Bіlous R. The method is an automated system dynamic load balancing in the network based on genetic algorithms.

The thesis proposed a method for constructing a dynamic adaptive load balancing system in a computer network based on genetic algorithms for solving the multi-criterion optimization problem. A genetic algorithm for finding optimal paths in the graph on several criteria is proposed. Solution encoding technique and peculiarities of genetic operations of crossover and mutation are formed. The described algorithm can be used for constructing multi-path routing protocol in computer networks. Network load balancing problem is formalized as a problem of optimizing the parameters of routing protocol and routing tables in order to minimize network links bandwidth consumption. A genetic algorithm to solve this problem proposed, the chromosome encoding technique and approaches of employing genetic operations are presented: heuristic search for common mutations, adaptive probabilities of crossover application operations and mutation. Implementation technique for modern computer networks using existing routing protocols and packet switching mechanisms of the proposed balancing systems described.