Computer networks operate as complex interconnected systems. Many types of data collected by network sensors can be modeled using a graph or hypergraph. For example, network traffic (commonly called NetFlow) can be modeled as a directed graph, while associations between users and the processes they run can be modeled as a hypergraph. In recent work my team has been studying how different topological measurements and their dynamics can be used to understand these relational structures in order to provide computer network situational awareness. I will report on our theoretical advances as well as provide some concrete examples of analysis on test data sets.
Topology of graphs and hypergraphs for computer network data
Emilie Purvine, Pacific Northwest National LaboratoryAuthors: Emilie Purvine
2022 AWM Research Symposium
Women in Computational Topology