You are here

Distributed Learning Intelligence at the Tiny Edge

Towards Distributed Embedded Sustainable Computing
We are interested in studying the design of federated learning for distributed low capacity computing devices. These devices include mini-PCs, Single-Board-Computers and tiny computing devices (embedded systems). The goal is to find more adaptive and decentralized federated learning designs.

Our approach is very practical. We use real devices in testbeds to experiment different designs of the federated learning protocol and their effects on resource consumption and accuracy.

Projects with code:
Adaptive federated learning with mini-PCs.
Federated learning with the Arduino Nano 33 BLE Sense.
Federated learning with the Arduino Portenta H7.
Dual-core usage in the Portenta H7 for machine learning.

Publications we made on this topic:
1. Marc Monfort Grau, Roger Pueyo Centelles, and Felix Freitag. On-Device Training of Machine Learning Models on Microcontrollers With a Look at Federated Learning. In Proceedings of the Conference on Information Technology for Social Good (GoodIT '21). Association for Computing Machinery, New York, NY, USA, 198–203. (postprint)
2. F. Freitag, P. Vilchez, Ch. Liu, L. Wei, M. Selimi. Testbed in Wireless City Mesh Network with Application to Federated Learning Experiments. ACM/SIGCHI International Conference on the Internet of Things (IoT 2021), Nov 2021, St.Gallen, Switzerland.
3. L. Ibraimi, M. Selimi, F. Freitag. BePOCH: Improving Federated Learning Performance in Resource-Constrained Computing Devices. IEEE Global Communications Conference (GLOBECOM 2021), Dec 2021, Madrid, Spain.
4. F. Freitag, P. Vilchez, L. Wei, C.H, Liu, M. Selimi, I. Koutsopoulos: Demo: An Experimental Environment Based On Mini-PCs For Federated Learning Research. CCNC 2022: 927-928
5. F. Freitag, P. Vilchez, Ch. Liu, L. Wei, M. Selimi. Performance Evaluation of Federated Learning over Wireless Mesh Networks with Low-Capacity Devices. International Conference on Information Technology & Systems (ICITS 2022), Feb 2022, San Carlos, Costa Rica.
6. Llisterri Giménez, N.; Monfort Grau, M.; Pueyo Centelles, R.; Freitag, F. On-Device Training of Machine Learning Models on Microcontrollers with Federated Learning. Electronics 2022, 11, 573. https://doi.org/10.3390/electronics11040573
7. N. Llisterri Giménez, Freitag, F., Lee, J. K., and Vandierendonck, H., “Comparison of Two Microcontroller Boards for On-Device Model Training in a Keyword Spotting Task”, in 2022 11th Mediterranean Conference on Embedded Computing (MECO), 2022, pp. 1-4.
8. N. Llisterri Giménez, Solé, J. Miquel, and Freitag, F., “Embedded federated learning over a LoRa mesh network”, Pervasive and Mobile Computing, vol. 93, p. 101819, 2023. https://doi.org/10.1016/j.pmcj.2023.101819
9. N. Llisterri Gimenez, Lee, J. K., Freitag, F., and Vandierendonck, H., “The Effects of Weight Quantization on Online Federated Learning for the IoT: A Case Study”, IEEE Access, vol. 12, pp. 5490-5502, 2024. https://doi.org/10.1109/ACCESS.2024.3349557