Distribution Grid Emulation incl. Topology

Description

This repository contains the code from our study on evaluating transport protocols, particularly MPTCP, in the context of reliable communication within energy distribution grids.

@inproceedings{2025_fink_mptcp-eval,
    author = {Fink, Ina Berenice and Ferlemann, Lennart and Dahlmanns, Markus and Thimm, Christian and Wehrle, Klaus},
    title = {{Emulating and Evaluating Transport Layer Protocols for Resilient Communication in Smart Grids}},
    booktitle = {Proceedings of the 2025 IEEE/IFIP Network Operations and Management Symposium (NOMS '25), May 12-16, 2025, Honolulu, HI, USA},
    year = {2025},
    publisher = {IEEE},
}

Simulation setup

Install rettij from https://gitlab.com/frihsb/rettij

Enter our project directory.

Generate a custom topology

python3 topology-generator/generator.py <topology.yaml> <topology-ips.txt>

<topology.yaml> and <topology-ips.txt> are the output files.

Starting rettij

1. Run the following command, where <topology.yaml> is a topology that has been generated with the topology generator. The original topology from our evaluation is provided by topologies/distribution-system-topo.yaml. Note that the setup can take about an hour for large topologies.

   rettij -t <topology.yaml> --components custom-components/

2. Set up the rettij interfaces, the necessary commands will be output by rettij while setting up the simulation. It might be necessary to delete old interfaces from a preceding simulation run using the displayed commands beforehand.

3. If desired, connect the rettij interfaces for control center computer (rettij.pc) and RTU (rettij.rtu) to the external devices using network bridges. For example for the bridges br0 and br1:

   sudo ip addr del 10.100.101.2/24 dev rettij.pc
   sudo ip addr del 10.100.102.2/24 dev rettij.rtu
   sudo ip link set rettij.pc master br0
   sudo ip link set rettij.rtu master br1

Note that we applied the following changes to k3s to successfully run rettij with our topology.

Modifications to Kubernetes (k3s)

• The root directory for kubelet has to be changed, because the root file system was too full, and an additional partition was created. This change was done in the file /etc/rancher/k3s/config.yaml.

  kubelet-arg:
  - "root-dir=/data/kubelet"

• The maximum number of pods had to be raised because of the size of our topology. This change was done in the file /etc/rancher/k3s/config.yaml and the file /etc/rancher/k3s/kubelet.yaml had to be added.

  kubelet-arg:
  - "config=/etc/rancher/k3s/kubelet.yaml"

  apiVersion: kubelet.config.k8s.io/v1beta1
  kind: KubeletConfiguration
  maxPods: 1000

• The size of the default network for the pods had to be increased. This change was done in the files /etc/rancher/k3s/config.yaml and /etc/rancher/k3s/kubelet.yaml respectively.

  cluster-cidr: "10.44.0.0/16"

  podCIDR: "10.44.0.0/16"

• The Kubernetes node needs to be exported, modified, deleted and re-created, to reflect the changed network.

  kubectl get node <node> -o yaml ><node.yaml>
  # adjust podCIDR and podCIDRs
  vim <node.yaml>
  kubectl delete node <node>
  kubectl create -f <node.yaml>