To date, evaluations of the energy efficiency of protocols and applications for small wireless devices have considered only the total energy utilization and ignored the dynamics of the battery itself.
This is true both for simulators and for platforms that measure energy consumption in real time. The battery is treated as a simple “bucket of mA-h,” which is depleted as various operations consume a given amount of current for a given amount of time.
In reality, a battery is a complex, non-linear electrochemical system. The way that charge is extracted from the battery affects how much of the nominal capacity can be used before the battery is depleted. Key effects include:
1) rate capacity – A lower current generally extracts charge from the battery more efficiently. That is, reducing the current by half may more than double the lifetime.
2) relaxation – An intermittent discharge is generally more efficient than a continuous one. That is, operating a 50% duty cycle may more than double the lifetime.
3) temperature – Batteries are generally less efficient at lower temperatures and more effficient at higher ones.
These effects have been extensively studied and modeled for many battery chemistries and structures, such as the Li-ion batteries used in mobile devices. This allows these systems to be speciffically optimized for battery lifetime.
By contrast, very little data is available about the behavior of the cheap, non-rechargeable batteries that are typically used in small wireless devices. Data is especially lacking about the behavior of these batteries under typical load patterns, with low duty cycles and relatively high peak currents, such as for the operation of a rceiver or sensor/actuator.
As a result, it is simply not known to what extent battery effects should be taken into account when evaluating protocols and applications for such systems.
We describe a testbed for automated measurement of the discharge behavior of Li-coin cells, give some preliminary results, and discuss sources of experimental uncertainty.
Additional work is clearly needed to analyse preliminary results and to better understand the testbed’s strengths and limitations. Nevertheless, our work has already demonstrated both the viability of the testbed and the importance of taking battery discharge behavior into account.
To read this external content in full, download the complete paper from the author online archives at Uppsala University.