Reliable acoustic communication between submersibles and surface vessels plays a critical role in deep-sea exploration. Adaptive turbo equalization can effectively combat the selective fading of underwater acoustic channels, thereby becoming one of the enabling technologies for single-carrier deep-sea vertical acoustic communications. Existing adaptive turbo equalizer designs are usually based on a minimum-mean-squared-error criterion or a minimum-mean-absolute-error criterion. These criteria are inherently suboptimal with respect to the achievable symbol error rate (SER). In this article, an improved proportionate normalized minimum-SER (IPNMSER) algorithm is proposed for adaptive turbo equalization in deep-sea vertical acoustic communications. The proposed algorithm utilizes the minimum-SER (MSER) criterion to derive the equalizer update equations, aiming to minimize the system's SER directly. In addition, because the deep-sea vertical channel has a sparse structure, which leads to a sparse equalizer, a sparsity-aware proportionate-type approach is therefore incorporated into the framework of the MSER criterion to achieve faster convergence. To investigate the effectiveness of the proposed algorithm, we conducted a deep-sea vertical acoustic-communication experiment in the Challenger Deep of the Mariana Trench. The results demonstrated that the proposed IPNMSER algorithm can outperform a conventional normalized MSER algorithm and other well-known proportionate-type algorithms, achieving error-free detection for all data blocks over a vertical communication range of approximately 10500 m.

中文翻译：

---

马里亚纳海沟水下声学通信的稀疏感知自适应涡轮均衡

潜水器和水面舰艇之间可靠的声学通信在深海勘探中起着至关重要的作用。自适应涡轮均衡可以有效对抗水声信道的选择性衰落，从而成为单载波深海垂直声通信的使能技术之一。现有的自适应涡轮均衡器设计通常基于最小均方误差标准或最小均值绝对误差标准。这些标准就可实现的符号错误率 (SER) 而言本质上是次优的。在本文中，提出了一种改进的比例归一化最小SER（IPNMSER）算法，用于深海垂直声通信中的自适应涡轮均衡。所提出的算法利用最小SER (MSER) 准则推导出均衡器更新方程，旨在直接最小化系统的SER。此外，由于深海垂直通道具有稀疏结构，导致稀疏均衡器，因此将稀疏感知比例型方法纳入MSER标准框架中，以实现更快的收敛。为了研究所提出算法的有效性，我们在马里亚纳海沟的挑战者深处进行了深海垂直声学通信实验。结果表明，所提出的IPNMSER算法可以优于传统的归一化MSER算法和其他著名的比例型算法，