Skip to content
/ rescue Public

A distributed super-resolution architecture for traffic matrix prediction with partial network traffic visibility.

Notifications You must be signed in to change notification settings

lorepap/rescue

Repository files navigation

ResCue

This is the code repository of our CNSM paper “Rescue: Inferring fine-grained traffic matrices via distributed deep residual networks".

Configuration

Code is tested with Python=3.8.13. Install requirements with

pip -r install requirements/requirements.txt

Configure training settings for federated learning by editing config/federated.yml.

Running

To run the ResCue code use the runner within the runners folder:

python runners/federated_runner.py

You can choose between multi-client or single-client trainings. When multi-client is set to true, multiple sequential threads are thrown setting an increasing number of clients - from client_range[0] to client_range[1] (see federated.yml). When multi-client is false, a thread will be thrown training a single federated model with the number of clients set as n_clients.

Plotting and results evaluation

Before plotting, launch the evaluation script to compute the metrics obtained by the inference process.

About

A distributed super-resolution architecture for traffic matrix prediction with partial network traffic visibility.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published