A K-means Optimized by Improved Grey Wolf Algorithm Anomaly Detection method for Wireless Sensor Networks

Anomaly detection in wireless sensor networks is crucial for the implementation of tasks such as fault diagnosis, intrusion detection, and event monitoring. Under existing research, a large number of anomaly detection algorithms use supervised or semi-supervised algorithms to solve a specific application scenario problem. However, in wireless sensor network multi-scenario applications, the feature definition for anomaly data is not available a priori. In this paper, an unsupervised algorithm based on K-means is proposed to solve this problem. Since the effect of K-means algorithm is sensitive to the number of clusters and the selection of initial points, it is easy to fall into local optimum. To enhance the reliability of the algorithm, the gray wolf algorithm is used to find the original cluster centers, and then the clustering results are compared with the weighted neighboring clusters to determine whether the edge points or edge clusters are abnormal data. The experimental results show that the accuracy, recall and F1 value of the algorithm are significantly improved compared with K-means algorithm and other algorithms optimizing K-means under different wireless sensor network data sets.