Integrated Power Management for Video Streaming to Mobile Handheld Devices

Op timizing user exp erience or streaming videoap p lications on handheld devices is a significant research challenge. In this p ap er, wep rop ose an integratedp ower managementap p roach that unifies low level architectural op timizations (CPU, memory, register), OSp ower-saving mechanisms (Dynamic Voltage Scaling) and adapf ve middleware techniques (admission control, op timal transcoding, network tra#c regulation). Sp ecifically, we identi y interactionp arameters between the di - erent levels andop timize them to significantly reducep ower consump( on. With knowledge o device configurations, dynamic devicep arameters and changing system conditions, the middleware layer selects an ap(W p riate video quality and fine tunes the architecture orop timized delivery o video. Our p er ormance results indicate that architectural opP mizations that are cognizant o user levelp arameters(e.g. transcoded video quality) canp rovide energy gains as high as 57.5% or the CPU and memory. Middleware adap tations to changing network noise levels can save as much as 70% o energy consumed by the wireless network inter ace. Furthermore, we demonstrate how such an integrated ramework, that sup p orts tight coup ling o inter-levelp arameters can enhance user exp erience on a handheld substantially