A Technique of Resource Allocation for Computationally Hard Optimization Problems Solving in Distributed Heterogeneous Dynamic Environments

In this paper a technique of resource allocation for computationally hard optimization problems solving in distributed heterogeneous dynamic environments is presented and described. Despite the wide range of various metaheuristic approaches to the optimization problem solving, including distributed ones, there is a lack of detailed research which spotlights the tight connection between the optimization problem solution quality, the method of metaheuristic algorithms instances distribution and resource allocation for algorithms instances runs. The technique proposed is based on the metaheuristics peculiarity to improve the solution quality with the increase of objective function calls number. The current research focuses on the resource allocation for the procedures of metaheuristic blocks forming and distribution and on the optimization problem processing on a heterogeneous set of computing nodes. Simulation results demonstrate the positive effect of developed technique usage, which consists in considerable optimization problem solution improvement.