Event detection framework for wireless sensor networks considering data anomaly

Event detection is a topic widely discussed in the wireless sensor network (WSN) community. The goal of this process is to identify when the collected data represents an event occurrence. The lack of uniformity when approaching this problem, such as the modeling of the data collected by sensors, is an obstacle to the comparison among the different proposals. In this work, we present an extension of the Diffuse framework for scenarios regarding event detection, while considering settings with inaccurate and fault-susceptible measurements made by non-ideal sensing devices in a noisy environment. Based on that framework, we compare a new proposed algorithm with a method based on the literature, obtaining hit rates of event detection above 72% in settings with 30% of sensor faults.