The Filttering Algorithm Research Based On UWB Target Detection And Recognition

UWB is a short-range wireless communication technology with strong resolution, detection and anti-jamming capability. UWB radar has been widely used in the transportation detection, bridge detection, medical detection, etc. However traditional UWB radar is used for target detection and identification by analyzing the echo signals. It can not content the requirement of accomplishing communication and target detection at the same time which is significant in individual combat and intrusion detection. In this paper, we propose a novel target detection and recognition method based on extreme learning machine which is different from the traditional UWB radar to content the above requirement.In response to above research topics, after analyzes a lot of simulation and actual measurement, we found that when there are different types of objects in the communication channel, the received signal will exhibits different characteristics. If we use these different characteristics of the received signal and extract these characteristic parameters in different environments, we can establish correspondence between the characteristic parameters of receive signals and the different target environments. This technology can content the requirement of accomplishing communication and target detection at the same time. There exist researches have use the self-learning ability of neural network algorithm to identify the type of target objects in the UWB communication channel. However, the above UWB target recognition researches use feedforward neural network and learning speed of this neural network is generally slow. And the above researches are carried out under condition without noise, it is difficult to meet the actual demand. This paper proposes a new UWB target identification method based on extreme learning machine. This method extracts the characteristic parameters of the received signals instead of the echo signals, and employ the extreme learning machine to identify target. According to Matlab simulation result, the new method is quite fast and effective in target identification. To make this method more effective in a low SNR environment, this paper use noise suppression for UWB signal so that this method can be used in a long distance or bad channel environment with a good recognition rate. Considering the empirical mode decomposition filtering has been widely used in wideband radar techniques, this method will use the empirical mode decomposition filtering for UWB signal in low SNR condition. filtering and de-noising, making this study UWB target identification methods can be applied with a longer distance. This makes this new UWB target identification method can be applied in a long distance environment.