Abnormal Data Detection Algorithms In Wireless Sensor Networks

As a bridge between subjective perception world and objective physical world,wireless sensor networks(WSN)is a new technology for information acquisition and processing.It is one of the most important supporting technologies of the Internet of Things.WSN has been rapidly developed and widely used,including national defense,military,medical care,environmental monitoring,manufacturing and many other areas.With the continuous expansion of the scale of Wireless Sensor Networks,anomaly detection in sensor networks becomes more and more critical.Especially for some emergencies,such as chemical leakage,fire and so on,timely warning and emergency measures are often needed.The main work of this paper is as follows:1.Due to the insufficient number of sensors,it is impossible to detect the abnormality of sensor data by mining the spatial correlation between sensor data and its neighbors.We propose an anomaly detection method based on time series data.Using k normal data collected by sensors,a reference interval is established to judge whether the next moment's data is abnormal.The data interval difference degree method based on distance is used to determine the source of anomalies.The experimental results show that the detection rate of the abnormal data in the sensor network is above 98% and the false alarm rate remains below 0.5%,which have some certain utilities and commonality.2.For the existing spatio-temporal correlation-based anomaly detection algorithm,the detection effect is poor under the high failure rate sensor network.We propose a fault-tolerant anomaly detection algorithm for sensor networks based on temporalspatial correlation.In the time correlation part,the probabilities of possible events or faults are obtained,and the status of nodes is preliminarily determined.In the space correlation part,sensor networks are divided into different neighborhoods,and different detection methods are used to achieve fault tolerance.The results of experiment show that under the condition that 45% of sensor nodes are failing,the hit rate of event detection remains at about 97% and the false negative rate of events is above 92%.