Cutting without Pain: Mitigating 3G Radio Tail Effect on Smartphones Fei Yu † , Guangtao Xue † , Hongzi Zhu † , Zhenxian Hu † , Minglu Li † , Gong Zhang ‡ † Shanghai Jiao Tong University, China ‡ Huawei Research, China † {fishfly 1008,gt xue,hongzi,huzhenxian,mlli}@sjtu.edu.cn; ‡ nicholas.zhang@huawei.com Abstract—3G technology has stimulated a wide variety of high- bandwidth applications on smartphones, such as video streaming and content-rich web browsing. Although having those applica- tions mobile is quite appealing, high data rate transmission also poses huge demand for power. It has been revealed that the tail effect in 3G radio operation results in significant energy drain on smartphones. Recent fast dormancy technique can be utilized to remove tails but, without care, can degrades user experience. In this paper, we propose a novel scheme SmartCut, which effectively mitigates the tail effect of radio usage in 3G networks with little side-effect on user experience. The core idea of SmartCut is to utilize the temporal correlation of packet arrivals to predict upcoming data, based on which unnecessary high-power-state tails of radio are cut out leveraging the Fast Dormancy mechanism. Extensive trace-driven simulation results demonstrate the efficacy of SmartCut design. On average, SmartCut can save up to 56.57% energy on average while having little side-effect to user experience. I. I NTRODUCTION Smartphones equipped with powerful CPUs and 3G wire- less communication modules have gained a large popular- ity nowadays, which stimulates a large spectrum of high- bandwidth applications such as video streaming, online games and content-rich web browsing booming on smartphones. As the capability of smartphones keeps soaring, battery tech- nology remains a bottleneck. Recently, the tail effect of 3G interface has gained much attention, which refers to the interface will keep at high-power states for a long time even after the completion of data transmission. Studies[1][2] have reported that the tail effect of 3G interface can consume up to 60% of the total energy. To eliminate the tail effect in 3G networks, however, is very challenging due to two reasons. First, the tails are designed to achieve fast response to upcoming data transmissions. Simply cutting the tails without care would cause non-negligible delays 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 for the required data transmission. Long promotion delay would seriously degrade user experience. Second, in order to determine energy-efficient tails, it is inevitable to know the precise data work load in the future. In practise, this is very hard, if not impossible. In this paper, we first collect a real trace of 3G network traffic from a metropolis in China, involving more than 65,000 users over a month. By analyzing the empirical data, we confirm that the tail effect contributes more than 60% to the total power consumption used for data communication. Furthermore, we examine the packet arrival time from traffic generated by three categories of popular applications, i.e., video streaming, web browsing and instant messaging and find that 3G traffic aroused by all those applications shows strong temporal correlation. Inspired by this observation, we propose an innovative scheme, called SmartCut, which integrates two key techniques to mitigate the tail effect of 3G networks while keeps most applications running on smartphones un-affected. The core idea of SmartCut is to train an autoregressive move average (ARMA) model using historical 3G traffic trace, based on which SmartCut consecutively predicts the arrival time of upcoming packets. With the estimated packet arrival time, SmartCut first adopts the fast dormancy mechanism to cut the unnecessary high-power-state tails off. Moreover, in order to reduce the side-effect on user experience, it also promotes the 3G interface in advance before the next packet arrives. Trace-driven simulations show that SmartCut can save 56.57% energy on average among different applications. The remainder of the paper is organized as follows. Section 2 presents the related work. Section 3 introduces the system model. In Section 4, we describe the collected 3G data trace and then analyze the impact of the tail effect. The SmartCut scheme design is presented in Section 5. In Section 6, we describe the methodology to evaluate the performance of SmartCut and present the result. Finally, we give a conclusion of our work in Section 7. II. RELATED WORK The tail effect, where the cellular radio remains in a high power state for a long duration, contributes a large fraction of total radio energy consumption[2]. How to decrease the tail effect is arousing significant attention. In the literature, several schemes have been proposed to reduce the impact of tail effect, trading off between minimal energy waste and good user experience. Basically, those schemes fall into two categories: tail-cutting and tail-sharing. Given the knowledge of future packets, tail-cutting schemes[3] remove the high- power-state tail by adopting the fast dormancy mechanism, under which a smartphone can ask for an immediate release of radio links and leave the high-power-states rapidly. Knowing 978-1-4673-5946-7/13/$31.00 ©2013 IEEE 2013 Proceedings IEEE INFOCOM 440