Target Detection Based On Data Fusion Under Nonideal Channels

The problem of target detection based on data fusion is the problem that the sensors of network capture the information of phenomenon interested and transmit them to fusion center using data fusion theory to obtain final decision, and has caused considerable concern. But most researches focused on the situation of ideal channels, or just studied one part of non-ideal channels incompletely. Thus our focus of this thesis is the target detection of non-ideal channels.Before this, it is necessary to establish the model of data fusion system which has considered the situation of non-ideal channels including noises, attenuation and the error probability of sensors. It is also necessary to establish the sensors model and non-ideal channels model. This non-ideal channels model includes the error probability which produce by sensors under non-ideal channels and we should choose a suitable quantitative measure to accomplish the fusion system. Then add noises and attenuation in this non-ideal model to finish the non-ideal data fusion system.The classic fusion using the non-ideal fusion system is studied, such as counting rule and likelihood ratio test. We derive the equations of global false alarm probability and detection probability, verify the formulas and do a comparison between these fusion rules according to the simulation experiment. And the characteristic of global false alarm probability and detection probability by varying SNR.The scan statistics of fusion rule under non-ideal channels are also the focus of this thesis. We derive the equations of global false alarm probability of grid distribution and uniform distribution because of the difference of ideal channels. And many simulation experiments is done for verifying the formulas, comparing the scan statistics and counting rule and analyzing the characteristic of global false alarm probability and detection probability with varying SNR.Due to the complexity of the problem under non-ideal channels, we simplify the problem by adding a threshold in the fusion system, and find out the optimal threshold. Through the simulation experiments we obtain the difference of these two systems.