A Hybrid Genetic Algorithm for Hardware-Software Synthesis of Heterogeneous Parallel Embedded Systems

摘要

The paper includes a proposal of a new algorithm for hardware-software synthesis of heterogeneous parallel embedded systems. Optimal scheduling of tasks, optimal partition of resources and allocation tasks and resources are fundamental problems in this algorithm. In the former synthesis methods, software and hardware parts have been developed separately and then connected in the process of so-called concurrent synthesis. The objective of this research is to present the concept of coherent approach to the problem of system synthesis, i.e. a combined solution to task scheduling and resource partition problems. The approach is new and original and allowing synergic design of hardware and also software controlling the performance of a computer system. This is an approach which we call a coherent parallel synthesis. This paper shows the algorithm, based on genetic method assisted with simulated annealing strategy and shows the results of selected representative computational experiments into different instances of system synthesis problems which prove the correctness of the coherent synthesis concept and indicate methods solving these problems.