Abstract
In wireless sensor network-based event detection approaches, when the decision is taken based on the measurements of sensors, sensor-fault and noise-related measurement error should be taken into account. Using Bayesian approach to form a judgment is problematic without additional information or assumptions (for example, the difficulty of knowing prior probabilities in practice). In making the final decision using the majority decision rule (as well as the k-out-of-n rule), measurement of every sensor is considered as fully reliable. However, due to sensor fault and environmental noise, the preciseness of all measurements may not be guaranteed in real-life applications. This paper presents a Dempster-Shafer theory of evidence-based structural health monitoring protocol using wireless sensor networks that overcome these limitations. Our proposal effectively discounts the unreliable observer's (sensor's) measurements. Extensive simulations show significant improvement in terms of detection accuracy as compared to other well-known approaches.