脉冲噪声非线性变换设计的研究综述

Review of nonlinear transformation design for impulsive noise

由于脉冲噪声下的最优检测要求非线性运算,信号检测常采用零记忆非线性变换接匹配滤波的检测结构。本文系统地综述了脉冲噪声下的非线性变换设计,梳理了国内外研究在噪声模型、非线性函数和设计方法3个方面的工作,并总结研究路线和规律。首先,脉冲噪声模型常采用对称α稳定分布、Class A分布、高斯混合分布及其他混合分布。其次,以拖尾函数为特征,非线性函数模式可分为传统削波/置零、多区域组合拖尾、单参数特定拖尾、单参数非分段函数以及双参数可变拖尾。然后,在设计方法上,分析近似和正态变换的思路比较直观,而最大信噪比和最大效能准则与检测性能有直接联系。之后,总结了常见研究路线和主要研究成果的共性规律。最后,探讨了噪声分布未知和效能寻优简化的问题,展望未来可能的研究点。

Because of the optimal detection under impulsive noise requiring nonlinear operation,the detection structure of zero memory nonlinear transformation followed by the matched filter is often used.In this paper,the nonlinear transformation design under impulsive noise is systematically reviewed,and the research work in the noise model,the nonlinear function and the design method at home and abroad is summarized.Firstly,the impulsive noise model usually uses the symmetricαstable distribution,the Class A distribution,the Gaussian mixture distribution and other mixture distributions.Secondly,characterized by the tailing function,the nonlinear function model can be divided into traditional clipping/blanking,multi-region combined tailing,single parameter-based specific tailing or non-piecewise function,and bi-parameter variable tailing.Thirdly,in the design method,the idea of analysis-based approximation and normalization-based transformation is intuitive,and the criteria of maximum signal to noise ratio and maximum efficacy are directly related to the detection performance.Then,the common research routes and the common laws of the main research results are summarized.Finally,the problems of unknown noise distribution and efficacy optimization are discussed,and the possible research points in the future are prospected.