Uncoded acoustic communication in shallow waters with bursty impulsive noise

Abstract

The shallow underwater acoustic channel offers a challenging environment. Besides long delay spreads caused by multiple surface-bottom reflections, the channel is time variant as well. In tropical waters, the problem is compounded further by impulsive noise created by snapping shrimp. Conventionally, the noise process is modeled by white impulsive noise. However, in reality, snapping shrimp noise depicts memory and is thus bursty as well. We investigate the performance of a high-rate uncoded acoustic communication system operating in tropical shallow waters. The stationary α-sub-Gaussian noise with memory order m (αSGN(m)) model is employed as it characterizes both the temporal and amplitude statistics of snapping shrimp noise. We show that there is a stark deviation from the expected performance of the white maximum-likelihood (ML) detector when the ambient noise is αSGN(m) (m > 0). Moreover, we derive the ML detector for the shallow water acoustic channel with additive αSGN(m) and compare its performance to that of its white counterpart.