Rate Adaptation Technique for Media Streaming Over QUIC With Limited Backhaul

Abstract

The rapid proliferation of mobile devices has caused a significant surge in Internet traffic, mainly due to video streaming services, which account for up to 65% of the total volume. As a primary contributor to Internet traffic, video streaming services must use bandwidth efficiently to avoid overwhelming networks with unnecessary traffic. It is not trivial for media streaming services to utilize bandwidth efficiently while achieving scalability and interactivity. Media over QUIC (MoQ) is a promising new standard for realizing both scalable and interactive streaming services. However, due to its infancy, the selection of a proper bit rate for each video segment to be transmitted (also known as rate adaptation) in MoQ has yet to be explored. Inappropriate video bit rate selection can lead to a deterioration of the user’s quality of experience (QoE) and inefficient bandwidth utilization. Furthermore, the rate adaptation problem becomes more complex due to the limitation of backhaul bandwidth. The rate adaptation technique should consider the client parameters (i.e., client bandwidth) and backhaul bandwidth usage known only to the server (server parameters). This paper proposes the rate adaptation technique designed for MoQ that considers both the client and server parameters. We have implemented the proposed scheme and evaluated its performance under realistic scenarios. Our experimental results reveal that the proposed technique has a 70% lower packet delay rate and a 10% higher bandwidth utilization compared to the IFA-based, RENO-based, and DASH-Google. These results demonstrate the novelty and the benefits of the proposed scheme.

Keywords

• Bit rate adaptation
• media streaming
• QUIC
• dynamic adaptive streaming over HTTP (DASH)
• limited backhaul
• mixed integer linear programming (MILP)