A persistent homological analysis of network data flow malfunctions. (24th August 2017)

• Record Type: Journal Article
• Authors: Scoville, Nicholas A; Yegnesh, Karthik
• Is Part Of: Journal of complex networks. Volume 5:Number 6(2017)
• Page Start: 884
• Page End: 892
• Publication Date: 2017-08-24
• Subjects: hierarchical recurrent networks -- data flow -- persistent homology -- network packet delivery systems; Numerical analysis -- Periodicals; Computer networks -- Periodicals; Social networks -- Periodicals; 518.05
• DOI: 10.1093/comnet/cnx038
• Languages: English
• ISSNs: 2051-1310

Abstract

Abstract: Persistent homology has recently emerged as a powerful technique in topological data analysis for analysing the emergence and disappearance of topological features throughout a filtered space, shown via persistence diagrams. In this article, we develop an application of ideas from the theory of persistent homology and persistence diagrams to the study of data flow malfunctions in networks with a certain hierarchical structure. In particular, we formulate an algorithmic construction of persistence diagrams that parameterize network data flow errors, thus enabling novel applications of statistical methods that are traditionally used to assess the stability of persistence diagrams corresponding to homological data to the study of data flow malfunctions. We conclude with an application to network packet delivery systems.