基于四阶累积量极性迭代的自适应动态谱线增强快速算法

NEW FAST ALGORITHMS OF FOURTH-ORDER CUMULANT BASED SIGN ITERATIVE ADAPTIVE DYNAMIC LINE ENHANCER

针对最小均方 (LMS)算法抑制高斯噪声效果差和高阶累积量计算量大等问题 ,研究了四阶累积量递推算法与计算量的关系 ,提出了基于四阶累积量对角切片极性迭代的自适应动态谱线增强快速算法。并用水下目标辐射谱线的实测数据进行了动态仿真。理论分析和仿真结果表明 :该算法的计算量比基于四阶累积量对角切片非极性迭代的自适应谱线增强算法减少约 5 0 % ;该算法抑制高斯噪声的性能优于基于LMS算法的自适应谱线增强算法。因此 。

To reduce computational complexity in cumulant estimates and to overcome the ineffectiveness in suppressing Gaussian noice of LMS algorithms, the paper analyzed the relationship of cumulant recursive algorithm with computational complexity and developed new fast algorithms of fourth-order cumulant diagonal slice- based sign iterative dynamic line enhancer. In these algorithms, coefficients of adaptive filters are updated by sign iterative algorithm. Performance of the proposed algorithms in suppressing Gaussian noise was much better than that of the LMS-based adaptive line enhancer and cumulant-based non-sign iterative algorithm. The computational complexity of cumulant-based sign iterative algorithms is only half that of the cumulant-based non-sign iterative algorithm. Computer simulation with underwater target- radiated data was also included. Its results testify the effectiveness of these algorithms and show that sign iterative algorithm of non-input signal is the best among all sign iterative algorithms in suppressing Gaussian noise.