Energy Efficiency in Cyber-Physical Systems

The DC will study the use of RC for programming low-energy cyber-physical systems, such as fixed-wing aerial robots or IoT devices, where a large proportion of the energy consumption can be from computation rather than actuation or communication. Reversible programming languages will be used to implement relevant algorithms, and they will be integrated into a simulated implementation of each cyber-physical system. Algorithms are expected to include both fully and partially reversible algorithms, for example compression of data collected from sensors and image analysis algorithms. Energy consumption of the cyber-physical system as a whole will be estimated by a dual approach combining energy estimates of the RC hardware and of the cyber-physical system as a whole. A key research question lies in determining the extent to which energy savings depend on the I/O properties of the system, e.g., lossless compression of sensor data in an IoT device versus extracting information from an input image for control of an aerial robot.

1) Identification of energy-critical fully or partially reversible algorithms relevant for use in low-energy cyber-physical systems; 2) Modelling framework for estimating the energy impact of reversibility on cyber-physical systems; 3) Proof of concept implementation of aerial robot and IoT systems using reversible algorithms.

M15, UNIBO, I. Lanese, partially reversible algorithms; M32, UoM, M. Lujan, modelling framework.