continuous data, reducing storage volume while maintaining or improving projection accuracy through structured data organization.

Inventive Principle:

Principle #1Segmentation

Data Source

PatentUS12107736B1Systems and methods for visualization based on historical network traffic and future projection of infrastructure assets

Publication Date: 2024.10.01 FANNIE MAE

AI summary

Disclosed herein are systems, methods, and storage media for network traffic projection and visualization. A computing system includes at least one circuit structured to receive network traffic data. A neural network is generated based on the network traffic data and includes a network traffic projection. The network traffic projection is displayed, via a graphical user interface, to a system administrator. In some embodiments, the computing system includes at least one circuit structured to receive change management data, such as application- and outage-related information. The change management data is combined with the network traffic projection data in a change window simulator, which recommends one or more change windows.