Network-wide energy optimization for adaptive embedded systems

This paper discusses network-wide energy optimization of embedded systems which can adapt by switching configurations. We model applications and their task chains in a network of embedded devices, including sleep modes and change of configuration, which provides a basic adaptation mechanism. We present a formal model of the energy consumption for such systems and apply it to automotive embedded systems. In particular, we develop new potential for network-wide energy savings as well as optimizations for adaptive systems. For instance, we show that non-optimal configurations may lead to a globally optimal system setup, if a system adapts regularly.