Section: New Results

Energy in Wireless Sensor Networks

Participants : Emmanuel Nataf [contact] , Patrick-Olivier Kamgueu.

The energy sources of sensors in a wireless network rely mainly on batteries and are very limited in their capacity. Several research efforts are focalized on trying to limit the energy consumption in such networks. This is particularly the case in protocol design. Indeed, the communication consumes a large majority of the available energy. To be realistic and efficient, all proposed approaches need to know the energy available at any time in the systems. Unfortunately, most sensors do not provide such information because it requires additional built-in hardware that would drastically increase their cost. Over the last decade very accurate physical battery models that encompass consumption and recovery have been designed. The complexity of these models is however too high to be implemented inside simple sensors. Recent research results have shown that this integration could be possible if some approximations are integrated in the models.

We have worked on integrating such an approcimated model in the sensor operating system. This work allows the simulation of such sensors and the deployement on real devices that will be aware of their remaining energy level without requiring any additionnal costly equipment. A first implementation on simulation tool has given very promising results; sensors can access their energy level and take decision based on this estimate. Firstly, we have studied energy consumption of a sensors network collecting and routing data toward a single destination. Energy cost of the network deployment has been computed and so the network life as a whole. An other result of our work is the comparison of several common link layer access protocols and several data rate transmits [31] .