Reliable Point-to-Point Underwater Acoustic Data Transfer: To Juggle or Not to Juggle?

Reliable data transfer speeds using underwater acoustic communication systems are limited by long propagation delays, small link data rates, and high bit error rates. We consider the practical problem of transferring a data file or data stream reliably from one half-duplex underwater node to another. In a typical automatic repeat request (ARQ) approach, a node transmits one or more packets and waits for the corresponding acknowledgments (ACKs). With long propagation delay, the long waiting time for ACKs results in low average throughput. The long propagation delay, however, presents an opportunity for two nodes to simultaneously transmit data and ACKs toward each other in a juggling-like approach, potentially reducing the average waiting time for ACKs. The approach needs to satisfy certain timing constraints, and its performance is largely dependent on the network settings and chosen parameters. Through analytical and numerical studies, we provide key insights into appropriate choice of ARQ strategies and protocol parameters under different internodal propagation delays. We show that the juggling-like ARQ (J-ARQ) provides good data streaming throughput but performs poorly for small file transfers. We propose a novel rate-less code-based J-ARQ protocol that overcomes this limitation and offers high data transfer speeds for small files in long propagation delay environments.