Joint signal processing and coding for underwater acoustic communications using ORBGRAND
TL;DR
This work presents a joint signal processing and coding (JSPC) framework that iteratively refines channel estimates using decoded codewords, enabling reliable communication with minimal parity overhead and outperforming differentially coherent detection, particularly at low signal-to-noise ratio (SNR).
Abstract
Achieving high-rate and reliable underwater acoustic communications remains challenging due to severe multipath propagation, extended delay spreads, and strong Doppler-induced time variability. Consequently, channel coding plays a critical role in balancing reliability and spectral efficiency; however, reported performance results are often difficult to compare due to heterogeneous experimental platforms, channel conditions, and system configurations. In addition, the lack of standardized underwater acoustic channel models limits systematic evaluation and reproducible benchmarking. To address these challenges, we present a joint signal processing and coding (JSPC) framework that iteratively refines channel estimates using decoded codewords, enabling reliable communication with minimal parity overhead. The proposed framework employs Ordered Reliability Bits Guessing Random Additive Noise Decoding (ORBGRAND), a soft-decision variant of the Guessing Random Additive Noise Decoding (GRAND) algorithm. Additionally, to enable reproducible benchmarking across diverse underwater conditions, the proposed approach is evaluated on measured channel impulse responses from a publicly accessible repository, using channels from two distinct experiments that span contrasting regimes, from long-range mobile links to short-range shallow-water links. Results demonstrate approximately 6–8 dB of coding gain over the uncoded baseline with as few as 8 parity bits, while outperforming differentially coherent detection, particularly at low signal-to-noise ratio(SNR).