Abstract

Wireless video streaming is becoming more prevalent in this day and age. When streaming video, network signal fluctuations can cause interrupts in the video playback. These fluctuations are even more likely to occur when the network is wireless and the device is mobile. To counteract this, web video players use buffers to store the video data ahead of time. However, if the buffer is too small or the network is too unstable, interruptions may still occur. One popular streaming service that implements client buffers is YouTube. In this paper, YouTube video playback when streaming to a mobile wireless device on the WPI campus is analyzed. The experiment starts out by streaming video while walking along predetermined paths. Throughput traces are captured using a network sniffer and run through a video play back simulator with varying buffer sizes. A cost function is developed by weighing different variables that attribute to the quality of the viewing experience such as initial buffer size, interrupted time, interrupt frequency and duration of playback before the first interrupt. Using simulator results, the initial buffer size that results in the lowest cost is determined.

Download:

• report (pdf)