V-edge: Fast Self-constructive Power Modeling of Smartphones Based on Battery Voltage Dynamics
Authors:
Fengyuan Xu, College of William and Mary; Yunxin Liu, Microsoft Research Asia; Qun Li, College of William and Mary; Yongguang Zhang, Microsoft Research Asia
Abstract:
System power models are important for power management and optimization on smartphones. However, existing approaches for power modeling have several limitations. Some require external power meters, which is not convenient for people to use. Other approaches either rely on the battery current sensing capability, which is not available on many smartphones, or take a long time to generate the power model. To overcome these limitations, we propose a new way of generating power models from battery voltage dynamics, called V-edge. V-edge is self-constructive and does not require current-sensing. Most importantly, it is fast in model building. Our implementation supports both component level power models and per-application energy accounting. Evaluation results using various benchmarks and applications show that the V-edge approach achieves high power modeling accuracy, and is two orders of magnitude faster than existing self-modeling approaches requiring no current-sensing.
USENIX is committed to Open Access to the research presented at our events. Papers and proceedings are freely available to everyone once the event begins. Any video, audio, and/or slides that are posted after the event are also free and open to everyone. Support USENIX and our commitment to Open Access.
BibTeX
@inproceedings {180294,
author = {Fengyuan Xu and Yunxin Liu and Qun Li and Yongguang Zhang},
title = {V-edge: Fast Self-constructive Power Modeling of Smartphones Based on Battery Voltage Dynamics},
booktitle = {10th USENIX Symposium on Networked Systems Design and Implementation (NSDI 13)},
year = {2013},
isbn = {978-1-931971-00-3},
address = {Lombard, IL},
pages = {43--55},
url = {https://www.usenix.org/conference/nsdi13/technical-sessions/presentation/xu_fengyuan},
publisher = {USENIX Association},
month = apr
}
Power models report on precisely how much power our applications and components consume on a mobile device. Ideally, all users could model their devices' power consumption and pinpoint how their most precious resource is being consumed. The problem is that measuring power consumption to any reasonably fine granularity is difficult, typically requiring special hardware components that many phones simply do not have. So, while power models work great in a lab, they cannot be deployed widely onto normal users' mobile devices. Until this paper, that is.
"At a first glance, this paper felt like magic." This was one of the reviewer's reactions to this paper, and it underlies both why the PC accepted it and why you will probably be as skeptical as we were at first blush. The magic in this paper is a method of inferring changes in current—something that many mobile phones do not measure—from changes in voltage—something most phones do measure. The authors demonstrate that this is possible because the batteries in most mobile phones today follow a model that permits a linear relationship between voltage and current changes. Sure, V=IR, but this "shouldn't" apply here; the battery's internal resistance "should" change with respect to remaining capacity. It does change, but for common batteries today, it turns out the change is very small. Voila.
Fortunately, good science isn't good magic; the authors dispel the illusion with a full implementation and evaluation of their approach, called V-edge, on popular mobile devices. The evaluation uses new power models that the authors admit aren't terribly accurate; there are arguably better models out there (for all but screens, anyway), and a proper deployment should probably use them. But the fact remains that *no* power model could have been applied to these unmodified devices prior to V-edge. V-edge inspires hope that as new power models are developed, they can be deployed and evaluated on a far wider set of devices and user environments.
connect with us