Research Projects

This proposal aims to design a video streaming framework that can sustain low bandwidth networks, particularly in developing countries like India. The new HTTP protocol, HTTP/3, uses QUIC Quick UDP Internet Connection at its core, which is advantageous over TCP for heavy-tailed latency characteristics and supports consistent quality of experience (QoE) for applications like video streaming. However, some studies argue that QUIC suffers in comparison to TCP, but these observations are not clearly identified. The proposal will analyze HTTP/3's behavior over an experimental setup to answer open questions about QUIC's performance over the Internet and develop a sustainable video streaming framework for low bandwidth networks. The current internet is full of middleboxes, such as firewalls and proxies, which often block UDP frames for network security purposes. To alleviate this problem, QUIC adapts its behavior by racing a TCP connection with QUIC, with the failed QUIC connection marked as broken for a timeout duration. However, this fallback mechanism may have potential side effects causing unexpected impacts on an application's performance. The fallback mechanism does not explicitly check whether the QUIC connection failure is due to the presence of a middlebox blocking the underlying UDP segments. It is highly possible to have a momentary connection failure due to poor end-to-end network connectivity between remote hosts. The study will investigate the frequency of fallbacks over a QUIC-enabled stream from an HTTP/3-supported browser, its long-term performance compared to TCP, the severity of the middlebox-adaptive behavior of QUIC's fallback to TCP, and potential solution approaches for preventing unnecessary fallbacks.

0