Analysing End-to-End Packet Delay and Loss in mobile ad hoc networks for interactive audio applications

Interactive audio applications such as audio conferencing and telephony require high constraints on delay, jitter and loss. The packets of these applications must be received without significant loss, with low delay and jitter. When packet loss rate exceeds 10% and one way delay exceeds 150 ms, speech quality can be quite poor. Human conversation tolerates a maximum end-to-end delay of between 150 and 300 milliseconds. In addition, these packets must have a small delay variation to maintain constant rate for successive audio packets at the destination. In ad hoc networks, many factors such load traffic, codec bit rate, routing protocol configuration and mobility speed have an impact on packet loss rate, delays, and jitter which degrade the quality of the received audio signal. In this paper, we analysed how these factors influence packet delay and loss observed at the reciever side. A best knowledge of this behaviour is important to develop more effective mechanisms for dynamic adjustment of playout delays or throughput, in order to improve the perceptual quality of audio applications running in ad hoc networks. At the receiver side, we distinguish two operating phases of an audio flow, normal phase and reconfiguration phase. During a normal phase the flow reaches the destination from the source on the same route without link breakage, while during a reconfiguration phase, nodes move on the current route and the audio flow transfer is interrupted for delay required from a routing protocol to establish a new route towards the destination.