Analysis Of Energy Consumption On3G Terminals

Smartphones equipped with powerful CPUs and3G wire-less communicationmoduleshavegainedalargepopularitynowadays, whichstimulatesalargespectrumofhigh-bandwidth applications such as video streaming, online games and content-richweb browsing booming on smartphones. As the capability of smartphones keeps soar-ing, battery technology remains a bottleneck. Having those applications mobile is funbut how to preserve energy on smartphones is of great importance. Recently, the tailefect of3G interface has gained much attention, which refers to the interface will keepat high-power states for a long time even after the completion of data transmission.Studies[1][2] have reported that the tail efect of3G interface can consume up to60%of the total energy.To eliminate the tail efect in3G networks, however, is very challenging due tothe following reason. The tails are designed to achieve fast response to upcoming da-ta transmissions. Simply cutting the tails without care would cause non-negligibledelays since the interface needs time to switch from low-power (idle) states to high-power (data-transmission-ready) states each time when the interface is not ready forthe required data transmission. Long promotion delay would seriously degrade userexperience. Therefore, a trade-of between minimal energy waste and good user expe-rience needs to be considered twice when dealing with the tail efect. There exists nosuccessful solution, as far as we know, to resolving the problem of tail efects in3Gnetworks.In this paper, we frst collect a real trace of3G network trafc, involving1,000users over two weeks. By analyzing the empirical data, we confrm that the tail efectcontributes more than60%to the total power consumption used for data communica-tion. Furthermore, in order to fnd whether there exist clear patterns embedded in3Gtrafc, we examine the packet arrival time from trafc generated by three categories of popular applications, i.e., video streaming, web browsing and instant messaging. Wefnd that3G trafc aroused by all those applications shows strong temporal correlation.Inspired by this observation, we propose an innovative scheme, called SmartCut,which integrates three key techniques to mitigate the tail efect of3G networks whilekeeps most applications running on smartphones un-afected. The core idea of Smart-Cut is to train an autoregressive move average (ARMA) model using historical3Gtrafc trace, based on which SmartCut consecutively predicts the arrival time of up-coming packets. With the estimated packet arrival time, SmartCut frst adopts the fastdormancymechanismtocuttheunnecessaryhigh-power-statetailsof. Moreover,ital-so promotes the3G interface in advance before the next packet arrives. In the case thatSmartCut makes a wrong prediction, variation of past estimations is utilized to reducethe infuence. The advantage of SmartCut is two-fold. First, SmartCut can signif-cantly reduce the energy waste caused by tail with little side-efect on user experience.Second, SmartCut is light weighted and easy to implement without the requirementof any changes to up-layer applications. We implement a prototype on Android-basedsmartphones and conduct small-scale experiments. The experiment results on the pro-totype demonstrate the feasibility of SmartCut. Furthermore, we conduct extensivetrace-driven simulations based on the large3G trace that we have collected and resultsshow that SmartCut can save56.57%energy on average among diferent applications.