基于复高斯模型的样本缺失窄带雷达信号重构算法

该文提出一种针对窄带雷达信号存在样本缺失情况下的信号重构算法。由于窄带雷达体制下，目标回波近似服从复高斯分布。在这一前提下，首先建立描述样本缺失观测信号与未知完整信号间关系的概率模型，然后根据贝叶斯准则推导出在给定样本缺失观测信号条件下完整信号的后验分布，最后利用期望最大(Expectation Maximization, EM)算法得到模型中参数的最大似然估计，进而得到完整信号的重构。该方法的优势是只需利用样本缺失观测信号就可以重构出未知的完整信号，除了复高斯分布的假设，不需要其他任何样本信息和先验假设帮助参数学习。基于实测数据的实验结果和与现有算法的比较结果表明该方法能够获得较好的重构性能。

Reconstruction Method for Narrow-band Radar Returns with Missing Samples Based on Complex Gaussian Model

This paper proposes a new signal reconstruction method for the signals with missing samples obtained by narrow-band radar. For the narrow-band radar system, the target echoes can be assumed to follow the complex Gaussian distribution. Based on this precondition, first the probabilistic model between the observed signal with missing samples and the unknown complete signal is formulated. Then the posterior distribution of the complete signal is obtained via the Bayes' theorem. Finally, the maximum likelihood estimation of the model parameters is obtained with the Expectation Maximization (EM) algorithm and the reconstruction of the complete signal can be obtained. The advantage of the method is that the reconstruction of the complete signal only using the observed signal with missing samples based on the complex Gaussian distribution assumption, while no other signal and prior information are needed in the parameter learning process. Experiments based on the measured data and the comparation results with other state-of-the-art approaches show that the proposed method can achieve good reconstruction performance.