Adaptive Digital Twin and Communication-Efficient Federated Learning Network Slicing for 5G-enabled Internet of Things

TL;DR

This work tackles dynamic network slicing for 5G-enabled IIoT under privacy constraints by combining a Graph-Attention Network–based Digital Twin (DT) for real-time demand forecasting with a Federated Multi-Agent Reinforcement Learning (DT-MAFL) framework that allocates resources via Deep Deterministic Policy Gradient (DDPG) while preserving slice privacy. The DT captures spatiotemporal traffic patterns and learns adaptive graph structures to forecast slice demand $d_m$, while MAFL enables decentralized policy learning with occasional global aggregation to maximize the per-slice QoS utility $U_m$ and utilization $\,\Omega_m$ under resource constraints. Key contributions include the DT for cross-slice demand forecasting with dynamic adjacency learning, a MAFL architecture that couples local DDPG updates with a weighted global model, and comprehensive simulations showing improved forecast accuracy and reduced communication overhead compared to baselines. The framework offers a scalable, privacy-preserving approach for realistic IIoT network operation, enabling better resource utilization and QoS in heterogeneous slices. Future work explores policy distillation to handle Non-IID slice requirements and further enhance learning robustness.

Abstract

Network slicing enables industrial Internet of Things (IIoT) networks with multiservice and differentiated resource requirements to meet increasing demands through efficient use and management of network resources. Typically, the network slice orchestrator relies on demand forecasts for each slice to make informed decisions and maximize resource utilization. The new generation of Industry 4.0 has introduced digital twins to map physical systems to digital models for accurate decision-making. In our approach, we first use graph-attention networks to build a digital twin environment for network slices, enabling real-time traffic analysis, monitoring, and demand forecasting. Based on these predictions, we formulate the resource allocation problem as a federated multi-agent reinforcement learning problem and employ a deep deterministic policy gradient to determine the resource allocation policy while preserving the privacy of the slices. Our results demonstrate that the proposed approaches can improve the accuracy of demand prediction for network slices and reduce the communication overhead of dynamic network slicing.

Figures (4)

• Figure 1: DT-MAFL dynamic network slicing architecture.
• Figure 2: Forecasting Performance by various models.
• Figure 3: Convergence performance.
• Figure 4: Resource Utilization and QoS satisfaction performance.