Software-Hardware Partitioning Strategy Using Hybrid Genetic and Tabu Search

One of the most crucial steps in the design of embedded systems is deciding which components of the system should be implemented in software and which ones in hardware. Inspired by genetic algorithm (GA) and tabu search (TS), this paper puts forward a hybrid strategy (GATS) to solve the software-hardware partitioning problem in embedded system. The main frame of GATS is provided by genetic algorithm and the tabu search is taken as the mutation operator. Here the tabu search is used for the solution space in the process of mutation. And the results show that GATS has multiple starting-points, strong mountain-climbing ability and memory function instead of inferior mountain-climbing ability of GA and the single starting-point feature of TS. The experimental results indicate that GATS is superior to the single GA and TS in terms of both required time and system cost, which testify the effectiveness of GATS and produce better portioning results.