Environment-aware video streaming optimization of power consumption

Abstract

Streaming services gradually support high-quality videos for the better user experience. However, high-quality video streaming on mobile devices consumes a considerable amount of energy. This thesis presents the design and prototype of two systems, EVSO and NeuSaver, which achieve power savings by applying adaptive frame rates to parts of videos with a little degradation of the user experience. Both EVSO and NeuSaver utilize a novel perceptual similarity measurement method based on human visual perception specialized for a video encoder. EVSO's streaming server preprocesses the videos into several videos according to the similarity intensity of each part of the video and then provides the client with the processed video suitable for the battery status of the client's mobile device. Meanwhile, NeuSaver's streaming server uses Reinforcement Learning (RL) to select the appropriate frame rates for each video chunk based on previous observations. NeuSaver automatically reinforces the RL model in response to the feedback of the environments, without relying on pre-defined models or policies. The EVSO and NeuSaver were implemented on the commonly used H.264/AVC encoder. We conduct various experiments and a user study with ten videos. Our experimental results show that EVSO effectively reduces the energy consumption of mobile devices by 22% on average and up to 27% while maintaining the quality of the user experience.