Abstract: It is a hot research topic in the current military field that target detection is executed with many sensors in a networking manner. Target detection is also the fundamental function of a sensor network. However it is difficult for the current mechanism to describe the target discovery event which is based on an output signal threshold of the target. According to the mathematical statistics theory, the problem of target detection is modeled as a statistical inference process in the paper. A target detection mechanism based on the hypothesis testing method is proposed. In this mechanism the testing parameters can be determined by a sample training process, and moreover a data fusion method is adopted for the detection result optimization. The experimental results show that this mechanism can avoid the noise interference in the detection process, and improve the exactness performance of target catch in sensor networks.

Key words: sensor network, target detection, data fusion, statistical hypothesis testing