A Comparison of Request and Measurement Based End to End Video Target Rate Adaptation Methods

In this paper, we compare a request based video target rate adaptation method to a measurement based adaptation method. It is shown that by using methods, that use both estimation of the transport delay and its variation and packet loss jointly, it is possible to reduce packet losses and transport delay variation, when the transport channel is temporarily congested or suffering from other bad conditions (like fading radio access). The lower packet loss and transport delay variation consequently improves the perceptual quality of the video signal significantly. Further it is shown by simulated experiments, that request based video adaptation method is less sensitive to the nominal RTCP sender report (SR) interval than a measurement based adaptation, when same mathematical algorithms for target rate estimation are used.