Streaming video content over cellular connectivity impacts the battery consumption of a client (e.g., a smartphone). The problem is exacerbated when the channel quality is poor because of a large number of retransmissions, moreover, streaming high quality video in such cases can negatively impact user experience (e.g., due to stalling). In this paper, we develop an analytical framework which can provide the user with an estimate of "how much" energy she can save by choosing to view a lower quality stream of the video she wishes to view. The framework takes as input the network conditions (in terms of packet error rate or PER) and a coarse characterization of the video to be viewed (slow versus fast motion, resolution), and yields as output the energy savings with different resolutions of the video to be viewed. Thus empowered, the user can then make a quick, educated decision on the version of the video to view. We validate that our framework is extremely accurate in estimating the energy consumption via both simulations, and experiments on smartphones (within ~5% of real measurements). We find that switching to a lower resolution video can potentially lead to ~418 mW (23.2%) decrease in the consumed power for slow motion video, and ~480 mW (26%) for fast motion video in bad channel conditions. This translates to an energy savings of 376.2 J and 432 J respectively, for video clips that are 15 minutes long.
展开▼
