Sequential Bayes Factor Testing: A New Framework for Decision Fusion
Juan Parras, Santiago Zazo · 2019
We propose using a Bayes factor sequential hypothesis test for the decision fusion problem in wireless sensor networks, in which several sensors send a report to a fusion center so that a global decision is taken. This problem is frequently modeled in current literature as a hypothesis test from Bernoulli samples. We propose using a sequential composite hypothesis test based on Bayes Factor using Beta distributions as prior distributions. We obtain closed form expressions for the distributions, which allows us to develop a very efficient algorithm to implement our approach. When we validate our approach via simulations, we observe that, when compared to the common counting rule, our algorithm provides a lower average error and requires a smaller number of samples to make a decision.