CSR: Small: Towards Energy-efficient Internet Mobile Streaming

如今，越来越多的人在移动设备上随时随地享受流媒体服务。然而，在移动设备上接收流媒体服务从根本上受到移动设备上有限的电池供电的限制。在流会话中，通过无线网卡接收流数据，用CPU解码接收到的数据，并将解码后的数据显示在屏幕上，持续消耗大量电池电量。尽管在节省电池功耗方面已经做出了巨大的努力，但是由于设备硬件、操作系统和应用算法等的异构性，现有的流实践带来了许多新的复杂性。这些复杂性导致能源效率下降。本项目旨在研究在不同阶段（从数据接收到图像显示）促进节能互联网移动流媒体服务的新方案。该项目主要通过研究移动设备上的数据传输调度，研究编解码感知的动态电压和频率缩放，以及设计流媒体会话中移动设备的动态内容自适应显示方案来解决这些挑战。新设计的方案可以直接帮助移动设备显著提高流媒体服务的能源效率。在其他耗电量大的移动应用中，它们也有很大的潜力来节省电池功耗。研究成果将纳入课程发展，参与的本科生和研究生将被训练成具有计算机系统研究经验的下一代计算机科学家和工程师。

Today more people are enjoying streaming services on their mobile devices from anywhere at anytime. However, receiving streaming services on mobile devices is fundamentally constrained by the limited battery power supply on a mobile device. In a streaming session, a large amount of battery power is consumed continuously for receiving the streaming data via the wireless network card, decoding the received data with the CPU, and displaying the decoded data on the screen. Although significant efforts have been made to save the battery power consumption, the existing streaming practice brings many new complexities due to the heterogeneity of device hardware, operating systems, and application algorithms, etc. These complexities lead to degraded energy efficiency. This project aims to investigate new schemes to facilitate energy-efficient Internet mobile streaming services at different phases, ranging from the data receiving to the image displaying. The project addresses the challenges by mainly looking into the data transmission schedule on mobile devices, studying codec-aware dynamic voltage and frequency scaling, and devising dynamic content-adaptive displaying schemes for mobile devices in streaming sessions. The newly designed schemes can directly help mobile devices significantly improve the energy efficiency in streaming services. They also have great potentials to save battery power consumption in other power-hungry mobile applications. The research outcomes will be integrated into curriculum development and the participating undergraduate and graduate students will be trained with computer system research experience as the next generation computer scientists and engineers.