A Discrete Multi-Objective Optimization Method for Hardware/Software Partitioning Problem Based on Cuckoo Search and Elite Strategy

摘要

This paper attempts to provide a desirable solution to hardware/software partitioning of the embedded system. For this purpose, the author developed a discrete multi-objective optimization method based on the cuckoo search (CS) algorithm (MODCS) and the elite strategy of stratification and congestion degree comparison. Then, the MODCS was compared with two other typical simulation algorithms. The results show that the MODCS is superior to typical optimization algorithms in terms of many indices, including diversity, stability and generational distance (GD) of optimal solution. The superiority is positively correlated with the number of modules. The findings shed new light on the bionic optimization of hardware/software partitioning.