Node Reliability: Approximation, Upper Bounds, and Applications to Network Robustness

TL;DR

This paper introduces efficient Monte Carlo and stochastic methods for approximating node reliability in graphs, providing formulas for specific graph types, analyzing phase transitions, and proposing upper bounds to assess network robustness.

Contribution

It presents novel Monte Carlo and stochastic approximation techniques for node reliability, along with formulas for Erdos-Renyi and Random Geometric graphs, and new upper bounds based on degree distributions.

Findings

Monte Carlo method effectively estimates node reliability

Derived formulas for Erdos-Renyi and Random Geometric graphs

Proposed upper bounds improve reliability assessment

Abstract

This paper discusses the reliability of a graph in which the links are perfectly reliable but the nodes may fail with certain probability p. Calculating graph node reliability is an NP-Hard problem. We introduce an efficient and accurate Monte Carlo method and a stochastic approximation for the node reliability polynomial based solely on the degree distribution. We provide the formulas for the node reliability polynomial of both Erdos-Renyi graphs and Random Geometric graphs. The phase transition in the node reliability of Erdos-Renyi graphs is also discussed. Additionally, we propose two increasingly accurate upper bounds for the node reliability polynomial solely based on the graph's degree distributions. The advantages and disadvantages of these two upper bounds are thoroughly compared. Beyond the computation of node reliability polynomials, we also estimate the number of cut sets and present a solution to the reliability-based network enhancement problem.