Joint Delay-Sensitive and Power-Efficient Quality Control of Dynamic Video Streaming Using Adaptive Super-Resolution

Abstract

As mobile devices exhibit exceptional processing capabilities, super-resolution (SR) technology for video quality enhancement has developed. However, SR is still computationally-intensive, and the integration of SR into video streaming systems remains a nascent domain, with limited recent studies. To address this concern, we design an adaptive SR model that can dynamically control quality enhancement and computational demands in response to the available computational resources and input quality. Furthermore, we propose an integrated strategy for video segment delivery and resource allocation within video streaming systems aided by adaptive SR. This technique optimizes key parameters encompassing transmitter decisions (e.g., transcoding rate, the number of delivering segments, and transmit power) and receiver decisions (e.g., quality enhancement rate and computing resources). Our data-intensive simulation results demonstrate the adaptive SR-centric video streaming system that strikes a harmonious balance across diverse performance metrics. These metrics encompass average video quality, playback stall rates, transmission power, and computational resource consumption.