Highlights:
- AI-integrated Digital Twins enhance real-time forecasting in 5G and 6G networks.
- LSTM neural networks predict network traffic with high accuracy.
- Proactive resource management enables more efficient, resilient, and autonomous network operation.
- New research by John Sengendo and Fabrizio Granelli showcases superior performance against traditional methods.
TLDR:
A new AI-powered Digital Twin framework, developed by John Sengendo and Fabrizio Granelli, integrates LSTM neural networks to predict network traffic and manage resources dynamically, paving the way for intelligent self-optimizing 5G and 6G infrastructure.
As global communication networks prepare for the transition from 5G to 6G, researchers are developing innovative solutions to handle unprecedented levels of data demand and service diversity. The latest advance comes from John Sengendo and Fabrizio Granelli, who have introduced an AI-enabled Digital Twin (DT) framework designed to forecast network traffic and manage resources proactively. Their study, recently accepted for presentation at the IEEE Global Communications Conference (GLOBECOM) 2025, highlights how the fusion of artificial intelligence and digital twinning is reshaping the architecture of next-generation networks.
Traditional network management relies heavily on heuristic models, which struggle with the complexity, agility, and scalability required in ultra-dense network deployments. The authors propose a transformative alternative: a Digital Twin Network (DTN) that mirrors the real-world infrastructure virtually. This DTN allows network operators to model, analyze, and test different operational scenarios without affecting live services. By embedding an AI-driven engine within the DT framework, the system continuously learns and adapts to evolving traffic conditions, optimizing resource allocation in real time and minimizing latency and service degradation.
At the core of this innovation lies a Long Short-Term Memory (LSTM) neural network—an advanced deep learning algorithm tailored for temporal forecasting. Trained on historical traffic data, the LSTM model captures long-term dependencies, identifying complex patterns in traffic behavior. This predictive capability empowers operators to anticipate network congestion and automatically reallocate bandwidth or power resources before service quality is impacted. Analytical experiments conducted by Sengendo and Granelli demonstrate that their AI-enhanced DT framework achieves superior performance benchmarks compared with baseline approaches, marking a significant leap toward fully autonomous, self-optimizing 6G networks.
Their work underscores a critical direction for the telecommunications industry: embedding intelligence within digital twins to build adaptive, high-performance networks capable of meeting the real-time needs of emerging applications such as autonomous vehicles, smart cities, and industrial IoT. This fusion of AI and DT technology promises not only operational efficiency but also sustainability, as it optimizes energy and spectrum use across the network landscape.
Source:
Sengendo, John & Granelli, Fabrizio (2025). “AI-Enabled Digital Twins for Next-Generation Networks: Forecasting Traffic and Resource Management in 5G/6G.” arXiv:2510.20796 [cs.NI]. https://doi.org/10.48550/arXiv.2510.20796