Hybrid algorithms for hardware/software partitioning and scheduling on reconfigurable devices

Abstract Hardware/software (HW/SW) partitioning and scheduling are essential to embedded systems. In this paper, a hybrid algorithm derived from Tabu Search (TS) and Simulated Annealing (SA) is proposed for solving the HW/SW partitioning problem. The annealing procedure of SA is employed to accelerate the updating of Tabu tables for task scheduling, and the virtual hardware resource is also set to implement the customized TS. The Earliest-Deadline-First (EDF) strategy is introduced to describe the reconfiguration of FPGA. Moreover, an algorithm combining the Breadth-First-Search (BFS) with Depth-First-Search (DFS) is proposed for HW/SW task scheduling to fit the features of reconfigurable systems. Experimental results show that the improvement over the latest efficient combinatorial algorithm is up to 50%. The proposed virtual hardware expanding technique makes the performance increase up to 97.51% on random graphs. The improvement on task scheduling is by 50% in comparison to popular algorithms cited in this paper.