Automatic Generation of Digital Twins for Network Testing
Shenjia Ding, David Flynn, Paul Harvey
TL;DR
The paper addresses the challenge of validating software for autonomous networks, proposing automatic generation of digital twins (DTs) to streamline testing within ITU-T's experimental evaluation framework. It introduces a unit-twin concept and leverages Automated Machine Learning (AutoML) with AutoGluon and Auto-sklearn to build data-driven DTs from synthetic network data, aiming to reduce manual design effort and validation time. Preliminary experiments on a simple diamond topology show DTs can achieve near-penetrating accuracy (≈0.99 on Path1, ≈0.92–0.94 on Path2) with DT inference orders of magnitude faster than full Mininet simulations, and a substantial overall pipeline speed-up when considering data generation and training. The findings suggest AutoML-driven DT generation is a feasible, scalable approach for validating autonomous network controllers, with future work expanding topology complexity, dynamic traffic, and generalizability across scenarios.
Abstract
The increased use of software in the operation and management of telecommunication networks has moved the industry one step closer to realizing autonomous network operation. One consequence of this shift is the significantly increased need for testing and validation before such software can be deployed. Complementing existing simulation or hardware-based approaches, digital twins present an environment to achieve this testing; however, they require significant time and human effort to configure and execute. This paper explores the automatic generation of digital twins to provide efficient and accurate validation tools, aligned to the ITU-T autonomous network architecture's experimentation subsystem. We present experimental results for an initial use case, demonstrating that the approach is feasible in automatically creating efficient digital twins with sufficient accuracy to be included as part of existing validation pipelines.