Open AccessA Fuzzy Clustering Based Anomaly Node Detection Method for Publish/Subscribe Distributed Systems
Author(s) -
Defeng Wang,
Zhuowei Shen,
Wenjun Wu
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1813/1/012046
Subject(s) - computer science , data mining , anomaly detection , node (physics) , publication , cluster analysis , anomaly (physics) , fuzzy clustering , fuzzy logic , artificial intelligence , engineering , physics , structural engineering , advertising , business , condensed matter physics
Timely and accurate understanding of the node running status can effectively control the propagation of faults in publish/subscribe distributed systems, which is of great significance to ensure the reliable operation of applications. A method of anomaly node detection based on fuzzy k-means clustering algorithm is proposed. Compared with the traditional K-means algorithm, this algorithm introduces fuzzy membership matrix to realize fuzzy clustering, and uses the idea of local reachability density to improve the selection method of cluster centers. Experimental results show that this method can effectively detect anomaly node in publish / subscribe distributed systems with higher accuracy, recall and F-measure than traditional K-means algorithm. The precision of publish anomaly detection and network anomaly detection is improved by 10.53% and 38.6% respectively.