Machine Learning Node Model Generation for Network Simulation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing network modeling technologies fail to accurately generate models considering internal node configurations and control protocol processing, limiting their effectiveness in simulating network performance.

Innovation Solution

A machine learning device that inputs setting information, applies traffic, measures performance, and generates node models using collected data, including internal configuration and control protocol processing information, to create a network model that reflects actual node behavior.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If network modeling technology from Non Patent Literature 1 is used, then network topology can be modeled, but internal node configuration (queues, processing speed, parallel execution) cannot be considered

Engineering Contradiction:

Improvenetwork topology modeling capabilityVSAvoidnode internal configuration accuracy

Core Design Contradiction:

Adaptability or versatilityVSManufacturing precision

Solution Approach 1:

The patent segments the network modeling process into two distinct components: (1) network topology modeling that captures node connections and structures, and (2) node internal configuration modeling that captures queues, processing speeds, and parallel execution capabilities. This segmentation allows each component to be modeled with appropriate detail and accuracy independently.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent introduces machine learning models as intermediary components that bridge the gap between topology-level abstraction and detailed node configuration. The ML models learn node behavior patterns from data and generate accurate internal configuration parameters without requiring direct measurement of every node attribute.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Measurement precision

If network modeling technology from Non Patent Literature 2 is used, then single router performance can be predicted, but network topology and control protocol processing cannot be considered

Engineering Contradiction:

Improvesingle node performance prediction accuracyVSAvoidnetwork topology and control protocol consideration

Core Design Contradiction:

Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The patent creates a universal modeling framework that can handle multiple functions simultaneously: (1) single node performance prediction, (2) network topology modeling, and (3) control protocol processing analysis. The machine learning models are designed to be multi-functional, processing various types of network data and generating comprehensive performance predictions that incorporate all three aspects.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Solution Approach 2:

The patent merges previously separate modeling approaches into a unified system. It combines topology modeling capabilities with node performance prediction and control protocol analysis into a single integrated machine learning framework that processes network data holistically and generates comprehensive performance predictions.

Inventive Principle:

Principle #5Merging (Combining)

3Ease of manufacture

If traditional modeling approaches are used, then model generation is simplified, but accuracy in reflecting actual network performance characteristics is reduced

Engineering Contradiction:

Improvemodel generation simplicityVSAvoidnetwork performance characteristics accuracy

Core Design Contradiction:

Ease of manufactureVSManufacturing precision

Solution Approach 1:

The patent replaces traditional mechanical/mathematical modeling approaches with machine learning-based modeling. Instead of manually constructing models based on theoretical equations and assumptions, the system uses ML algorithms to automatically learn network performance characteristics from data, achieving higher accuracy while maintaining ease of model generation through automated training processes.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS20250005437A1Machine learning device, machine learning program, simulation device, and simulation program

Publication Date: 2025.01.02 NIPPON TELEGRAPH & TELEPHONE CORP

AI summary

A node model generation unit (2) includes a node setting unit (29) that inputs predetermined setting information to a node (11) of a network, a traffic generation unit (21) that applies predetermined traffic to the node (11) on the basis of predetermined applied traffic information, a performance measurement/capture unit (25) that acquires a performance measurement result obtained by measuring performance of the node (11) to which the predetermined setting information is set and to which the predetermined traffic is applied and captures (acquires) output traffic, a learning data input processing unit (24) that collects the predetermined setting information, the predetermined applied traffic information, a performance measurement result of the node (11), and the output traffic and registers the collected information as learning data, and a learning execution unit (27) that generates a node model of the node (11) by machine learning based on learning data registered by the learning data input processing unit (24).