Building a Low-cost Network Digital Twin for the IoT-Edge-Cloud Continuum Using Open-Source Tooling
Josevany do Amaral, Rute C. Sofia · 2026
Read Building a Low-cost Network Digital Twin for the IoT-Edge-Cloud Continuum Using Open-Source Tooling by Josevany do Amaral, Rute C. Sofia (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-24853.txt.
Introduction
Josevany do Amaral and Rute Sofia build a low cost digital twin for industrial networks using only open source tools. Testing network changes and failure cases on live Internet of Things infrastructure risks real disruption. A digital twin offers a safe replica for such experiments. Their twin targets edge deployments in the IoT edge cloud continuum. It is assembled from Containerlab, Open vSwitch, the ONOS controller, and a Prometheus and Grafana monitoring stack. Topology emulation, software defined traffic control, and live telemetry sit in one package.
Validation compares the twin against a physical Raspberry Pi edge wireless network. Round trip time medians nearly match, differing by 0.4 milliseconds. UDP throughput differs by only 0.03 megabits per second. Two measures diverge more: TCP throughput and packet loss. Amaral and Sofia trace those gaps to identifiable virtualisation artifacts. For each they give root causes and remediation paths.
Catalog record
| CREATOR | Josevany do Amaral, Rute C. Sofia |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 38 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2606.24853
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Building a Low-cost Network Digital Twin for the IoT-Edge-Cloud Continuum Using Open-Source Tooling Josevany do Amaral, Rute C. Sofia Abstract Validating network configurations and testing failure scenarios in IoT-edge-cloud environments without disrupting live infrastructure remains an open operational challenge. This paper presents a low-cost, fully open-source Network Digital Twin (NDT) for IIoT edge deployments, built on Containerlab, Open vSwitch, ONOS, and a Prometheus/Grafana observability stack. The framework integrates container-native topology emulation, SDN-driven traffic engineering, and real-time telemetry in a single deployable artefact. Validation against a physical Raspberry Pi edge WLAN shows strong distributional convergence on RTT median ( Δ = 0.4 \Delta=0.4 ms) and UDP throughput ( Δ = 0.03 \Delta=0.03 Mbps). Remaining divergences on TCP throughput and packet loss are attributed to identifiable virtualisation artefacts, with root causes and remediation paths provided. I Introduction The IoT-Edge-Cloud continuum spans resource-constrained sensor nodes, wireless access networks, edge compute clusters, and cloud backends, creating multi-tier environments where network behavior is inherently heterogeneous and non-stationary. Validating configurations, stress-testing failure recovery, and verifying Quality-of-Service (QoS) policies in such environments without disrupting production traffic is a fundamental operational challenge. Network Digital Twins (NDTs) address this challenge by providing a virtual replica of the physical network, one that mirrors behavioral characteristics closely enough to serve as a safe sandbox for experimentation. The IETF/IRTF Network Management Research Group defines an NDT as “an advanced platform for network emulation, serving as a tool for scenario planning, impact analysis, and change management” [ 12 ] , characterized by five elements: data , models , mapping , interfaces , and logic . A key property distinguishing an NDT from classical simulation is interactive virtual-real mapping: the twin is calibrated against, and continuously compared with, the physical network [ 12 ] . Despite this clear positioning, practical open-source low-cost NDT implementations that integrate container-native topology emulation, programmable SDN control, and a live observability pipeline in a single deployable framework remain scarce. This paper fills that gap. We present an NDT built progressively from first principles using only open-source components: Containerlab, Open vSwitch, ONOS, Prometheus, and Grafana. The framework is validated against a far Edge small testbed deployed in the fortiss IIoT Labs, a Wi-Fi connected cluster of Raspberry Pi edge nodes representative of an IoT-edge deployment. The progression from basic virtual networking to full Digital Twin validation mirrors the learning arc of the tooling itself, making the framework reusable as both a research platform and a teaching artefact. The specific contributions are: 1. An open-source NDT architecture for the IoT-edge-cloud continuum, mapping directly onto the IRTF reference model [ 12 ] . 2. Four calibrated Wi-Fi impairment profiles (Baseline, Open, Typical, Congested) derived from empirical CODECO measurements and applied via a reusable wifi-profile.sh toolchain. 3. A five-experiment evaluation campaign: baseline characterization, congestion impact, SDN traffic engineering, resilience testing, and NDT-vs-physical fidelity validation. 4. Identification and mitigation of a TBF segment starvation artefact observed in Containerlab/OVS environments when iPerf3 uses default MTU-sized transfers, with discussion of the conditions under which it recurs and its secondary effect on TCP throughput fidelity. 5. A critical analysis of convergence failures and their root causes, with concrete directions for future work. The paper is organized as follows. Section II covers related work. Section III provides background. Section IV presents the NDT architecture. Section V describes the laboratory environment. Section VI defines the experimental methodology. Section VII reports results. Section VIII analyzes findings and limitations. Section IX concludes. II Related Work The concepts of digital twins originating in industrial manufacturing [ 4 ] have been progressively adapted to networking contexts. Early network emulation platforms such as GNS3 and CORE enabled topology virtualization but lacked integration with real-time telemetry pipelines and SDN control planes. Mininet established a standard for SDN experimentation [ 6 ] , but its kernel-space architecture limits container diversity, image flexibility, and scale. Containerlab [ 3 ] addresses these limitations by providing a declarative, container-native topology engine supporting a wide range of network operating systems and lightweight Linux nodes. Unlike Mininet, Containerlab treats the container as the first-class abstraction, allowing each node to carry its own networking stack, monitoring agents, and configuration. Regarding NDT validation methodology, prior work has focused primarily on datacenter scenarios or analytical delay models [ 2 ] . NDT frameworks targeting IIoT or wireless edge environments are substantially less common, and those that exist typically do not incorporate SDN control-plane testing or long-term behavioral distribution matching. Closest to this is the work of [ 10 ] , which uses Mininet for IIoT scenario emulation but does not address Wi-Fi profile calibration, persistent telemetry, or SDN-driven traffic engineering within the twin. The use of tc netem for wireless impairment injection is established practice [ 9 ] , but the interaction between netem queue disciplines and Linux Token Bucket Filter (TBF) shapers in containerized environments, particularly the segment starvation artifact described in Section IV , has not been systematically characterized in prior work. III Background III-A The IoT-Edge-Cloud Continuum The IoT-Edge-Cloud continuum integrates varied cyber-physical systems and networks. It comprises far Edge devices/clusters and near Edge devices and clusters, spanning to the Cloud. Network paths traverse multiple administrative domains and integrate varied networking technology, e.g., Wi-Fi, Ethernet, and cellular, among others. Traffic engineering is applied at each tier. This heterogeneity makes the continuum particularly difficult to model: a policy change at the edge can have cascading effects on cloud-bound telemetry streams. Container orchestration platforms such as Kubernetes are increasingly deployed at the Edge [ 8 ] to manage workloads across this continuum, but network-layer validation tooling has not kept pace. More recently, extensions to container orchestration addressing cross-layer context-awareness, such as CODECO [ 11 ] , are bridging the divide between communication and computation, aiming at providing a holistic view of what infrastructures for the IoT-Edge-Cloud continuum need to support. III-B Network Digital Twins The IRTF NDT reference architecture [ 12 ] defines five functional elements. Data captures real-time and historical state from the physical network. Models provide emulation abstractions over that data. Mapping establishes the virtual-real correspondence, which may be one-to-one (continuous sync) or one-to-many (federated twins). Interfaces standardize integration between the twin and external systems. Logic encodes analysis, diagnosis, and control functions. The draft also enumerates five construction challenges: large-scale data management, interoperability across heterogeneous devices, data modelling complexity, real-time synchronization, and security. At an IIoT-Edge scale, the first three are most urgent and are directly addressed by our framework (Section IV ). III-C Container-Native Network Emulation: Containerlab Containerlab [ 3 ] is a declarative topology engine that provisions container-based network nodes such as routers, switches, and endpoints, hosts, connected by virtual links, all described in a single YAML manifest. Each node is a standard Docker container, giving it a full Linux networking stack and the ability to run arbitrary monitoring agents. Topology deployment and teardown take seconds, enabling reproducible experimental cycles. Compared to Mininet [ 6 ] , Containerlab supports a broader range of node images, integrates naturally with existing container infrastructure (including Kubernetes worker nodes), and does not require kernel patches. A known limitation in Containerlab/OVS environments is the interaction between tc netem and the Linux Token Bucket Filter (TBF) shaper: standard MTU-sized transfers can exhaust the TBF token