Improving the Quality of Experience by Enhancing the Efficiency of Viewing Consecutive Short Videos Over HTTP (DASH)

Abstract

This study focuses on optimizing the Dynamic Adaptive Streaming over HTTP (DASH) technique to improve the viewing experience of consecutive short videos, with a particular emphasis on enhancing the Quality of Experience (QoE) for users. A pioneering example of such an approach can be found in Özbek & Arıcıoğlu (2019), where user-defined adaptation priorities significantly improved subjective QoE ratings. A comprehensive analysis of user behavior—including swiping actions and feedback—is conducted to identify the critical factors affecting video streaming performance. Based on these insights, innovative modifications to segment loading and prefetching strategies are proposed, aiming to deliver a smoother, interruption-free, and high-quality viewing experience even under constrained network conditions. To validate segment prioritization and evaluate perceived quality across resolutions, objective video quality metrics such as VMAF, PSNR, and SSIM are considered essential. Recent datasets mapping segment-level features to these metrics across resolutions provide valuable benchmarks for DASH-based encoding strategies (Micó-Enguídanos et al., 2024). Similar energy-aware optimization strategies for quality and resource trade-offs in DASH-based systems have also been explored through deep reinforcement learning in transcoding environments (Kim et al., 2024). A prototype DASH client and server are developed to simulate various network scenarios and evaluate the proposed enhancements. The experimental framework employs both objective metrics— such as startup delay and buffer events—and indirect inferences of user-perceived quality (Mok et al., 2011) improvements. This approach ensures a practical assessment of both technical performance and viewing continuity. The anticipated contributions of this research include the development of scalable, user-centric solutions that enhance the overall efficiency of video streaming platforms, ultimately benefiting both academic research and industrial practices in the field of adaptive streaming technologies.