基于最大相关熵准则的恒模盲均衡算法

A Constant Modulus Blind Equalization Algorithm Based on Maximum Correntropy Criterion

  • 摘要: 针对恒模算法(constant modulus algorithm, CMA)在脉冲噪声环境下性能退化的问题,本文基于最大相关熵准则(maximum correntropy criterion, MCC)对恒模算法中基于最小均方误差(mean square error, MSE)准则的代价函数进行修正,推导出适用于脉冲噪声环境的基于MCC准则的恒模盲均衡算法(MCC_CMA)。该算法利用通信信号的恒模特性,首先得到发送信号与均衡器输出信号模值的误差信号,再通过使模值误差信号的相关熵最大来获得其迭代误差调节项,避免了传统高阶统计量算法在脉冲噪声环境下性能退化的问题。对高斯噪声以及α-稳定分布和混合高斯分布两种脉冲噪声环境下的信道均衡问题的仿真实验表明,相对于经典的自适应恒模盲均衡算法,MCC_CMA算法不依赖噪声的先验知识就能获得较快的收敛速度、较低的剩余码间干扰和误码率,并且在不同脉冲强度的脉冲噪声环境下都能够得到较好的均衡结果,表明MCC_CMA算法具有很好的鲁棒性。

     

    Abstract: In this paper, the mean square error (MSE) criterion used in the In this paper, the mean square error (MSE) criterion used in the constant modulus algorithm (CMA) is modified by the maximum correntropy criterion (MCC) to solve the problem of performance degradation of the CMA blind equalization algorithm under impulse noise environment, and then the a constant modulus blind equalization algorithm based on MCC is derived, which is referenced as MCC_CMA. By utilizing the constant modulus property of the communication signals, the proposed algorithm first obtains the modulus difference signal between the transmitted signal and the equalizer output signal, and then the iterative error adjustment term is achieved by maximizing the correntropy of the modululs difference signal, thus the problem of the performance degradation of the traditional high-order statistics based algorithms under impulse noise environment is avoided. Under Gaussian noise environment and two kinds of impulse noise environment: α-stable distribution and mixture Gaussian distribution, the simulation experiments of channel equalization problem show that the MCC_CMA algorithm not only can obtain faster convergence speed, lower residual intersymbol interference and bit error ratio without relying on the prior knowledge of noise, comparing with the classical adaptive constant modulus blind equalization algorithm; but also has good robustness, that is, it can get good equalization results in impulse noise environments with different impulsiveness.

     

/

返回文章
返回