Energy-efficient computing

Computing systems, from embedded sensors to data centers, are limited by energy consumption. Our research focuses on designing and building computer systems that manage and minimize energy consumption. We develop algorithms and methods for resource-constrained intelligent devices, optimizing at different levels. This includes using compressed sensing and event-based sampling for data acquisition and exploiting modern edge/fog/cloud architectures for efficient machine-learning workload distribution during training and execution.

We study optimizations at the system level to quantify adequately the impact on the energy efficiency of strategies, like moving the data processing closer to its source, to reduce communication during data transmission.


Federated machine learning over fog/edge/cloud architectures