Complexity constrained layering of broadcast video for heterogeneous mobile receivers

It is well known that advances in battery technology for mobile devices have not kept pace with advances in memory, graphics, and processing power. As a result, power consumption remains a top challenge for mobile systems and applications. One of the most power hungry components of mobile video is the computational complexity of the video codec. This complexity is increasing with newer codecs such as the High Efficiency Video Codec (HEVC) making it more challenging to build mobile video apps that don't drain the mobile device's battery quickly. In this paper, we address the specific application of video broadcasting to a large number of mobile users using scalable video coding (SVC). We propose an approach for determining the number of required layers in SVC and optimizing the quality of experience of end users given the heterogeneous complexity constraints of the receivers. Our approach leads to lower battery consumption and longer video playing time by adhering to each receiver's complexity constraint. We focus on the motion compensation process, including motion vector prediction and interpolation, because it is the largest power consumer. We formulate the optimization problem and propose two heuristics as practical solutions. Our results show that our methods can achieve near-optimal solutions with negligible error.