中文摘要：

针对浅海水声信道固有的随机时一空一频变、高噪、强多径等特性及变化的多径时延扩展，在变步长最小均方（LMS）平行滤波器组（PFB—LMS）算法的基础上提出了一种新的水声通信自适应均衡算法。该算法将变阶数和变步长的调整结合起来，降低了算法对迭代步长和均衡器阶数的敏感度。仿真结果表明，新算法在参数适应性方面优于传统LMS及PFB-LMS算法。

英文摘要：

The shallow water acoustic channel is characterized as a complex time, space and frequencyvariant channel with several negative factors, e.g. narrow band, high ambient noise, multipath distortion and polytropic muhipath time delay which pose serious difficulty for the underwater acoustic communication. A novel two-parameter adjustable least mean square （LMS） equalization algorithm is presented based on the classic parallel filter banks LMS （PFB-LMS） algorithm. The new algorithm enables hybrid adjustment of step-size and tap-length so that the sensitivity of step-size and tap-length parameter selections is alleviated. Simulation results show that the new algorithm outperforms the traditional LMS and PFB-LMS algorithms in parameter robustness under time varying channels.