Empirical Distribution-Based Event Detection in Wireless Sensor Networks: An Approach Based on Evidence Theory

Event detection is a central issue in wireless sensor networks. In this paper, we propose a novel event-detection scheme which utilizes empirical distribution and the combination method of evidence theory. Unlike conventional methods where the sensor uses energy detection to determine the presence or absence of an event, our scheme uses a goodness-of-fit (GOF) test to measure the distance between the observed data and the empirical distributions of both the presence and absence hypotheses. Multiple types of such two-sided GOF tests are combined to create a well-adapted detector using evidence theory. The simulation results show that the proposed detector is more accurate than conventional detectors in different kinds of noisy environments.