Physical‐layer security threats have evolved from malicious attacks in wireless systems, due to their furtive nature, make wireless communication systems vulnerable. In this work we proposed a centralized modulated wideband converter (C-MWC) combined with classifier detector based Mahalanobis distance ( MD S ) based classical estimator S and robust distance ( RD MCD ) based MCD estimator. The received signal at each radio receiver in each channel pass by different steps to realize sub-Nyquist sampling rate. Every receiver gives minimum sampling. All compressed observations from each channel are collected in compressed data matrix, which is considered directly as the input of the proposed M D S - RD MCD classifier-based ROC curve in the level of fusion center (FC). The performance evaluation is performed in terms of anomaly detection rate-based threshold value of each distance. By employing one of the machine learning (ML) techniques MD S - RD MCD classifier based PCA using ROC curve, the performance of this new proposed system shows good.