Cross-layer analysis via Markov models of incremental redundancy hybrid ARQ over underwater acoustic channels

Abstract

Underwater acoustic networks make it possible to wirelessly convey information, e.g., coming from measurements and sensing applications from under water to the surface. However, underwater communications are characterized by long delays, small available bandwidths and high error rates. These aspects may significantly affect the design of a reliable data-link layer for such systems. In this paper, we assess the performance of hybrid automatic repeat request error control schemes and we evaluate their application to improve the reliability of time-varying underwater acoustic links. We employ a parametric Markov model, which has been trained over channel traces collected during at-sea experiments. The results, based on both experimental data and analysis, suggest that parametric probabilistic representations, such as the considered Markov model, are good candidates for describing the correlated underwater acoustic channel dynamics, and may be employed to achieve a realistic evaluation of the data-link layer performance for underwater acoustic scenarios. Analytical and simulation results confirm that incremental redundancy improves the throughput of underwater acoustic links, even when real channel conditions, such as those encountered in the considered experiments, have wide dynamics over time. Finally, this kind of evaluations, beyond the data-link design, can also be employed at the network level for routing and network deployment considerations.