Delay analysis for AIMD flows in wireless/IP networks

End-to-end delay and delay jitter are critical QoS (quality of service) parameters for time sensitive applications. In wireless/IP hybrid networks, the wireless link is presumably the bottleneck. Because of the time-varying and error prone channel, the wireless channel throughput is random. On the other hand, with closed-loop AIMD (additive increase and multiplicative decrease) congestion control, the arrival process to a queue is not on-off. In this paper, given the measured channel profile and the transmission scheme used in the link layer, the wireless channel throughput distribution is derived. By analyzing the queue in the wireless domain with an AIMD controlled flow, the queuing delay distribution is derived. It is found that by appropriately setting the protocol parameters for TCP-friendly AIMD flows, the wireless spectrum can be efficiently utilized, and the QoS requirements of time sensitive applications can be statistically guaranteed.