[Paper Review] Mutually connected component of network of networks.

Ginestra Bianconi, S. N. Dorogovt︠s︡ev+1 more|arXiv (Cornell University)|Feb 2, 2014

Distributed systems and fault tolerance5 citations

TL;DR

This paper investigates the emergence of a giant mutually connected component in networks of networks where nodes are interdependent only on their replica nodes across layers. It proves that when all nodes in one layer depend on nodes in the same other layer, the mutually connected component becomes independent of the network of networks' topology and equals that of a fully connected multiplex network built from the same layers.

ABSTRACT

We describe the emergence of the giant mutually connected component in networks of networks in which each node has a single replica node in any layer and can be interdependent only on its replica nodes in the interdependent layers. We prove that if in these networks, all the nodes of one network (layer) are interdependent on the nodes of the same other interconnected layer, then, remarkably, the mutually connected component does not depend on the topology of the network of networks. This component coincides with the mutual component of the fully connected network of networks constructed from the same set of layers, i.e., a multiplex network.

Motivation & Objective

• To understand the structural conditions under which a giant mutually connected component emerges in networks of networks.
• To investigate the role of interdependence topology in determining the size and stability of the mutually connected component.
• To identify when the mutual connectivity is independent of the network of networks' architecture.
• To prove that interdependence between replica nodes across layers leads to a universal mutual component size, regardless of network topology.

Proposed method

• Modeling networks of networks as multiplex structures with interdependencies restricted to replica nodes across layers.
• Defining the mutual component as the set of nodes that remain functional after cascading failures due to interdependence.
• Proving analytically that if all nodes in one layer are interdependent on nodes in a single other layer, the mutual component size is invariant under changes in the network of networks' topology.
• Using graph-theoretic arguments to show that the mutual component equals that of a fully connected network of networks constructed from the same layers.
• Demonstrating that the result holds regardless of the individual layer's topology, provided the interdependence structure is symmetric and replica-based.
• Applying the proof to multiplex networks to establish equivalence between general interdependence structures and the fully connected case.

Experimental results

Research questions

• RQ1 Under what conditions does the mutual component in a network of networks become independent of the network of networks' topology?
• RQ2 How does restricting interdependence to replica nodes affect the size and robustness of the mutually connected component?
• RQ3 Can the mutual component size in a general network of networks be bounded or predicted by the fully connected case?
• RQ4 What structural properties of interdependence lead to universal behavior in mutual connectivity?

Key findings

• The mutually connected component size is independent of the topology of the network of networks when all nodes in one layer are interdependent on nodes in a single other layer.
• The mutual component in such systems is identical to that of a fully connected network of networks built from the same layers.
• The result holds regardless of the individual layer's topology, including random, scale-free, or regular structures.
• Interdependence restricted to replica nodes leads to a universal behavior where the mutual component is maximized and topology-invariant.
• The proof establishes that the mutual component is determined solely by the interdependence structure and not by the underlying network architecture.
• This topology-invariance implies that robustness analysis can be simplified by focusing on the fully connected case without loss of generality.