Application of estimation of distribution algorithm in HW/SW partition

摘要

Hardware/software (HW/SW) partitioning problem is NP hard problem. An improved algorithm based on estimation of distribution algorithms is proposed to solve HW/SW partitioning problem. Estimation of distribution algorithm is good in globe search but poor in local search and may suffer from “premature convergence” beacause of diversity loss. The improved algorithm strengthens the local searching ability by cloning and searching the elite solutions and improves the diversity loss by correcting the probability model. Numerical simulation is carried out and compared with existing algorithm, the results show the effectiveness of the improved estimation of distribution algorithm in solving HW/SW partitioning problem.