17th European Performance Engineering Workshop, EPEW 2021, and the 26th International Conference on Analytical and Stochastic Modelling Techniques and Applications, ASMTA 2021, Virtual, Online, 13 - 14 December 2021, vol.13104 LNCS, pp.472-487
© 2021, Springer Nature Switzerland AG.We present an analytical framework to obtain the distribution of frozen, low quality, and high quality video in a setting with two video quality levels where the channel is dynamic and the data rate the user can achieve varies with time. The presented model, which is based on multi-regime Markov fluid queues, is also capable of producing the distribution of the excess data present in the playout buffer at the end of the video session duration, which will be wasted. The playout control is assumed to be hysteretic, and the effects of the values of thresholds selected for starting playout, switching to low/high quality levels, and pausing/resuming download on the distribution of video quality and excess data is investigated. The presented model can be extended to quality levels more than two.