This repository contains the code and notebooks used to generate FaaS-edge node networks, prepare node-level datasets, and run training or Gossip Learning experiments for edge resource usage forecasting.
The recommended workflow is:
- Generate a FaaS-edge node network with
src/1_network_generation.ipynb. - Prepare node datasets with
src/functions_data_preparation.py. - Run experiments with either:
src/train.pyfor training-based experiments.src/traingossip.pyfor Gossip Learning experiments.
Each step depends on the previous one, so the network should be generated first, then the datasets prepared, and finally the desired experiment launched.
The project is organized around Python-based simulation and experiment
scripts, so you should use a Python environment with the dependencies
required by the repository (see src/requirements.txt).