Fusion of likelihood ratios in distributed Bayesian detection
H. Deliç, Demetrios Kazakos · 2002
A discrete-time Bayesian detection model is considered in which sensors collect data records, and it is assumed that any two sensors are statistically independent. A binary hypothesis is tested, and it is assumed that a state-variable random process model represents the two hypotheses for each sensor. The optimum Bayesian detector is constructed by the computation of the local likelihood ratios at each sensor in real time and the subsequent transmission of these ratios to the fusion center. A study is made of the performance for various parameters; in particular, the effects of the data size and the number of sensors on the performance of the detection scheme.>