单载波水声通信均衡技术研究

With the increased interest in ocean observation and exploration, it is imperative to build underwater sensor networks that can be deployed rapidly and flexibly. The high-rate reliable underwater acoustic communication is the key enabling technique to achieve the goal. Single-carrier technique has the potential to meet the future underwater communication requirement, attributed to its high transmission efficiency as well as fast channel-tracking capability. Even though, due to the increased demand on the rate and reliability, existing single-carrier direct adaptive equalization (DAE) faces serious challenges, and theoretical and technical breakthroughs are needed. This study proposes to explore two schemes to improve the equalization technique. First, the sparsity of the equalizer is utilized to improve the convergence and performance as well reduce the complexity of the DAE. To achieve this goal, the sparse adaptive algorithm attributed to the combination of proportionate updating and sparse norm regularization, and the dynamic compressed sensing (DCS) technique will be investigated. Second, the channel estimation based equalization (CEE) generally outperforming the DAE in performance and robustness, will be studied. To address the bottleneck in the complexity and the channel knowledge utilization of the conventional CEE, new mechanisms like the Kalman filtering can be used to achieve low-complexity high-performance equalization. Last, the equalizers are endowed the soft-input soft-output capability so as to implement the turbo equalization, which enables robust performance even under low signal-to-noise ratio conditions.

随着海洋探测需求的不断增长，构建快速灵活的水下传感器网络势在必行，而高速率可靠移动水下通信是实现这一目标的关键。单载波技术具有传输效率高、对信道变化跟踪能力强等优点，具备满足未来水下通信要求的潜力。尽管如此，速率及可靠性要求的提高使得现有单载波直接自适应均衡技术面临严峻的挑战，迫切需要理论和技术上的突破。本项目提出两个突破方案：一方面，探索利用均衡器的稀疏性提高自适应均衡收敛速度和性能同时降低计算复杂度，拟采用结合比例调节与稀疏范数正则化的稀疏自适应滤波技术以及动态压缩感知技术；另一方面，探索性能和鲁棒性优于直接自适应均衡的基于信道估计的均衡，针对此类方案在计算复杂度和信道信息利用方面的瓶颈，拟采用新的机制如卡尔曼滤波获得低复杂度高性能的均衡。最后，赋予均衡器软输入软输出能力实现Turbo均衡，在低信噪比条件下也能获得鲁棒性能。

单载波水声通信技术具有传输效率高、对信道变化跟踪能力强等优点,具备满足未来水下通信要求的潜力。尽管如此，速率及可靠性要求的不断提高使得现有单载波直接自适应均衡技术面临严峻挑战。本项目拟通过改进现有自适应均衡方案及探索具有实用条件的基于信道估计的均衡方案来获得理论和技术上的突破。主要研究内容分为三方面：直接自适应均衡技术、基于信道估计的均衡技术以及Turbo 均衡技术。取得的重要结果如下：（1）对稀疏系统逆系统的稀疏特性进行深入研究，在此基础上提出均衡器广义稀疏性的观点，为改进现有自适应均衡技术奠定理论基础；（2）提出稀疏自适应均衡技术，利用均衡器的广义稀疏性实现同时降低复杂度并提高性能。具体地，采用门限法和动态压缩感知算法实现均衡器重要抽头选择并通过稀疏自适应滤波算法对重要抽头系数值进行跟踪，设计了SZA-IPNLMS、SpAdOMP-APA及SP-IPAPA等一系列稀疏自适应均衡方案；（3）提出RLS类型稀疏自适应滤波算法PRLS及l0/l1-PRLS，并在理论上证明其在稀疏系统下优于标准RLS算法。进一步设计了算法的快速实现并用于水声通信信道估计及均衡；（4）首次将卡尔曼滤波技术用于水声通信信道均衡，试验数据表明基于信道估计的卡尔曼均衡性能优于经典的自适应均衡；（5）提出期望最大（EM）等效噪声功率估计方案，提高了向量近似消息传递（VAMP）信道均衡方案的鲁棒性和性能。（6）提出EM-VAMP及稀疏自适应Turbo均衡方案，对其收敛性及性能进行理论分析及试验验证。关键数据方面：相比于传统的自适应均衡，所提出的稀疏自适应均衡方案能够在提高性能的同时有效降低复杂度，海试数据结果表明最好的情况下可以节约高达80%的计算量。总之，本研究解答了学术界对均衡器稀疏性的疑问，为低复杂度高性能稀疏自适应均衡器设计奠定基础，提出了稀疏类RLS自适应滤波算法并作理论性能分析，证实基于信道估计的均衡技术优于自适应均衡技术，具有重要的科学意义。