Modelling, quantification, optimisation - energy efficiency in distributed, embedded systems

Abstract

During the last decade, energy efficiency became one of the most challenging research fields in the IT domain. After optimising large-scaled hardware infrastructures, e.g. computing centres or server clusters, nowadays the focus has changed rapidly. Current R&D projects deal with two major areas. On the one hand embedded, energy self-sufficient system architectures and on the other hand wireless communication technologies to enable mobile operation. Due to strongly limited energy resources within these platforms, the system runtime represents a critical parameter. Especially in case of cooperative scenarios with multiple systems in a distributed environment, efficient communication features and the availability of all systems are essential to ensure a robust operation. Accordingly, optimisation strategies are necessary to prolong the runtime and to improve the communication behaviour. In this paper, we critically discuss energy efficiency in distributed, embedded systems with focus on the communication aspects. We introduce an easy to use estimation model as well as the respective cost functions for quantifying the energy efficiency. In this context, we evaluate several optimisation strategies on different abstraction layers regarding their energy efficiency and their impact on the communication behaviour. We analyse wake-up receiver technologies as well as a modern data aggregation approach - PLANetary. The implemented real-world test scenario represents a wireless network topology with specific, wake-up receiver enabled hardware platforms. The results clarify, that if the configuration scheme fits to the environmental conditions, significant improvements of the energy efficiency are possible.