Description
A Master of Science thesis in Electrical Engineering by Nada Abdelhafez entitled, “Rate Adaptation in Dynamic Adaptive Video Streaming Over HTTP”, submitted in November 2021. Thesis advisor is Dr. Mohamed S. Hassan and thesis co-advisor is Dr. Taha Landolsi. Soft copy is available (Thesis, Completion Certificate, Approval Signatures, and AUS Archives Consent Form).
Abstract
Video streaming stands out as the most significant traffic type consumed by mobile devices. This increased demand has been a major driver for research on bitrate adaptation algorithms. Bitrate adaptation ensures high user-perceived quality, which, in turn, correlates with higher profits for content providers and delivery systems. Dynamic Adaptive Streaming over HTTP (DASH) is a widely adopted video streaming standard utilized by service providers to provide competitive Quality of Experience (QoE). It is capable of providing seamless streaming via uncertain network conditions by switching across different video qualities and their corresponding video segment bitrates. The complexity of the video streaming environment makes it a good candidate for different learning-based approaches. Accordingly, this thesis proposes and assesses different learning-based adaptation approaches. The first proposed approach is the hybrid QoE-based algorithm, which manages the trade-off between video quality, re-buffering and quality switching events that impact the user-perceived quality. The objective is to optimize the QoE metric which is constrained by the network throughput, segment size, and buffer occupancy to continuously select the optimum bitrate levels with low complexity. The second proposed approach is the state-of-the-art DQNReg, a reinforcement learning based technique that enhances the classical deep Q-learning method. A segment-wise QoE-based reward function is established so that the learning strategy can converge towards maximizing the QoE outcome. The proposed adaptation approaches have been thoroughly evaluated using trace-based simulation for fixed and mobile networks. The first hybrid QoE-based approach performance surpasses that of the benchmark algorithm and the DQNReg algorithm outperforms other existing methods. The evaluations show that the DQNReg significantly reduces the re-buffering duration while maintaining higher video quality and relatively lower quality variations.