Measurement-based statistical fusion methods for distributed sensor networks

Chow [8], who showed that a weighted majority fuser is optimal in combining outputs from pattern recognizers under statistical independence conditions. Furthermore, the weights of the majority fuser can be derived in closed form in terms of the individual detection probabilities of patten recognizers. A simpler version of this problem has been studied extensively in political economy models (for example, see [3] for an overview). Under the Condorcet jury model of 1786, the simple majority rule has been studied in combining the 1-0 probabilistic decisions of a group of N statistically independent members.