Ant Colony Optimization for Coherent Synthesis of Computer System
Mieczysław Drabowski · InTech eBooks · 2011
The goal of high-level synthesis of computer systems is to find an optimum solution satisfying the requirements and constraints enforced by the given specification of the system. The following criteria of optimality are usually considered: costs of system implementation, its operating speed, power consumption and dependability. A specification describing a computer system may be provided as a set of interactive tasks (processes, functions). The partition of the functions between hardware and software is the basic problem of synthesis. Such partition is significant, because every computer system must be realized as result of hardware implementation for its certain tasks. In the synthesis methods so far, the software and hardware parts were developed separately and then connected in process the co-called co-synthesis, which increased the costs and decreased the quality and reliability of the final product. The resources distribution is to specify, what hardware and software are in system and to allocate theirs to specific tasks, before designing execution details. The problems of tasks scheduling are one of the most significant issues occurring at the procedure synthesis of operating systems responsible for controlling the distribution of tasks and resources in computer systems. 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 model and approach are new and original proposals allowing synergic design of hardware and software for performing operations of the computer system. This is approach, which we called a par-synthesis (coherent co-synthesis). This research shows the results selected of computational experiments for different instances of system par-synthesis problems proving the correctness of the coherent synthesis concept and shows the methods solving these problems. Due to the fact that synthesis problems and their optimizations are NP-complete we suggest meta-heuristic approach: Ant Colony Optimization. Coherent co-synthesis of computer systems, as well as synergic design methodology their structures and scheduling procedures may have practical application in developing the tools for automatic aided for rapid prototyping of such systems.