Communication networks are inherently dynamic and the changes are often due to unpredicted causes, for instance, failures of the devices or bulks of user requests. To guarantee the continuation of the services, the providers should keep the typical activities of control and management of the network aligned with respect to these changes. They should handle both the evolution of the network and complexity of the infrastructure, while, actually, most of the existing technologies do not adopt update mechanisms or deal with the problem only for specific categories of networks. We propose a data mining approach to analyze evolving communication data while accounting for the whole network and its parts (devices and connections). The approach is able to detect changes that denote substantial and statistically evident variations in the communication modalities. Changes correspond to variations appearing in the frequent sub-networks discovered from evolving communication data: variations in the frequent sub-networks denote changes occurring in the raw data. We perform experimental evaluation on both real and synthetic networks and provide quantitative and qualitative results.
CITATION STYLE
Impedovo, A., Loglisci, C., Ceci, M., & Malerba, D. (2020). Exploiting Pattern Set Dissimilarity for Detecting Changes in Communication Networks. In Studies in Computational Intelligence (Vol. 880, pp. 137–152). Springer. https://doi.org/10.1007/978-3-030-36617-9_9
Mendeley helps you to discover research relevant for your work.