Reducing Backlight Power Consumption for Streaming Video Applications on Mobile Handheld Devices

Mobile handheld devices have stringent constraints on power consumption because they run on batteries that have a limited lifetime. Conserving power to prolong battery life is of primary importance for these devices. Several factors such as backlight intensity, the hard disk, the CPU, the network interface and the nature of the application contribute significantly towards power consumption for a mobile device. While significant research effort has been made to optimize power consumption at the application, network and processor levels, comparatively little work has been done to reduce or adapt to the power consumed by the backlight. In this paper, we propose an adaptive middleware based approach to optimize backlight power consumption for mobile handheld devices when playing streaming MPEG-1 video, without significantly compromising on video quality. Our performance results indicate that up to 60% of the power consumed by the backlight can be saved by using the proposed approach. 1. Motivation Modern handheld mobile devices are increasingly being used to provide users with streaming multimedia content. These devices due to their modest sizes and weights have limited computing, storage and battery resources. Moreover multimedia applications tend to be extremely resource hungry, due to which streaming multimedia content onto low-power mobile handheld devices introduces a significant research challenge. Typically, the CPU, the network interface and the display constitute the three primary sources of power consumption in low power devices. In this work, we focus on achieving energy savings from the backlight display of the device, without significantly compromising on the quality of the streamed video. Researchers have proposed several schemes in recent years, to optimize power consumption of low-power devices in mobile environments. Techniques such as dynamic voltage scaling (DVS) (2,3), dynamic power management of disks (10), network interfaces (8) and compiler/OS/middleware based adaptation (4,5,11) attempt to reduce power consumption at various computational levels. However, efforts to reduce the power consumption of the backlight