Understanding networks through induced sub-structures

Abstract In this working paper we document the progress made concerning a new characterisation of directed networks using induced substructures (triads). We introduce the concepts of open and closed edges, which provide a mechanism to study and differentiate between triads. These concepts relate to the overt and covert nature of communication within a triad. We use the concept of edge-overlap between triads to create weights that measure the number triads in which an edge is open or closed. Using Erdos-Renyi graphs, we study the minimum weighted path problem applied to open weights, while varying the network size and edge-density. We find that when edge density is sparse, there is considerable difference between shortest paths and least weighted open paths in terms of path length. In other words, pursuing topologically covert communication within a network induces a cost in terms of number of hops. We outline current plans for the work and present details of the code, including handling of multiple objectives for this analysis.
Authors
  • Lauren Hudson (Cardiff)
  • Liam Turner (Cardiff)
  • Roger Whitaker (Cardiff)
  • Diane Felmlee (PSU)
  • Don Towsley (UMass)
Date Sep-2019
Venue Annual Fall Meeting of the DAIS ITA, 2019