Graph Generator Algorithm for Network Dynamics Simulation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current technologies lack effective methods for generating and simulating complex graphs that represent real-world networks, particularly social graphs, which are crucial for understanding and mitigating the spread of phenomena like epidemics and malware within networks.

Innovation Solution

A graph generator algorithm that takes input data such as a starting model, target distributions, and a limit value to iteratively refine a graph, allowing for the creation of a graph that accurately represents network connections and dynamics, enabling simulations of real-world applications like infection spread and malware propagation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If a graph is generated to represent network connections, then the ability to simulate network dynamics is improved, but the complexity of the graph generation process increases

Engineering Contradiction:
Improveability to simulate network dynamicsVSAvoidgraph generation process complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The graph generation process is segmented into distinct iterative steps: initializing a graph with N objects, calculating a temporary distribution, comparing with target distribution, modifying the graph if needed, and repeating until convergence. This segmentation makes the complex generation process manageable and systematic.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The algorithm modifies graph parameters (connections between objects) iteratively based on the difference between temporary and target distributions. The graph structure is dynamically adjusted by adding, removing, or modifying connections to achieve the desired target distribution, thereby generating complex graphs through controlled parameter changes.

Inventive Principle:
Principle #35Parameter changes

2Measurement precision

If the graph represents social networks with many objects, then the accuracy of network modeling is improved, but the computational resources required increase

Engineering Contradiction:
Improvenetwork modeling accuracyVSAvoidcomputational resources
Core Design Contradiction:
Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The algorithm generates graphs with a specified number of objects N, which may be less than the complete set of real-world objects. This partial generation approach provides sufficient accuracy for simulation purposes while significantly reducing computational resource requirements compared to modeling every individual object in the real network.

Inventive Principle:
Principle #16Partial or excessive action

Solution Approach 2:

The algorithm creates a simplified copy of the real network structure that captures essential characteristics (target distribution) without replicating every detail. This abstract representation maintains modeling accuracy while reducing computational complexity through generalization rather than exhaustive representation.

Inventive Principle:
Principle #26Copying

3Manufacturing precision

If the graph is modified iteratively to match target distribution, then the quality of graph representation is improved, but the generation time increases

Engineering Contradiction:
Improvegraph representation qualityVSAvoidgraph generation time
Core Design Contradiction:
Manufacturing precisionVSLoss of time

Solution Approach 1:

The algorithm incorporates feedback by comparing the temporary distribution against the target distribution at each iteration. Based on this comparison, the graph is modified in a directed manner to reduce discrepancies, ensuring convergence toward the target representation. This feedback mechanism guarantees quality while controlling generation time through systematic improvement.

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The algorithm performs preliminary actions by initializing the graph with a random structure before iterative modification. This initial setup provides a starting point that requires fewer iterations to converge to the target distribution, thereby reducing overall generation time while maintaining representation quality.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4071654A1Method and graph generator for generating a graph of the relationships between objects in networks
Publication Date: 2022.10.12 DEUTSCHE TELEKOM AG
  • EP4071654A1 patent drawingFigure 1~2
  • EP4071654A1 patent drawingFigure 3
  • EP4071654A1 patent drawing

AI summary

Techniques for generating a graph, wherein the generated graph represents connections, in particular logical connections, between objects, comprising the following steps: • Providing a computing unit with a processor suitable for executing an algorithm for generating a graph, wherein the algorithm is implemented on the computing unit; • Passing the following input data to the algorithm: o an initial output model of a graph; o an initial target distribution of the objects of the graph at a time Tn, wherein the initial target distribution represents initial properties of the objects; o a limit value GW as a termination criterion when comparing target distributions; • Performing the following steps by the algorithm: o Computing a temporary distribution of the objects based on the initial model of the graph;• Comparing the temporary distribution and the first target distribution and calculating a value J, where the value D quantifies a difference between the temporary distribution and the first target distribution; • Outputting the initial graph model as the generated graph G(Tn) if the value J falls below the specified limit GW, otherwise modifying the initial graph model and repeating the aforementioned steps using the algorithm.