Network node grouping algorithm and evaluation model based on clustering and Bayesian classifier

Abstract

Identifying key nodes of the network helps design network protection policies and improves network robustness and reliability. This paper proposes a network node grouping algorithm and a grouping performance evaluation based on clustering and Bayesian classifiers. Before grouping, this paper defines a distance suitable for measuring the degree of location difference between network nodes. The definition of the attribute and the distance among nodes, the Clustyes algorithm can get a better grouping effect. The experimental results show that the network group obtained by Clustyes is better than the random grouping algorithm. The Clustyes algorithm can solve the problem that the random grouping algorithm has certain randomness. At the same time, it can solve the problem that the benchmarking node affects the grouping effect of the grouping algorithm. It can be seen from the experimental results that the network node group obtained by Clustyes has better packet communication performance, packet stability, and close physical distance, which can provide a better application for multi-group network. Compared with the related research results, the proposed improved method can not only accurately search for the optimal sequence in the large solution space, but also has stronger versatility and higher accuracy than other methods.