Real-time application clustering in wide area networks. (July 2020)

• Record Type: Journal Article
• Authors: Takyi, Kate; Bagga, Amandeep
• Is Part Of: Computers & electrical engineering. Volume 85(2020)
• Publication Date: 2020-07
• Subjects: Machine learning -- Packet loss -- Clustering -- Traffic classification -- Quality of service; Computer engineering -- Periodicals; Electrical engineering -- Periodicals; Electrical engineering -- Data processing -- Periodicals; Ordinateurs -- Conception et construction -- Périodiques; Électrotechnique -- Périodiques; Électrotechnique -- Informatique -- Périodiques; Computer engineering; Electrical engineering; Electrical engineering -- Data processing; Periodicals; Electronic journals; 621.302854
• DOI: 10.1016/j.compeleceng.2020.106691
• Languages: English
• ISSNs: 0045-7906

Abstract

Highlights: Datasets of higher-dimensional space reveal the true performance of a classifier. Clustering in real-time; a better technique for traffic classification in networks. Packet loss impacts inter-arrival time between packets flows in classification. Networks with limited resources have poor quality of serhvice during classification. Abstract: Network traffic classification employing Machine Learning and Statistical approaches have contributed to the understanding of the dynamic nature of traffic. For further improvement, all phases of networks, including when the requirements of the network exceed its current resources, must be considered. With scenarios of networks with low-speed links, fragmentation and loss of packets leading to poor quality of services are highly expected, resulting in few flows being classified at a time with the features extracted. Training a classifier with few features inhibits the overall classification accuracy with real-time traffic traces. We propose a Real-Time Application Clustering (R-TAC) strategy which can classify application flows utilizing the limited flow features extracted. Results from evaluation reveal that our proposed clustering approach performs better in terms of classification accuracy (96.40%) and precision metrics (85–99%) than the existing state of the art methods, and the best classification accuracy when validated with an existing dataset. Graphical abstract: Image, graphical abstract