Abstract The article considers the tasks of the study and analysis some factors danger and intensity of a set of security events in multiservice local and corporate public communication networks. Based on the study, a new approach is proposed for creating a mathematical model danger and intensity of a set security events in multiservice telecommunications networks under the threat of subscriber and network unauthorized access. As a result of the study, analytical formulas were obtained for the formal dependence integrated criteria for the quality of indication as single and multiple security events and hazard factors with a parameter, components of the information security vector and the conditional probability intensity of a set of security events. For a comparative analysis, a numerical calculation was carried out and a graphical dependence of the probability of the degree resistance to security threats in a communication system on the probability danger and the intensity of a set of security events was constructed when using different feedback coefficients. The relevance of using machine learning technology in the tasks of assessing the effectiveness of management based on security events in multiservice communication networks using the k -nearest neighbours (KNN) method, which classifies an object as a class, is considered. Based on the modeling, the metrics for assessing the quality of classification by the KNN algorithm are presented and a graphical dependence True Positive Rate (TPR) and False Positives Rate (FPR) is constructed for comparative analysis.
Ibrahimov et al. (Wed,) studied this question.