*2.1. K-RXD*

The K-RXD, denoted by *<sup>δ</sup>K*−*RXD*(*r*), is specified as follows:

$$\delta\_{\mathbf{K}-\mathbf{R}\mathbf{X}\mathbf{D}}(\mathbf{r}) = (\mathbf{r} - \boldsymbol{\mu})^T \mathbf{K}^{-1} (\mathbf{r} - \boldsymbol{\mu}) \tag{1}$$

where *μ* = (1/*N*) ∑*Ni*=<sup>1</sup> *ri* is the global sample mean and *K* = (1/*N*) ∑*Ni*=<sup>1</sup>(*<sup>r</sup><sup>i</sup>* − *μ*)(*<sup>r</sup>i* − *μ*)*<sup>T</sup>* is the sample data covariance. The form of *δK*−*RXD* in (1) is actually the well know Mahalanobis distance between the data sample being detected and global sample mean. It should be pointed out that the model assumes the data arise from two normal probability density functions with the same covariance matrix but different means.
