Improving Robustness and Reducing Control Overhead via Dynamic Clustering in Tactical SDN

This study presents a dynamic clustering approach to address the challenges of managing modern and dynamic tactical communication using diverse radio links in Software-Defined Tactical Networks (SDTNs). State-of-the-art conventional Software-Defined Networking (SDN) protocols, which rely on centralized controllers, are unsuitable for tactical scenarios due to their inherent demands for high-throughput, low-latency control planes, and static nodes. To enable SDTN, we propose a clustering solution designed to distribute network nodes across multiple remote controllers, thereby reducing the control message overhead created by the OpenFlow protocol. Moreover, we propose a flexible controller assignment strategy to dynamically balance the network load during runtime and improve network robustness using backup controllers. Our comparative analysis suggests a significant reduction of control overhead across heterogeneous topologies with VHF and UHF links using our clustering solution. Furthermore, we replicate a real-world scenario by emulating a convoy of military vehicles demonstrating the effective distribution of low control overhead among the controllers and maintaining robust network connectivity, even in the event of a controller failure.

Please cite:

@INPROCEEDINGS{zibner2023improving,
  author={Zißner, Philipp and Rettore, Paulo H. L. and Santos, Bruno P. and Lopes, Roberto Rigolin F. and Loevenich, Johannes F. and Sevenich, Peter},
  booktitle={MILCOM 2023 - 2023 IEEE Military Communications Conference (MILCOM)}, 
  title={Improving Robustness and Reducing Control Overhead via Dynamic Clustering in Tactical SDN}, 
  year={2023},
  volume={},
  number={},
  pages={491-496},
  keywords={Protocols;Runtime;Network topology;UHF measurements;Radio links;Robustness;Topology;Software-defined Tactical Network;Control Overhead;Clustering},
  doi={10.1109/MILCOM58377.2023.10356288}
}

Philipp Zißner; Paulo H. L. Rettore; Bruno P. Santos; Roberto Rigolin F. Lopes; Johannes F. Loevenich; Peter Sevenich

Founding agencies: CNPq/CAPES/FAPEMIG.