Target detection and classification by UWB communication signal based on fourth-order cumulants

Experimental study is undertaken with a novel method of the feasibility of ultra-wideband (UWB) communication system to concealed obstacles detection and classification. Totally different from traditional method using UWB radar echoes, the recognition of target can be achieved by received UWB-IR signals from the UWB communication system, which can transmit the information and sense the surrounding environment simultaneously. In this paper, a fourth-order statistic method is proposed to extract features that are representative of the target types from the received signals. The extracted features are based on the 1-D diagonal slice of fourth-order cumulant. Then, we use support vector machine (SVM) to realize the obstacle identification. The detection performance is compared with that of feature extraction method based on statistical characteristics of received signal [1, 2]. According to the experiment based on real data collected by the received signals of UWB communication, the results indicates that the detection method based on fourth-order cumulant is better than that based on statistical characteristics and is particularly effective when used in low signal-to-noise ratio scenarios.