D. Amodei and D. Hernandez, Ai and compute, 2018.

K. Anderson, The inconvenient truth of carbon offsets, Nature, vol.484, issue.7392, pp.7-7, 2012.

L. Anthony, B. Kanding, and R. Selvan, Carbontracker: Tracking and predicting the carbon footprint of training deep learning models, 2020.

, AWS. Aws and sustainability, 2020.

R. F. Berriel, A. T. Lopes, A. Rodrigues, F. M. Varejo, and T. Oliveira-santos, Monthly energy consumption forecast: A deep learning approach, 2017 International Joint Conference on Neural Networks (IJCNN), pp.4283-4290, 2017.

D. J. Beutel, T. Topal, A. Mathur, X. Qiu, T. Parcollet et al., Flower: A friendly federated learning research framework, 2020.

M. Hodak, M. Gorkovenko, and A. Dholakia, Towards power efficiency in deep learning on data center hardware, 2019 IEEE International Conference on Big Data (Big Data), pp.1814-1820, 2019.

M. Hodak, M. Gorkovenko, and A. Dholakia, Towards power efficiency in deep learning on data center hardware, 2019 IEEE International Conference on Big Data (Big Data), pp.1814-1820, 2019.

K. Hsieh, A. Phanishayee, O. Mutlu, and P. Gibbons, The non-iid data quagmire of decentralized machine learning, 2019.

, Climate change 2014 synthesis report, 2014.

J. Leonard, E. J. Kish, and . Topol, Unpatientswhy patients should own their medical data, Nature biotechnology, vol.33, issue.9, pp.921-924, 2015.

A. Krizhevsky and G. Hinton, Learning multiple layers of features from tiny images, 2009.

A. Lacoste, A. Luccioni, V. Schmidt, and T. Dandres, Quantifying the carbon emissions of machine learning, 2019.

K. Lepak, G. Talbot, S. White, N. Beck, and S. Naffziger, The next generation amd enterprise server product architecture, IEEE hot chips, vol.29, 2017.

T. Li, A. Kumar-sahu, M. Zaheer, M. Sanjabi, A. Talwalkar et al., Federated optimization in heterogeneous networks, 2018.

X. Li, K. Huang, W. Yang, S. Wang, and Z. Zhang, On the convergence of fedavg on non-iid data, 2019.

W. Lim, C. Nguyen, D. Luong, Y. Hoang, Y. Jiao et al., Federated learning in mobile edge networks: A comprehensive survey, IEEE Communications Surveys & Tutorials, 2020.

B. Mcmahan and D. Ramage, Federated learning: Collaborative machine learning without centralized training data, Google Research Blog, vol.3, 2017.

B. Mcmahan, E. Moore, D. Ramage, S. Hampson, and B. Aguera-y-arcas, Communication-Efficient Learning of Deep Networks from Decentralized Data, Proceedings of Machine Learning Research, vol.54, pp.20-22, 2017.

. Microsoft, Datacenter fact sheet -microsoft download center

, How to stop data centres from gobbling up the worlds electricity, Nature, 2018.

T. Nishio and R. Yonetani, Client selection for federated learning with heterogeneous resources in mobile edge, ICC 2019 -2019 IEEE International Conference on Communications (ICC), pp.1-7, 2019.

. Rajendra-k-pachauri, K. Gomez-echeverri, and . Riahi, Synthesis report: summary for policy makers, 2014.

R. Pryss, M. Reichert, J. Herrmann, B. Langguth, and W. Schlee, Mobile crowd sensing in clinical and psychological trials -a case study, 2015 IEEE 28th International Symposium on Computer-Based Medical Systems, pp.23-24, 2015.

J. Qian, X. Fafoutis, and L. K. Hansen, Towards federated learning: Robustness analytics to data heterogeneity, 2020.

O. Russakovsky, J. Deng, H. Su, J. Krause, S. Satheesh et al., Imagenet large scale visual recognition challenge, International journal of computer vision, vol.115, issue.3, pp.211-252, 2015.

H. Shen, Smartphones set to boost large-scale health studies, Nature News, 2015.

R. Smith, Nvidia announces jetson tx2: Parket comes to nvidia's embedded system kit, 2017.

E. Strubell, A. Ganesh, and A. Mccallum, Energy and policy considerations for deep learning in nlp, 2019.

A. Uchechukwu, K. Li, and Y. Shen, Energy consumption in cloud computing data centers, International Journal of Cloud Computing and Services Science (IJ-CLOSER), vol.3, issue.3, pp.145-162, 2014.

. Uptimeinstitue, data center industry survey results, 2019.

, IID and non-IID terms are employed to distinguish between clients that have an evenly distributed set of samples containing all the classes (IID) and clients that have more samples of certain classes (non-IID). The "Max Acc, Table 5: Details of the results obtained on CIFAR10 with multiple FL setups