We consider adaptive detection of a signal of interest when two sets of training samples are available, one sharing the same covariance matrix as the data under test, the other set being mismatched. The approach proposed in this letter is to whiten both the data under test and the matched training samples using the sample covariance matrix of the mismatched training samples. The distribution of the whitened data is then derived and subsequently the generalized likelihood ratio test is obtained. Numerical simulations show that it performs well and is rather robust.

中文翻译：

---

当一些训练样本发生协方差不匹配时，使用白化数据进行自适应检测

当有两组训练样本可用时，我们考虑对感兴趣的信号进行自适应检测，一组与被测数据共享相同的协方差矩阵，另一组不匹配。这封信中提出的方法是使用不匹配训练样本的样本协方差矩阵对被测数据和匹配训练样本进行白化。然后导出白化数据的分布，随后获得广义似然比检验。数值模拟表明它表现良好，并且相当稳健。