Hardware/Software Partitioning Based on Dynamic Combination of Genetic Algorithm and Ant Algorithm

Abstract

Genetic algorithm can do colony global searching quickly and stochastically, but can’t efficiently get to optimal results, since it slows down when solving to certain scope. On the other hand, ant algorithm gets to optimal results efficiently, but lacks initial pheromone at the beginning. To solve the hardware/software bi-partitioning problem in embedded system and system-on-a-chip design, the authors put forward a new algorithm based on dynamic combination of genetic algorithm and ant algorithm. The basic idea is: (1) using genetic algorithm to generate preliminary partitioning results, converting them into initial pheromone distribution for ant algorithm, and then using ant algorithm to search for optimal partitioning scheme; (2) while running genetic algorithm, dynamically determining the best combination time of genetic algorithm and ant algorithm to avoid too early or too late termination of the genetic algorithm. The algorithm utilizes the advantages of the two algorithms and overcomes their disadvantages, and it introduces a dynamic combination strategy between them. Experimental results show the algorithm excels genetic algorithm and ant algorithm in performance, and it is discovered that the bigger the partitioning problem is concerned, the better the algorithm performs.