Abstract: Borehole rescue technology has been widely used as a new rescue technique for the rapid construction of life way. UWB radar can be used to detect life information under non-visual conditions, but the UWB radar echoes are easily affected by the noise environment, background clutter and detecting the movement of a target in the underground, which makes it difficult to obtain effective information about the lives of trapped people. This paper summarizes the current status of UWB radar echo information processing for borehole rescue: the current status of clutter filtering is summarized from three aspects including clutter filtering methods, algorithms and models; the current status of effective feature extraction is summarized from four aspects including time/frequency domain analysis methods, time-frequency analysis methods, digital processing techniques and other processing methods. The main problems with current UWB radar echo information processing in borehole rescue were derived: the filtering technology and filtering type are relatively single and lack of post-filtering information verification, the lack of multiple types of micro-Doppler feature extraction, fewer studies on UWB radar echo information processing in mine disaster environments, and the existing information processing equipment cannot meet the demand of mine borehole rescue. The development trend of UWB radar echo information processing in borehole rescue is proposed: combining time-frequency analysis methods and multi-domain processing for clutter filtering research; using neural network models to study the UWB radar echo components based on borehole rescue and comparing with the surface environment; using ICA algorithm to extract multiple micro-Doppler features, using synchronous squeezed short-time Fourier transform to extract motion features, inter-correlation-based joint multi-distance gate signal for sign extraction algorithm to enhance heartbeat signal extraction; development of multi-disciplinary technical equipment to meet drilling and rescue needs, expand downhole echo data training set, and establish a deep self-learning effective feature database.