Network Digital Twin for Route Optimization in 5G/B5G Transport Slicing with What-If Analysis

TL;DR

The paper tackles dynamic QoS-aware route optimization in 5G/B5G transport networks by introducing an experimental Network Digital Twin (NDT) that pairs a GNN-based virtual twin with a what-if recommendation system to test routing decisions before deployment. The NDT is evaluated on 8-, 16-, and 30-node synthetic topologies, achieving mean absolute percentage error (MAPE) latency predictions around $1\%$ for smaller topologies and about $2\%$ for the largest, demonstrating accurate foresight into network performance when routes are changed. It integrates a physical testbed (ONOS, Mininet), real-time monitoring, and an end-to-end closed loop where the virtual twin informs policy choices that are then enacted in the live network. Two routing strategies are compared: a Random baseline and an AI-based PPO reinforcement learning approach, highlighting the practical utility of the NDT in guiding SLA-compliant decisions and reducing trial-and-error in network management. The work advances transport slicing by enabling proactive, data-driven decisions through what-if analyses, potentially reducing SLA violations and optimizing resource use in 5G/B5G networks.

Abstract

The advent of fifth-generation (5G) and Beyond 5G (B5G) networks introduces diverse service requirements, from ultra-low latency to high bandwidth, demanding dynamic monitoring and advanced solutions to ensure Quality of Service (QoS). The transport network - responsible for interconnecting the radio access network and core networks - will increasingly face challenges in efficiently managing complex traffic patterns. The Network Digital Twin (NDT) concept emerges as a promising solution for testing configurations and algorithms in a virtual network before real-world deployment. In this context, this work designs an experimental platform with NDT in a transport network domain, synchronizing with the virtual counterpart and a recommendation system for what-if analysis, enabling intelligent decision-making for dynamic route optimization problems in 5G/B5G scenarios. Our NDT, composed of a Graph Neural Network (GNN), was evaluated across three different network topologies consisting of 8, 16, and 30 nodes. It achieved lower MAPE values for URLLC and eMBB slices, comparing latency predictions with actual latency after the solution implementation. These values indicate high accuracy, demonstrating the solution's effectiveness in generating precise insights into network performance if a particular solution were implemented.

Figures (5)

• Figure 1: The figure above shows our architecture with NDT for transport networks. The figure's colors are distributed as follows: green for the Control Plane, blue for the virtual twin, purple for the Monitoring System, pink for the recommendation system, and yellow for the Control Functions of the Network Application layer.
• Figure 2: The representation of decision-making with the network operator options (accept or reject the solution) highlight in yellow.
• Figure 3: The suggested routes by the recommendation system, for the AI-based and Random solution, and the GNN latency predictions for each solution, composing the what-if analysis, for the 16-node topology. The house and star symbol represent the source and destination switch, respectively.
• Figure 4: Performance analysis of the URLLC slice in terms of average latency, when applying different routing methods in a 30-node topology.
• Figure 5: Performance analysis of the eMBB slice in terms of average latency, when applying different routing methods in a 30-node topology.