Glowworm Swarm Optimization Algorithm for Computer Equipment Partitioning

The paper is devoted to the solution of one of the main microelectronics design problem in the design stage - a partitioning of ECE elements. The quality of solution directly affects the subsequent interconnection traceability and, as a result, the performance of the final product. The partitioning problem belongs to NP-complete combinatorial logic problems, which means that an exact solution can be obtained only by exhaustive search. The paper proposes a new heuristic population partitioning algorithm, based on modeling the glowworm swarm behavior. The main idea of the algorithm is the principle of choosing a pair of alternative solutions based on modeling the luciferin luminescence, produced by glowworms during the mating period. The paper presents the partition problem formulation, and also an optimization criterion is proposed. The glowworm swarm behavior in wildlife is described in detail. The scheme of the optimization algorithm, based on the glowworms behavior, is constructed, and the example of new alternative solutions generation is considered. To estimate the time complexity of the developed algorithm the authors has developed software product. Series of experimental researches has conducted, so results obtained by the ECE partitioning algorithm based on a glowworm swarm behavior and the well-known hMetis and MLPart algorithms are compared.