A network generator for covert network structures

Area of Science:

• Network Science
• Computational Social Science
• Cybersecurity

Background:

• Covert networks, such as criminal or terrorist organizations, operate with incomplete data due to members' efforts to conceal activities and associations.
• Understanding the organizational structures of these networks is crucial for intelligence and security analysis.
• Existing methods struggle to generate realistic synthetic networks that capture the complexities of covert structures.

Purpose of the Study:

• To introduce a novel method for generating statistically similar synthetic covert networks.
• To model both the edge structure and hierarchical organization of covert networks.
• To provide tools for analyzing network stability and for anonymizing sensitive network data.

Main Methods:

• A novel rewiring method for covert networks, parameterized by edge connectivity standard deviation.
• Modeling higher-level organizational structures using multi-layer networks.
• Utilizing the Stochastic Block Model for the lowest network level.
• Generating numerous synthetic networks from original covert network data.

Main Results:

• Generated synthetic networks are statistically similar to themselves and the original network.
• Modeling edge structure and hierarchy together is essential for generating realistic networks.
• A small percentage (18%) of synthetic network structures were consistently repeated, indicating stable organizational patterns.
• Identified frequently repeating structures as strong candidates for ground truth network architecture.

Conclusions:

• The developed method effectively generates synthetic covert networks that preserve statistical properties of the original.
• The approach allows for the identification of stable, frequent structures within covert networks, aiding in analysis.
• Synthetic networks can be utilized for anonymization and testing software in open research settings.