抑制α稳定噪声的改进常数模盲均衡算法

Improving AECCMA(Adaptive Error-Constrained Constant Modulus Algorithm) for Blind Equalization to Make it Suitable in α-Stable Noise

为了改善α稳定噪声环境中盲均衡器的性能,文章提出一种改进常数模盲均衡算法(MAEC-CMA)。对均衡器输入信号进行软限幅,并对原自适应误差受限常数模盲均衡算法(AECCMA)的误差信号进行非线性变换,有效地抑制了α稳定噪声的影响。采用2种水声信道,在高斯噪声与α稳定噪声的情况下对算法进行了计算机仿真。结果表明:在高斯噪声环境中,MAECCMA算法与AECCMA算法具有相似的性能,相对于常数模(CMA)算法和归一化最小平均绝对偏差(NLMAD)算法它具有较快的收敛速度;在α稳定噪声环境中,文中提出的MAECCMA算法性能优于其它3种算法。

Aim.The AECCMA algorithm proposed by S.Choi et al in Ref.5 is better in convergence than CMA algorithm in Gaussian noise.We now propose a modified AECCMA（MAECCMA） algorithm that can also suppress α-stable noise while retaining its good convergence in Gaussian noise.Section 1 of the full paper briefs α-stable noise.Section 2 briefs the normalized least mean absolute deviation（NLMAD） algorithm.Section 3 explains our MAECCMA algorithm;it discusses how to modify the AECCMA algorithm and then presents the details of modification in two steps：（1） adding to the software the capability of limiting the amplitude of the equalizer input,（2） transforming nonlinearly the error signals of the AECCMA algorithm to suppress the α-stable noise.Section 4 presents the computer simulation of our MAECCMA algorithm respectively in Gaussian noise and in α-stable noise,using two underwater acoustic channels.The simulation results,given in Figs.1 through 4,show preliminarily that：（1） the performance of our MAECCMA algorithm is almost the same as that of AECCMA algorithm in Gaussian noise;（2） both algorithms have faster convergence speed than the CMA algorithm and the NLMAD algorithm in Gaussian noise;（3） the performance of our MAECCMA algorithm is the best among the above four algorithms in α-stable noise;thus it can steadily suppress the residual inter-symbol interference caused by the multi-path effect of an underwater acoustic channel.