AMBIENT-6G D4.1: Initial report on energy-aware computing and applications

This deliverable reports on the state-of-the art and initial AMBIENT-6G designs for energy-aware computing for Ambient-IoT (A-IoT). It starts with an energy analysis of the use cases defined in D5.1 giving a basis for the requirements, opportunities and boundary conditions that must be taken into account when developing of energy-aware applications. It consolidates early results in end-to-end energy modeling and prediction: starting with the radio frequency (RF) wireless power transfer (WPT) channel, RF energy harvesting and storage, continuing with sensors and actuators and energy-aware task execution on the microcontroller, and methods for energy prediction based on time series, energy pattern identification and load balancing. The deliverable further explores energy-aware computing on the level of the energy-neutral device (END) examining energy-aware devices operations and on-device inference and learning. The deliverable concludes with advanced strategies for energy-aware computing over the device-edge-cloud continuum. It investigates trade-offs in offloading computationally intensivetasks, distributed and hybrid approaches for tiny machine learning (TinyML), and goal-oriented, energy-aware protocols tailored for dynamic and dependable edge node deployments. It reports on innovations in ultra-low power federated learning and orchestration mechanisms that enable scalable, intelligent A-IoT systems. Overall, this deliverable lays the foundation for further research in energy-aware computing in A-IoT environments, which will be reported in upcoming deliverables D4.2 and D4.3.