An ultra-wideband impulse radar was studied for the detection of buried life in coal mines. An improved Empirical Mode Decomposition (EMD) method based on a cross-correlation filter was proposed for reduction of multipath and noise interference. Multipath interference was first removed by cross-corre- lation filtering. Then the delays of each pulse in every echo were summed. An EMD algorithm was used for noise reduction for the total delay of each echo. The corresponding EMD results of every echo were then summed and averaged. Finally, evidence for the existence of buried life and their position were obtained from amplitude-frequency curves of the averaged EMD results. Detailed simulation experi- ments are presented to validate the effectiveness of this proposed method. The experimental results show that this method can efficiently eliminate multipath interference and reduce noise interference in echoes, which makes detection and location of buried life in coal mines more accurate. An ultra-wideband impulse radar was studied for the detection of buried life in coal mines. An improved Empirical Mode Decomposition (EMD) method based on a cross-correlation filter was proposed for reduction of multipath and noise interference. Multipath interference was first removed by Then the delays of each pulse in every echo were summed. An EMD algorithm was used for noise reduction for the total delay of each echo. The corresponding EMD results of every echo were then summed and averaged. Finally, evidence for the existence of buried life and their position were obtained from amplitude-frequency curves of the averaged EMD results. Detailed simulation experi- ments are presented to validate the effectiveness of this proposed method. The experimental results show that this method can efficiently eliminate multipath interference and reduce noise interference in echoes, which makes detection and location of buried life in coal mines more accurate.