Research And Application Of Ultra Wideband Target Identification Based On Selected Bispectra

It is a very important practical application in the strong clutter environment to identify a certain kind of target, such as in the environment with foliage, blocked by buildings and so on. It is also a very valuable research topic for the military and civilian applications. However, in this kind of scenes, environment noise, multipath fading and occluding objects are making it difficult to get analytic solutions, and the channel modeling is also very difficult to be widely applied due to the complex environment.Aiming at this problem, this paper uses selected bispectra algorithm to extract target features from the signal waveforms transmitted around the Wireless Sensor Networks, and then uses different classifiers to learn and recognize the target features. Basing on the selected bispectra and the classifiers, several target identification algorithms with application potentials have been promoted. In this paper, a dataset of waveforms transmitted between UWB-IR transceivers in the real world foliage environment with different targets has been built, including no target, human target, wood board target and iron box target. Moreover, the promoted algorithms are tested and compared with each other using this dataset.Besides, a prototype system with an experimental platform exploiting UWB-IR transceivers and a laptop has been built, based on the GUI design in Matlab. The system adopts the selected bispectra and ELM target identification algorithm, which is promoted in the paper with a potential of online training and online testing. This promising platform is an initial attempt to apply the promoted algorithms in real world application in the future.