中文摘要：

提出一种新的快速LMS-Newton算法，并将该算法用于水声信道均衡。首先引入归一化因子以提高算法收敛速度，在此基础上，引入一种新的变步长迭代方程进一步降低算法收敛后的稳态误差。通过仿真分析步长迭代方程中6c和声的取值原则及对算法收敛性能的影响，并在2种水声信道环境下，采用2种调制信号将该算法与其他LMS类和RLS类算法的收敛性能进行比较。研究结果表明：该算法实现简单，收敛速度快，稳态误差小；相比XENLMS算法，对于水声信道和调制信号的变化，新算法的适应性或者鲁棒性更强，且随着信道环境和调制信号的复杂化，新算法的收敛性能均与RLS类算法的相当。

英文摘要：

An improved fast LMS-Newton algorithm applied to underwater acoustic equalization was proposed. A new normalized factor was＇introduced to improve the convergence speed, and then a new variable step-size iterative equation was applied to reduce steady-state error. The adoption principle about various a and fl and their influence on convergence ability of the presented algorithm were analyzed by simulation. After that, the comparison of convergence abilities were carried out among the new algorithm, several LMS class algorithms and several RLS class algorithms by using two modulation signals in two different underwater acoustic channels respectively. The results show that the new algorithm is of simple structure, fast convergence and less steady-state error. The adaptability of the new algorithm is superior to that of XENLMS algorithm when underwater channel and modulation signal change. At the same time, the convergence performance of new algorithm is similar to that of RLS with the complication of channel circumstance and modulation signal.