Predictive Selective Repeat - an Optimized Selective Repeat for Noisy Channels

Abstract

Automatic Repeat reQuest (ARQ) is a technique used in two-way communication systems to make sure that the transmitted data is received properly without any errors. The underlying mechanism on which ARQ operates is the acknowledgment (ACK) sent by the receiver to the sender on the orderly arrival of data frames. Some protocols which have been employed include Stop and Wait ARQ, Go-Back-N ARQ, and Selective Repeat ARQ. There have been many attempts to optimize these protocols to decrease the number of Round-Trips and increase the throughput. Predictive Selective Repeat (PSR) ARQ is the technique introduced in this paper to send multiple copies of frames without waiting for ACK. The main goal of PSR ARQ model is to optimally predict the number of required copies for a particular frame so as to reduce the number of Round-Trips by using a Time series Forecasting Long Short-Term Memory (LSTM) model. © 2023 IEEE.