H2BR: An HTTP/2-based Retransmission Technique to Improve the QoE of Adaptive Video Streaming Minh Nguyen minh.nguyen@aau.at Institute of Information Technology Alpen-Adria-Universität Klagenfurt Klagenfurt, Austria Christian Timmerer christian.timmerer@aau.at Institute of Information Technology Alpen-Adria-Universität Klagenfurt & Bitmovin Klagenfurt, Austria Hermann Hellwagner hermann.hellwagner@aau.at Institute of Information Technology Alpen-Adria-Universität Klagenfurt Klagenfurt, Austria ABSTRACT HTTP-based Adaptive Streaming (HAS) plays a key role in over-the- top video streaming. It contributes towards reducing the rebufering duration of video playout by adapting the video quality to the cur- rent network conditions. However, it incurs variations of video quality in a streaming session because of the throughput fuctua- tion, which impacts the user’s Quality of Experience (QoE). Besides, many adaptive bitrate (ABR) algorithms choose the lowest-quality segments at the beginning of the streaming session to ramp up the playout bufer as soon as possible. Although this strategy decreases the startup time, the users can be annoyed as they have to watch a low-quality video initially. In this paper, we propose an efcient retransmission technique, namely H2BR, to replace low-quality segments being stored in the playout bufer with higher-quality versions by using features of HTTP/2 including (i) stream priority, (ii) server push, and (iii) stream termination. The experimental re- sults show that H2BR helps users avoid watching low video quality during video playback and improves the user’s QoE. H2BR can decrease by up to more than 70% the time when the users sufer the lowest-quality video as well as benefts the QoE by up to 13%. CCS CONCEPTS · Information systems → Multimedia streaming; · Networks → Application layer protocols. KEYWORDS HTTP adaptive streaming, DASH, ABR algorithms, QoE, HTTP/2 ACM Reference Format: Minh Nguyen, Christian Timmerer, and Hermann Hellwagner. 2020. H2BR: An HTTP/2-based Retransmission Technique to Improve the QoE of Adap- tive Video Streaming. In Packet Video Workshop (PV’20), June 10ś11, 2020, Istanbul, Turkey. ACM, Istanbul, Turkey, 7 pages. https://doi.org/10.1145/33 86292.3397117 Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for proft or commercial advantage and that copies bear this notice and the full citation on the frst page. Copyrights for components of this work owned by others than ACM must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specifc permission and/or a fee. Request permissions from permissions@acm.org. PV’20, June 10ś11, 2020, Istanbul, Turkey © 2020 Association for Computing Machinery. ACM ISBN 978-1-4503-7946-5/20/06. . . $15.00 https://doi.org/10.1145/3386292.3397117 1 INTRODUCTION HTTP-based Adaptive Streaming (HAS) has become the de facto standard for delivering video over the Internet [1]. In HAS, multiple versions are generated from an original video at the server in order to adapt to the network. Each version is then divided into temporal segments of the same duration. To retrieve the video content, the client sends HTTP requests which fetch the most suitable versions for the next segments based on an adaptive bitrate (ABR) algorithm employed at the client. The ABR algorithm can rely on the client parameters such as bufer status, or network parameters such as throughput. There have been numerous ABR algorithms proposed to be im- plemented at the client side which can be clustered into bandwidth- based, bufer-based, and mixed adaptation [2]. The main goal of these methods is to adapt the video quality to the current network condition for the purpose of avoiding streaming issues like bufer drainage. However, they sometimes lead to video quality variations as the throughput fuctuates. These quality switches may signif- cantly impact the user’s Quality of Experience (QoE) [3]. Most of the ABR algorithms aim to request higher video quality versions when the throughput increases but ignore to improve the quality of already bufered low-quality, but not played-out segments even though the throughput may be enough to sustain more than one segment delivery at the same time. Meanwhile, a trace containing video streaming information collected from Akamai’s video con- tent delivery network (CDN) shows that almost 36% of the sessions sufer at least one downward variation in video quality [4]. Addi- tionally, many state-of-the-art ABR algorithms start a streaming session (i.e., the startup phase) with the lowest-quality segments to ramp up the bufer and, as a result, decrease the startup delay. However, users may quickly terminate a streaming session if the video quality is not good [5, 6]. HTTP/2 was released in 2015 by the Internet Engineering Task Force (IETF) [7]. It was issued as a higher performance alternative over its predecessor HTTP/1.1 and evaluated in the context of HAS by Mueller et al. [8]. It provides several key features including (i) server push, where website assets can be pushed to the client with- out user’s explicit requests for them, (ii) stream priority that enables the client to indicate how it favors the server to allocate resources when pushing concurrent data, and (iii) stream termination where unusable data can be terminated immediately by users. These fea- tures allow HTTP/2 to make improvements such as smaller latency and reduction of HTTP request overhead compared to HTTP/1.1.