NNSI — Network Node Significance Index

A composite centrality metric for identifying critical nodes and optimizing topology in Software-Defined Networks (SDN). Evaluated on 260 of 261 real-world networks from the Internet Topology Zoo.

📄 Reference paper: A Framework for Optimizing Network Topology Based on Graph Theory in Software-Defined Networking — Springer ICOMP'25 (citation below) 📊 Full 261-network run log: experiments/full_run_log.txt

Quick demo (2 minutes)

git clone https://github.com/DemigodDSK/nnsi-network-optimization.git
cd nnsi-network-optimization
python3 -m venv .venv && source .venv/bin/activate
pip install -e .
python -m nnsi --demo

The demo runs on 5 sample networks (Forthnet, UsSignal, Geant2012, Kdl, Esnet) and prints a comparison table.

Headline results (260 networks)

Metric	Value
Networks successfully analyzed	260 of 261
NNSI outperforms IVI baseline on	52 networks (20.0%)
Average NNSI advantage (impact difference)	0.0811
Top-node overlap with IVI (avg)	90.2%
Avg latency reduction post-optimization	5.86%
Avg resilience improvement	10.64%
Avg link-utilization improvement	14.59%
Full-run wallclock	4151s (~69 min)

NNSI formulation

NNSI(v) = CS(v) × FCS(v) × IPS(v)

• CS — Centrality Score (composite of degree, betweenness, closeness, eigenvector, PageRank)
• FCS — Failure Cascade Score (collective influence + clusterrank + local h-index)
• IPS — Information Propagation Score (neighborhood connectivity + ego-network density)

The product structure penalizes nodes that score high on only one dimension, surfacing nodes critical along multiple network axes. Compared against the IVI (Integrated Value of Influence) baseline.

Reproduce the full 261-network run

# Topology Zoo: http://www.topology-zoo.org/dataset.html (CC BY 4.0)
python -m nnsi --data-dir /path/to/your/topology_zoo

Expect ~70 min on a single CPU (the bundled experiments/full_run_log.txt records the original run).

Project layout

nnsi-network-optimization/
├── nnsi/
│   ├── __init__.py
│   ├── __main__.py              # CLI entry — python -m nnsi --demo
│   └── analyzer.py              # NetworkAnalyzer class (centrality, NNSI, IVI, optimization)
├── data/topology_zoo_sample/    # 5 GraphML files for the demo
├── experiments/full_run_log.txt # Original 261-network run output
├── pyproject.toml
└── requirements.txt

Stack

Python 3.11+ · NetworkX · NumPy · SciPy · pandas · matplotlib · seaborn · tqdm

Citation

If you use this work, please cite:

@inproceedings{naidu2025nnsi,
  title     = {A Framework for Optimizing Network Topology Based on Graph Theory in Software-Defined Networking},
  author    = {Naidu, Datta Sai Krishna},
  booktitle = {Proceedings of the International Conference on Computational Methods (ICOMP'25)},
  publisher = {Springer},
  year      = {2025}
}

Data attribution

Topology data is from the Internet Topology Zoo project (Knight et al.), licensed CC BY 4.0. See http://www.topology-zoo.org/

---

MIT License · © 2026 Datta Sai Krishna Naidu