A study of the stability of information and telecommunication networks under conditions of stochastic percolation of nodes

In-depth studies of the topological properties of information and telecommunication networks contribute to the understanding of their functional capabilities, including stability. The study of the stability of complex networks to failures in operation when their components fail is based on modeling by sequentially removing nodes or edges of the network (percolation). The paper presents a comparative analysis of sequential and stochastic variants of percolation of network nodes and statistical estimates of the complex two-criterion network stability coefficient. During the study, methods for calculating the average path length based on graph theory were used. In the statistical analysis of the network stability, we applied the analysis of variance and pairwise comparisons according to the Tukey criterion, based on the provisions of the theory of mathematical statistics. The simulation is performed using the Barabashi–Albert and Erdős–Rényi random graph models. The difference between the method of stochastic percolation and sequential percolation is shown. The performed statistical analysis proved the influence of the factor changing the structure of networks on their stability due to stochastic percolation. The dynamics of network stability reduction under stochastic percolation for different types of networks is shown. It is revealed that in some cases, for example, in networks with high density, the stochastic percolation method is the most preferable one. The study shows the possible options for assessing the stability of networks without a priori knowledge about the type of connections between nodes and with a priori knowledge about the type of connections between nodes. In the former case, knowing the number of network nodes, one can calculate the limit values of stability, in the same way as if the nodes were deleted accidentally. The latter option can be used to calculate the stability of networks that are subject to random node failures, for example, when diagnosing technical systems.