Model and Parametric Optimization of Proactive Protection of the Email Service from Network Intelligence

Abstract

The purpose of the study: to increase the security of the e-mail service of information systems in the conditions of network intelligence. Methods used: methods of mathematical statistics, random processes research, mathematical programming, heuristic optimization algorithms were used to achieve the research goal. The result of the study: a semi-Markov model of proactive protection of the e-mail service from network intelligence has been developed, which allows determining the probabilistic and temporal characteristics of the process of transmitting e-mail messages. Based on traffic analysis, statistical hypotheses about the types of distributions of the time of occurrence of events, under the influence of which the system under study evolves in a discrete set of states, are verified, point and interval estimates of the values of the parameters of these distributions are performed. The solution of the system of linear integral Volterra equations with integral kernels of the difference type was carried out using numerical methods of the Laplace transform. The problem of vector optimization is solved to determine the optimal parameters for configuring e-mail messages, allowing to maximize the effectiveness of the protection of the e-mail service, the robustness of the simulated system and minimize overhead costs under appropriate restrictions. The extremum of the objective functions was found using a bioinspired particle swarm algorithm. Scalarization of Pareto-optimal estimates was carried out using the ideal point method. Scientific novelty: it consists in developing a model and solving the problem of optimizing the parameters of the e-mail service in the conditions of network intelligence using the mathematical apparatus of semi-Markov processes, numerical methods of Laplace transformation, parametric evaluation of statistical characteristics of the model, scalarization of a multi-criteria optimization problem by the ideal point method and search for the extremum of objective functions using the particle swarm algorithm.