Network Identifiability: Advances in Separating Systems and Networking Applications

Viviana Arrigoni, Novella Bartolini

Abstract: This letter discusses the parallelism between the concept of network identifiability introduced by Boolean Network Tomography (BNT), and the theory of separating systems, highlighting applications of interest to research in networking. This letter evidences how recent results of BNT have direct implications to the formulation of new bounds to the size of separating systems over finite sets. Grounding on these theoretical results, we provide an efficient algorithm for the design of separating systems that meet the bound tightly. We extend the proposed results, bounds and algorithm, to networking applications, including network failure assessment, robust network design and compressed sensing.