Tabu-Search Algorithm for Optimization of Elastic Optical Network Based Distributed Computing Systems

In the paper we consider expanded routing and spectrum assignment problem in distributed computing system based on elastic optical network. Computational tasks generated by clients at network nodes are divided into subtasks and sent to data centers to provide computations. The problem consists in simultaneously dividing computational tasks into subtasks and for each subtask assigning data center, route and spectrum in elastic optical network. The considered traffic type is an extension of classical anycast traffic, since one on few data centers may be the target (destination) of each demand. In the paper, we formulate a new optimization problem as an ILP model and propose meta-heuristic algorithm based on tabu search method. Then, we present computational experiments for exemplifying network, reporting the quality of heuristic solutions an characteristics of considered optimization problem.