Automated Creation of Network Digital Twins

May 11, 2021 by Leslie Provenzano

The Need for Network Digital Twins

Digital twin technology offers significant value to an organization as it allows continuous access to the digital replica of a physical object or process throughout its lifecycle. The digital replica can be used for analysis to provide insights and actionable information to improve the process or product in terms of optimized performance, reduced cost, improved resilience, or better maintenance.

A network digital twin is a computer simulation model of the communication network integrated with its operating environment and the application traffic carried by it. To satisfy its intended purpose, the network digital twin must have sufficient fidelity to accurately reflect the network dynamics due to the interplay between the communication protocols, topology, traffic, and physical environment. A network digital twin can directly incorporate cyber vulnerabilities and defenses. Such a cyber-enhanced network digital twin can be used to assess the cyber resilience of the target system by subjecting the digital twin to live or simulated cyber attacks and analyzing its behavior and resilience under a diverse collection of attack scenarios.

SCALABLE’s Implementation of Network Digital Twins

Given the complexity of most networks, creating an accurate digital twin representing the topology, configuration, and traffic of an existing physical network can be challenging. SCALABLE’s emulation platform, EXata, provides several automated tools that facilitate high-fidelity digital twins of existing networks. Digital twin technology and tools can be used to create an “initial” or “baseline” digital twin of a network which can then be refined or extended as appropriate to suit the analyst’s needs.

Figure 1 depicts SCALABLE’s approach to creating network digital twins.


Creating A Network Digital Twins

The following sections outline the steps and tools for creating network digital twins. While many of the steps are common to all types of network digital twins, some steps and tools depend on the type of network that the twin represents (a physical or constructive network) and its intended use (network performance or cyber resilience assessment).

Scenario Designer is used to configure the network digital twin and includes:
•Network topology, device characteristics, and traffic.
•For wireless (sub)-networks, the network digital twin includes terrain, environmental conditions like weather, and platform mobility
• For cyber resilience assessment, the network digital twin may include cyber device configurations, vulnerabilities, firewall rules, and cyber-attacks

The Scenario Designer includes two major components: Scenario Importer and the SCALABLE GUI. The Scenario Importer automatically creates the network digital twin from external sources. The SCALABLE GUI can create a network digital twin from scratch or modify any part of an automatically-created network digital twin for what-if analysis.

Scenario Importer can import device, traffic, and cyber characteristics from an existing (As Is) network.
• Scenario Importer may use the Topology Converter to convert a network topology specified in other formats that include live network scans, MBSE tools, Visio diagrams, and network simulators
• Scenario Importer may use the Extractor to retrieve (a subset of) the relevant topology and device information from an external simulation (e.g., OneSAF, AFSIM, STK, VR-Forces, STAGE, ASCOT, and other mission simulations

The Scenario Designer stores multiple configurations in libraries to facilitate mixing and matching in Experiment Designer

Experiment Designer
•Chooses configurations from the libraries created by Scenario Designer (network, traffic, cyber, environment, mobility) to create experiments, e.g., running a live traffic recording on various wireless topologies
• Defines analysis objectives, including key performance parameters (KPPs) of interest (e.g., latency, packet drop, throughput, jitter)
• Specifies the execution environment for the network digital twin (local machine, server, cloud) and connections to live equipment or external simulations

EXata Engine
• Creates the network digital twin and executes the experiments, connecting to specified live equipment, live applications, and external simulations

• Interacts with the network digital twin at run-time to provide the ability to inspect and debug experiment configurations

• Superimposes networks on 3D environments to create a dynamic operational picture for visualizing entities, movement, terrain (including undersea), active network connections, packet routing, delivery, data rates, cyber state
• Enables human-in-the-loop interactions and launching of electronic warfare and cyber-attacks

• Used to display a variety of statistical measures (including the KPPs specified in the Experiment Designer)
• Displays heat maps, statistics over time
• Compares results across experiments

Report Generator
• Facilitates the generation of useful reports from the experiment statistics
• Reports in tabular and graphical form can be generated using external tools such as Excel or Tableau

Using digital twin technology and tools provides a cost-effective way to perform analysis, testing, and optimization. Live hardware and software applications can be seamlessly interfaced with, or integrated into, a network digital twin that executes in real-time. These real-time network digital twins can then be used to improve management, performance, and cyber resilience of networks in all domains, from commercial enterprise and IoT to multi-domain military networked systems operating from seabed to space.

More details on using digital twin technology for performance analysis can be found in the complete White Paper, including a detailed case study. In the paper, we describe our approach to creating network digital twins and using them for analyzing network performance through an example. In this example, we will analyze the performance of the gaming and training software VBS3 when it is run on a geographically distributed network.