Playout mechanism for voice packets in wide-area networks

Abstract

During the past several years, there has been a significant increase in the use of packetized voice over wide-area packet-switched networks. In such applications, a principal challenge in supporting interactive voice is the need to provide synchronous playout of voice packets in the face of stochastic end-to-end network delays. The voice synchronization was an active research area in the late 70``s and early 80``s. The early proposed synchronization methods assumed that both the propagation delay and the distribution of the variable component of network delay are known. So, a fixed playout delay can be computed such that no more than a given fraction of arriving packets are lost due to late arrival. But, it is difficult to know the end-to-end delay distribution of packets. Also, the distribution can change over relatively short time scales. In this paper, a method to adjust adaptively the playout delay of voice packets is supposed. The playout delay is calculated from the estimate of packet network delays and changed from one talkspurt to the next. Within a talkspurt, packets are played out in a periodic manner, thus reproducing their periodic generation at the source. A packet having the network delay larger than it``s playout delay is dropped. The proposed algorithm to choose the playout delay in the face of varying network delay can achieve a lower rate of lost packets. The evaluation for the algorithm consist of simulation by experimentally generated traffic. The generated traffic has characteristics of reported wide-area internet traffic.