快速收敛的低复杂度稀疏水声信道估计方法*

针对水声信道的稀疏特性,本文基于集员L1范数约束比例调节仿射投影算法,在不增加稳态误差的前提下,首先采用一种鲁棒集员滤波思想,通过设定动态误差门限加速算法收敛速度;另外针对其矩阵计算复杂度高问题,利用历史的比例调节矩阵优化信道更新方程,使得某些过程矩阵可通过递归方法更新,从矩阵运算角度降低了计算复杂度。湖试和海试数据处理结果表明,面对弱时变和强时变信道时,此方法相对已有稀疏水声信道估计方法略微降低稳态误差的同时具有更快的收敛速度,并从矩阵运算和迭代次数两个方面降低了计算复杂度。

Low complexity sparse underwater acoustic channel estimation method with fast convergence speed

In view of the sparsity of underwater acoustic channel (UAC), this paper proposes a sparse UAC estimation method with lower complexity based on set membership L1-norm constrained improved proportionate affine projection algorithm (SM-L1-IPAPA). Firstly, we borrow a robust SM (RSM) filtering idea to set a dynamic error threshold, which accelerates the convergence speed without increasing steady-state error. Then the channel update equation is optimized by using the historical proportionate matrix, so that some process matrices can be updated by recursive method, which reduces computational complexity from the perspective of matrix operation. The results of lake trial and sea trial data processing show that this method has faster convergence speed and slight lower steady- state error than other sparse UAC estimation methods when facing both weak and strong time-varying channels, and it can reduce the computational complexity from two aspects of matrix operation and iteration times.