Lifetime estimation method for connected devices
Connected devices are embedded electronic systems that are powered by batteries. Their lifetime is an important constraint to take into account at very early design stages. In the context of this work, the lifetime is defined as being the time elapsed from the device very first boot until its battery does not contain enough energy to supply it. Thus, in this work, we investigate lifetime estimation in early design stages.
A deep state-of-the-art's study showed that existing solutions were not accurate enough to do such estimation. Therefore, we introduce an original method oriented toward lifetime estimation in order to address this limitation. Our method relies on three fundamental aspects that are the software running on the device, its hardware platform and the battery that supplies it. This method has been implemented as a framework for the OMNeT++ network simulator.
We evaluate the simulation performances of our implementation as well as the precision of our method. Moreover, in the context of a scientific collaboration with the SICS (Sweden), this method has been applied to a case study that implies several TMote Sky nodes running ContikiOS with a RPL/6LowPan communication's stack. Actually, our method gave us access to useful insights. These latter were used to improve the lifetime from 4 months to more than a year.
We also compare our method’s simulation results to real experiment's measures. The absolute mean error is going from 6.17% to 11.56% depending on the model. In comparison, the state-of-the-art methods/solutions lead to an error that is greater than 35% - which highlights their inaccuracies. We also need to mention that the models that we have designed in our method were built using only the technical specifications and do not require any calibration.
Read more about Wilfried Dron here.