ENERGY AWARE NETWORKING PRINCIPLES

Formal Approaches to Model-Driven Quality of Service
Energy Aware Networking Principles
Validation of Multiprocessor Multithreaded Architectures
Quantitative Model Checking of Software

Energy Aware Networking Principles

Real-time, reactive and embedded systems are permeating society. At the same time, mobile and ubiquitous applications that were unthinkable only a few years ago are emerging. Many such applications are long-lived, continuously interacting with their environment, but operating under strict resource constraints. One particular concern for battery-powered devices is energy efficiency. Sensor networking protocols such as ZigBee are setting a trend there by relinquishing high data rates in favour of battery lifetime. This puts an entirely new perspective on many traditional networking solutions, where bits-per-seconds is replaced by bits-per-Joule. With his group, Holger Hermanns has lately started investigating truly classical networking principles, such as "slotted CSMA performs about two times better than unslotted CSMA" in light of this new perspective, with noticeably different answers.

 

Recent achievements of the Dependable Systems and Software chair in this context are:


  • Does Clock Precision Influence ZigBee's Energy Consumptions? Wireless embedded sensor networks are predicted to provide attractive application possibilities in industry as well as at home. IEEE 802.15.4 and ZigBee are proposed as standards for such networks with a particular focus on pairing reliability with energy efficiency, while sacrificing high data rates. IEEE 802.15.4 is configurable in many aspects, including the synchronicity of the communication, and the periodicity in which battery-powered sensors need to wake up to communicate. We are developing formal behavioural models for the energy implications of these options. The models are modularly specified (see here) using the language MoDeST, which has an operational semantics mapping on stochastic timed automata. The latter are simulated using a variant of discrete-event simulation implemented in the tool M÷bius. We managed to obtain estimated energy consumptions of a number of possible communication scenarios in accordance with the standards, and derive very interesting conclusions about the energy optimal configuration of such networks. As a specific fine point, we investigate the effects of drifting clocks on the energy behavior of various application scenarios. This is an ongoing joint work with Christian Gro▀ and Reza Pulungan. We are very enthusiastic about it. A paper reporting our results appeared in OPODIS 2007, Springer LNCS.