DTN-FL

This repository contains the single-region DTN-FL implementation used for blackhole detection in a Delay-Tolerant Networking environment. The code focuses on node-level federated learning, reactive blackhole classification, and pre-emptive canary-probe based detection.

Repository layout

notebooks/   Main experiments and model evaluation notebooks
testbed/     Single-region IBR-DTN/Mininet testbed helpers
src/         Reusable Python modules for preprocessing, models, training, and FL flow
data/        Dataset notes and expected input format
results/     Output folder for metrics, tables, and figures

Main notebooks

Notebook	Purpose
dtn_fl_single_region_fedavg.ipynb	Single-region federated learning experiment
reactive_blackhole_classification.ipynb	Reactive blackhole classification baseline
preemptive_canary_probe_detection.ipynb	Canary-probe based pre-emptive detection

Setup

Create a Python environment and install the required packages:

pip install -r requirements.txt

For Mininet/IBR-DTN experiments, run the testbed scripts on a Linux machine with Mininet and IBR-DTN installed.

Notes

Large datasets and generated outputs are not included by default. Place local datasets inside data/ and save generated figures or result files inside results/.