Browsing by Author "Saarikoski, Kasperi"

Network-intensive smartphone applications are becoming increasingly popular. Examples of such trending applications are social applications like Facebook that rely on always-on connectivity as well as multimedia streaming applications like YouTube. While the computing power of smartphones is constantly growing, the capacity of smartphone batteries is lagging behind. This imbalance has created an imperative for energy-efficient smartphone applications. One approach to increase the energy efficiency of smartphone applications is to optimize their network connections via traffic shaping. Many existing proposals for shaping smartphone network traffic depend on modifications on the smartphone OS, applications, or both. However, most modern smartphone OSes support establishing Virtual Private Networks (VPNs) from user-space applications. Our novel approach to traffic shaping takes advantage of this. We modified OpenVPN tunneling software to perform traffic shaping by altering TCP flow control on tunneled packets. Subjecting heterogenous network connections to traffic shaping without insight into traffic patterns causes serious problems to certain applications. One example of such applications are multimedia streaming applications. We developed a traffic identification feature which creates a mapping between Android applications and their network connections. We leverage this feature to selectively opt-out of shaping network traffic sensitive to traffic shaping. We demonstrate this by selectively shaping background traffic in the presence of multimedia traffic. The purpose of the developed traffic shaper is to enhance the energy efficiency of smartphone applications. We evaluate the traffic shaper by collecting network traffic traces and assessing them with an RRC simulator. The four performed experiments cover multimedia streaming traffic, simulated background traffic and concurrent multimedia and background traffic produced by simulation applications. We are able to enhance the energy efficiency of network transmissions across all experiments.