A model of multi-criteria optimization of the distribution of information flows in telecommunication networks

Gaipov Konstantin Eduardovich, Krikunov Ilya Leonidovich, Demicheva Alena Alekseevna

In this article, the solution of the problem of two-criteria optimization of the distribution of information flows is considered, the minimum of the total time spent by applications in the buffer of a telecommunications device and the minimum of total losses as a result of buffer overflow act as optimality criteria. Both optimality criteria are mutually exclusive, since an increase in one leads to a decrease in the other. The solution of this problem is shown by the example of a network with two possible routes using the tools of the MatLab mathematical modeling environment. Mathematical modeling was carried out by two methods: the first option using a genetic algorithm and two optimization parameters, the second option involves the use of the method of linear convolution of optimized functions and the use of traditional gradient algorithms for finding the extreme value of the objective function.

Keywords:multiobjective optimization, genetic algorithm, linear convolution algorithm, Pareto front.