Power-rate-quality optimized scalable video adaptation

Abstract

In this paper, we investigate the power-rate constrained scalable video adaptation for popular wireless video streaming application, where the wireless access network bandwidth and mobile remaining battery capacity are usually limited. Towards this goal, we have developed a scalable video decoding complexity model with the focus on the joint temporal and amplitude scalability, which can be translated to the power consumption model easily for mobile processor. Overall, there are three parameters for our proposed decoding complexity model. Currently, we propose to embed these parameters in the header field. We have validated our complexity model using various videos with different contents, resolutions, and bit rates. Results show that our proposed model can estimate the scalable video decoding complexity accurately with small root mean square error (RMSE) and high Pearson correlation (PC). Together with our rate and perceptual quality models for scalable video, we have made the power-rate constrained scalable video adaptation analytically tractable without requiring exhaustive search.