Adaptive Playout Mechanisms for Packetized Audio Applications in Wide-Area Networks
Reviewer: Jie Zhou
Audio applications over the packet-switched networks, like Internet, is much cheaper than over the traditional circuit-switched networks. However,
Internet delay measurements indicate that end-to-end delays may fluctuate rapidly and signicantly. This may affect the quality of interactive audio
severely. Addaptive algorithms are ways to solve this problem.
The paper investigates and compares the performance of four different algorithms for adaptively adjusting the playout delay for audio packets in the
face of varying network delays.
An adaptive algorithm which explicitly adjusts to the sharp, spike-like increase in packet delay can achieve a lower rate of lost packets for
both a given average playout delay and a given maximum buffer size.
I rate the paper as 3 (modest contribution), because it provides some insight of how receiver side of an audio application adjusts to the network
fluctuation. The author gives clear explanation and analysis of several addaptive algorithms, as well as extensive experimental data. Basically,
the paper is convincing.
The paper assumes that the sender sends the audio packet at a constant rate. If the spike-like increase in packet delay can be reduced by addaptive sending
rate, then the conclusion of this paper will become invalid. As illustrated in the paper, algorithm 4 performs a little worse than algorithm 1 when the
network has not large delay jitters.
In case of interactive applications on Internet, good service quality requires both low packet loss rate and short delay time.
How to quantify the distortion which occurs when silence periods are artificially contracted or expanded?
How about the comparation result of the discussed algorithms based on the distortion factor?