Channel estimation under short pilot length in deep-sea underwater acoustic communications.

Deep-sea underwater acoustic (UWA) channels typically exhibit pronounced sparsity and long delay spreads, with the sparsity pattern highly dependent on the locations of the transmitter and receiver. These characteristics bring significant challenges for channel estimator design. As a result, most existing methods require long pilot sequences to achieve accurate channel estimation. However, the use of lengthy pilots introduces considerable communication overhead and latency, which is not desirable in practice. To overcome this challenge, we propose a flexible, sparsity-aware channel estimation algorithm based on a generalized inverse Gaussian (GIG) prior. This approach eliminates the need of heavy parameter tuning, effectively accommodates diverse sparsity levels, and fully exploits the inherent sparsity of UWA channels. Consequently, the required pilot length can be reduced to approximately the channel length, while still ensuring accurate channel recovery and noise variance estimation. Simulation results demonstrate that the proposed GIG prior-based algorithm maintains high accuracy across a wide range of sparsity patterns, even when the pilot length is comparable to the channel length. Furthermore, experiments using real-world data from the South China Sea show that the proposed algorithm consistently achieves lower bit error rate than other state-of-the-art channel estimators, regardless of the equalizers used.