NetFL

NetFL is a framework that extends Fogbed by integrating Flower, enabling simulation of Federated Learning experiments within Fog/Edge computing environments. It supports the modeling of heterogeneous and resource-constrained edge scenarios, incorporating factors such as computational disparities among clients and dynamic network conditions, including bandwidth limitations, latency variations, and packet loss. This facilitates realistic evaluations of FL systems under non-ideal, real-world conditions.

Installation

Requirements: Ubuntu 22.04 LTS or later, Python 3.9.

1. Set up Containernet

Refer to the Containernet documentation for further details.

Install Ansible:

sudo apt-get install ansible

Clone the Containernet repository:

git clone https://github.com/containernet/containernet.git

Run the installation playbook:

sudo ansible-playbook -i "localhost," -c local containernet/ansible/install.yml

Create and activate a virtual environment:

python3 -m venv venv
source venv/bin/activate

Note: The virtual environment must be activated before installing or using any Python packages, including Containernet and NetFL.

Install Containernet into the active virtual environment:

pip install containernet/.

2. Install NetFL

While the virtual environment is still active, run:

pip install netfl

Running an Experiment with NetFL and Fogbed

Follow the steps below to set up and run an experiment using NetFL. This is an example using the MNIST dataset. You can find more examples in the examples folder:

1. Define the Dataset, the Model, and the Training Configurations

from keras import layers, models
from flwr.server.strategy import Strategy, FedAvg

from netfl.core.task import Dataset, Task, TrainConfig, DatasetInfo


class MNIST(Task):
    def dataset_info(self) -> DatasetInfo:
        return DatasetInfo(
            huggingface_path="ylecun/mnist",
            item_name="image",
            label_name="label",
        )

    def dataset(self, raw_dataset: Dataset) -> Dataset:
        normalized_dataset = Dataset(
            x_train=(raw_dataset.x_train / 255.0),
            x_test=(raw_dataset.x_test / 255.0),
            y_train=raw_dataset.y_train,
            y_test=raw_dataset.y_test,
        )
        return normalized_dataset

    def model(self) -> models.Model:
        model = models.Sequential([
            layers.Input(shape=(28, 28)),
            layers.Flatten(),
            layers.Dense(128, activation="relu"),
            layers.Dense(10, activation="softmax")
        ])
        model.compile(
            optimizer="adam",
            loss="sparse_categorical_crossentropy",
            metrics=["accuracy"],
        )
        return model

    def aggregation_strategy(self) -> Strategy:
        return self._aggregation_strategy_factory(FedAvg)
    
    def train_config(self) -> TrainConfig:
	    return TrainConfig(
            batch_size=32,
            epochs=1,
            fraction_evaluate=1.0,
            fraction_fit=1.0,
            learning_rate=0.001,
            min_available=4,
            max_available=4,
            num_rounds=10,
            seed=42,
            shuffle=True,
            test_size=0.2,
        )


class MainTask(MNIST):
    pass

2. Start Fogbed Workers and Define the Experiment Network Topology

Refer to the Fogbed documentation for detailed instructions on starting workers.

3. Create and Run the Experiment

from fogbed import CloudResourceModel, EdgeResourceModel, HardwareResources
from netfl.infra.experiment import Experiment
from task import MainTask

exp = Experiment(
    main_task=MainTask(),
    max_cpu=2.0,
    max_memory=3072,
)

worker = exp.add_worker(ip="192.168.0.100", port=5000)

cloud = exp.add_virtual_instance(
    name="cloud",
    resource_model=CloudResourceModel(max_cu=1.0, max_mu=1024)
)

edge_0 = exp.add_virtual_instance(
    name="edge_0",
    resource_model=EdgeResourceModel(max_cu=0.5, max_mu=1024)
)

edge_1 = exp.add_virtual_instance(
    name="edge_1",
    resource_model=EdgeResourceModel(max_cu=0.5, max_mu=1024)
)

server = exp.create_server(
    resources=HardwareResources(cu=1.0,  mu=1024),
    link_params={"bw": 1000, "delay": "2ms"},
)

edge_0_devices = [ 
    exp.create_device(
        resources=HardwareResources(cu=0.25,  mu=512),
        link_params={"bw": 100, "delay": "10ms"},
    ) for _ in range(2)
]

edge_1_devices = [ 
    exp.create_device(
        resources=HardwareResources(cu=0.25,  mu=512),
        link_params={"bw": 50, "delay": "5ms"},
    ) for _ in range(2)
]

exp.add_docker(server, cloud)

exp.add_docker(edge_0_devices[0], edge_0)
exp.add_docker(edge_0_devices[1], edge_0)

exp.add_docker(edge_1_devices[0], edge_1)
exp.add_docker(edge_1_devices[1], edge_1)

worker.add(cloud)
worker.add(edge_0)
worker.add(edge_1)

worker.add_link(
    cloud, 
    edge_0, 
    bw=10, delay="100ms", loss=1, max_queue_size=100, use_htb=True,
)

worker.add_link(
    cloud, 
    edge_1, 
    bw=5, delay="50ms", loss=1, max_queue_size=100, use_htb=True,
)

try:
    exp.start()    
    print("The experiment is running...")
    input("Press enter to finish")
except Exception as ex: 
    print(ex)
finally:
    exp.stop()

Running a Simple Example with a Basic Network Topology Using Docker

1. Create the Main Task

In the project root directory, create or modify a NetFL Task and name the file task.py. Refer to the examples in the examples folder for guidance on task creation.

2. Create the Infrastructure

Use Docker Compose to set up the infrastructure, including the server and clients:

docker compose up -d

3. View Training Results

To check the server logs, run:

docker logs server

Training logs are also stored in the logs folder within the project root directory.

4. Shut Down the Infrastructure

To stop and remove all running containers, use the following command:

docker compose down

More information

• NetFL on PyPI

• NetFL Docker Images