| Přispěvatelé: |
Elektroniikan ja tietoliikennetekniikan laitos – Department of Electronics and Communications Engineering, Tietotekniikan laitos – Department of Pervasive Computing, Tieto- ja sähkötekniikan tiedekunta - Faculty of Computing and Electrical Engineering, Tampere University of Technology, Hämäläinen, Timo |
| Popis: |
In the last decade, the enormous development of the semiconductor industry with ever-increasing complexities of digital embedded systems and strong market competition with fast time-to-market and low design cost demands have imposed serious difficulty to a conventional design method. Therefore, there emerges a new design flow named model-based system design, which is based on high-level abstraction models, heavy design automation, and extensive component reuse to increase productivity and satisfy the market pressure. This thesis presents reviews of ten high level academic system design frameworks and tools that have been proposed and implemented recently to support the model based design flow, namely System-on-Chip Environment (SCE), Embedded System Environment (ESE), Metropolis, Daedalus, SystemCoDesigner (SCD), xPilot, GAUT, No-Instruction-Set Computer (NISC), Formal System Design (ForSyDe), and Ptolemy II. These tools are then compared to each other in various aspects comprising objective, technique, implementation and capability. Following that, three design flow frameworks, including ESE, Daedalus, and SystemCoDesigner, are experimented for their real usage, performance and practicality. The frameworks and tools implementing the model-based design flow all show promising results. Modelling tools (ForSyDe, and Ptolemy II) can sufficiently capture a wide range of complicated modern systems, while high-level synthesis tools (xPilot, GAUT, and NISC) produce better design qualities in terms of area, power, and cost in comparison to traditional works. Study cases of design flow frameworks (SCE, ESE, Metropolis, Daedalus, and SCD) show the model-based method significantly reduces developing time as well as facilitates the system design process. However, most of these tools and frameworks are being incomplete, and still under the experimental stage. There still be a lot of works needed until the method can be put into practice. |
