Project summary and main technological topics
The project aims to develop a scalable and sustainable Federated Learning (FL) infrastructure for IoT applications in green scenarios, promoting greener, safer, and more sustainable production processes. The main objectives include:
- Identification of use cases through literature analysis and industrial contexts.
- Development of a federated infrastructure that supports efficient management of IoT devices.
- Implementation of energy-efficient algorithms, with advanced techniques for privacy preservation.
- Creation of an intuitive and accessible user interface.
Main technological topics:
- Federated Learning: design and use of a unified framework to implement FL in IoT contexts, ensuring sustainability and scalability.
- IoT (Internet of Things): management of interconnected devices for applications in green scenarios.
- Privacy-preserving Techniques: advanced techniques to protect sensitive data during federated learning.
- Energy-efficient Algorithms: development of optimized solutions to reduce environmental impact.
- Open Source Frameworks: adoption of open-source technologies to ensure accessibility and economic sustainability.
Dataset and validation:
The project will use two main datasets for the development and validation of algorithms:
- Dataset of household electrical load measurements: data collected in 20 homes over two years, at 8-second intervals, including measurements of aggregate loads and 9 appliances.
- Dataset of hourly load profiles: annual data from 24 facilities representing industrial, commercial, and residential sectors.
Expected innovations:
The proposed infrastructure aims to optimize IoT management in a scalable and secure way, leveraging advanced algorithms and sustainable approaches. The integration of cutting-edge technologies will support the evolution towards greater sustainability in production contexts.
