RWTH Aachen University
xPrio: Crosslayer Web Resource Prioritization Approach
Pages
16
Time to read
46 mins
Publication
Language
English
Pages
16
Time to read
46 mins
Publication
Language
English
This document is a technical report that presents xPrio, a novel approach to HTTP resource prioritization that leverages runtime data from both browser and transport layers. Traditional methods often rely on precomputed resource dependencies and network states, which can be costly and impractical due to the dynamic nature of web content. xPrio addresses this limitation by utilizing reinforcement learning to dynamically prioritize web resources as they are requested, based on real-time information. The report details how xPrio integrates browser prioritization signals with transport layer data, such as bandwidth and round-trip time (RTT), to optimize resource loading and improve webpage performance. The evaluation of xPrio demonstrates significant performance enhancements, achieving mean SpeedIndex speedups exceeding 15% on pages from the Alexa Top 500. This approach represents a shift towards more adaptive and efficient web resource management, reducing the overhead typically associated with traditional prioritization strategies.