Extended Abstract. Video streaming is currently one of the most popular web applications. While HTTP streaming delivers video with a static bitrate, DASH (Dynamic Adaptive Streaming over HTTP) can change the video bitrate during the video playback. These two methods provide video streaming service on the existing web architecture. The video clips are hosted as web objects and are accessed using HTTP. Currently, HTTP streaming and DASH tend to choose a low initial video bitrate to prevent

more »

... from suffering a long buffering time. Delivering at a low bitrate also under-utilizes network resources and cannot provide the best QoE (Quality of Experience) to users [4] . Even though DASH ramps up the quality for fast clients during the streaming [1], frequent switching of bitrate can hurt the QoE [5] . Knowing the network conditions before streaming is critical in selecting the bitrate. However, existing tools for measuring network performance either incur a high overhead or are difficult to deploy. For example, flooding-based tools (e.g. speedtest services) can be implemented with web technologies, but they often generate a high overhead. Probe-optimized measurement tools (e.g., Pathload) require extra software installation and cannot be easily deployed in browser [3] . Client Quality Oracle Probe Kit