Intelligent Network Security Monitoring Based on Optimum-Path Forest Clustering

IEEE Network. 2019. Vol. 33. No. 2. P. 126–131.

Guimarães R. R., Passos L., Filho R. H., Hugo C. de Albuquerque V., Rodrigues J. ., Komarov M. M., Papa J. P.

Distinguishing outliers from normal data in wireless sensor networks has been a big challenge in the anomaly detection domain, mostly due to the nature of the anomalies, such as software or hardware failures, reading errors or malicious attacks, just to name a few. In this article, we introduce an anomaly detection-based OPF classifier in the aforementioned context. The results are compared against one-class support vector machines and multivariate Gaussian distribution. Additionally, we also propose to employ meta-heuristic optimization techniques to finetune the OPF classifier in the context of anomaly detection in wireless sensor networks.

Keywords: обучение, training, оптимизация, optimization, беспроводные сенсорные сети, wireless sensor networks, forestry, temperature measurement, измерение температуры, prototypes, anomaly detection, выявление аномалий, кластеризация методом леса, прототипы