This example has been moved here: https://github.com/scaleoutsystems/examples/tree/main/Casa-IoT-keras

CASA test project

This classsic example of Human Daily Activity Recognition (HDAR) is well suited both as a lightweight test when learning FEDn and developing on FEDn in psedo-distributed mode. A normal high-end laptop or a workstation should be able to sustain at least 15 clients. The example is also useful for general scalability tests in fully distributed mode.

Provide local training and test data

For large data transfer reason we uploaded a data folder in this use case to archive.org. To test this use-case you need to download prepared data that composed 27 apartments (casa's), each apartment data are distributed over 11 clients, using this link: https://archive.org/download/data_20210225/data.zip

• Unzip the file
• Copy the content of the unzipped Archive to the data folder under casa directory

Configuring the tests

We have made it possible to configure a couple of settings to vary the conditions for the training. These configurations are expsosed in the file 'client/settings.yaml':

# Parameters for local training
test_size: 0.25
batch_size: 32
epochs: 3

Creating a compute package

To train a model in FEDn you provide the client code (in 'client') as a tarball (you set the name of the package in 'settings-reducer.yaml'). For convenience, we ship a pre-made package. Whenever you make updates to the client code (such as altering any of the settings in the above mentioned file), you need to re-package the code (as a .tar.gz archive) and copy the updated package to 'packages'. From 'test/casa':

tar -zcvf casa.tar.gz client
cp casa.tar.gz packages/

Creating a seed model

The baseline LSTM is specified in the file 'client/init_model.py'. This script creates an untrained neural network and serialized that to a file, which is uploaded as the seed model for federated training. For convenience we ship a pregenerated seed model in the 'seed/' directory. If you wish to alter the base model, edit 'client/models/casa_model.py' and regenerate the seed file:

Generate the configuration files

We provide the 'generate_clients.sh' bash file to generate all the configuration yaml files (docker_compose.clients.yaml, fedn-network.yal, extra-hosts.yaml) to run casa benckmark in the easiest way.

bash generate_clients.sh 

Configure and start a client using cpu device

The easiest way to start clients for quick testing is to use shell script.The following shell script will configure and start a client on a blank Ubuntu 20.04 LTS VM:

#!/bin/bash

# Install Docker and docker-compose
sudo apt-get update
sudo snap install docker

# clone the nlp_imdb example
git clone https://github.com/scaleoutsystems/FEDn-client-casa-keras.git
cd FEDn-client-casa-keras

# if no available data, download it from archive
# wget https://archive.org/download/data_20210225/data.zip
# sudo apt install unzip
# unzip -o data.zip
# sudo rm data.zip

# Make sure you have edited extra-hosts.yaml to provide hostname mappings for combiners
# Make sure you have edited fedn-network.yaml to provide hostname mappings for reducer
sudo docker-compose -f docker-compose.clients.yaml -f extra-hosts.yaml up --build

License

Apache-2.0 (see LICENSE file for full information).