Packet Delay Statistics of the Multichannel Selective-Repeat Automatic-Repeat-Request

Abstract

In this paper, we conduct stochastic modeling and analysis of the packet end-to-end delay in a multichannel selective-repeat automatic-repeat-request (MSR-ARQ) protocol. In this protocol, the transmitter continuously transmits packets over multiple parallel channels and retransmits erroneously received packets with either dynamic or static packet-to-channel scheduling policy. Under the assumption that packets are always supplied at the transmitter, denoted by the saturated traffic condition, we analyze the steady state probability distribution function of the delay of an arbitrary packet, which is measured by the duration between the instant at which the packet is transmitted for the first time and the time it departs from the resequencing queue at the receiver. Using the analysis result, we numerically compute the distribution function for chosen values of the number of channels and the error rates to demonstrate the computational effectiveness of the result. With numerical and simulation results, we then study the performance of MSR-ARQ in terms of the mean packet delay and compare the two scheduling policies. It is shown that the dynamic scheduling achieves a better packet delay performance than the static scheduling. With the dynamic scheduling and the presence of difference between the error rates of parallel channels, the mean packet delay decreases as the difference between channels’ error rates increases. Moreover, the number of parallel channels has an insignificant impact on the mean packet delay, which shows that the use of parallel channels is favorable for the wireless or mobile communications to increase the data transmission rate while keeping the mean packet delay at an acceptable level.