An Empirical Evaluation of VoIP Playout Buffer Dimensioning in Skype, Google Talk, and MSN Messenger

VoIP playout buffer dimensioning has been long a challeng-ing optimization problem as the buffer size should main-tain a balance between the conversational interactivity and speech quality. One of its challenges comes from the fact that many factors may affect the overall conversational qual-ity, and some of the factors may change over time. Although academic researchers have done numerous efforts in solving this problem, how their research results are applied in prac-tice remains unknown. In this paper, we investigate whether a gap between VoIP researchers and practitioners, from the perspective of play-out buffer dimensioning algorithms, exists. Targeting at three popular VoIP applications, namely, Skype, Google Talk, and MSN Messenger, we design experiments to sys-tematically measure how the applications adjust their play-out buffer sizes. By using an objective QoE (Quality of Ex-perience) metric, we show that all of the three applications do not adjust their buffer size very well. In other words, they could provide better QoE to users by simply replacing their buffer dimensioning algorithms. Also, all of them do not adapt the buffer size to the network loss rate, which should also be considered for optimal QoE provisioning