Data fusion utilization for distributed target detection with tree topology
Junhai Luo, Jing Ni, Qi Wu, Liying Fan · 2017
Multi-sensor fusion has been extensively studied in information fusion field, and the distributed target detection is one of the most important applications in the multiple sensor detection theories. In this paper, a data fusion algorithm for target detection is proposed based on tree topology combined with the orderly full binary tree and we discuss the optimal threshold fusion rule problem. Different from the conventional tree topology, the sensors in our topology are well ordered and the sensors with highest signal amplitude are selected as the fusion center. Moreover, we derive a fusion decision rule which takes the channel noise into consideration based on this topology. By introducing the probability of error, we prove that with equal prior probability there is a concave function of the likelihood ratio threshold used in the sensor decision rule. In the case of minimizing the probability of error, the optimal threshold of each level can be obtained. Finally, the distributed detection performance of tree topology is analyzed and comparisons with other topologies are drawn. These results show that the detection performance improves as the level decreases and our fusion rule can achieve a higher probability of detection.