Automatic-repeat-request protocols for data communication networks
Yet Keung Chang · cIRcle (University of British Columbia) · 2010
The performances of a number of Automatic-Repeat-Request (ARQ) protocols are compared on the basis of the expected wasted time per message incurred over random-error and Rayleigh fading channels. These include the standard Stop-And-Wait, Go-Back-N, and Selective-Repeat schemes. The reductions in expected wasted times achievable through the use of forward-error-correction (FEC) are demonstrated. It is found that in general, substantial improvements in performance can be obtained by using FEC. A new ARQ protocol proposed by Weldon is studied. It is shown that the throughput of Weldon's scheme can be increased by allowing multiple copies of a new data block to be sent. In order to maximize the throughput in Weldon's scheme, a number of parameters need to be selected optimally. An efficient method for choosing these parameters is obtained by exploiting the structure of a simplified expression for the throughput. The problem of the optimal block length that minimizes the expected wasted time per message is also considered. An exact analysis of the optimal block length is developed for the Stop-And-Wait scheme using an end-of-message character in the last block of a message. The optimal block length is a function of average message length, channel error rate, overhead per packet, acknowledgement delay and transmission rate. It is found that the optimal block length converges to a constant value when the average message length becomes large. Finally, the performance of an algorithm that computes the optimal block length in an adaptive way is examined.