A modular design space exploration framework for multiprocessor real-time systems
Abstract:
Embedded system designers often face a large number of design alternatives when designing complex systems. A designer must select an alternative which satisfies application constraints (e.g. timing requirements) while optimizing system level objectives such as overall energy consumption. The size of design space is often very large giving rise to the need for systematic Design Space Exploration (DSE) methods. In this paper we address the DSE problem for real-time applications that belong to two different domains: (i) streaming applications modeled using the synchronous dataflow graphs; (ii) feedback control tasks modeled using the periodic task model. We consider a heterogeneous multiprocessor platform in which processors communicate through a predictable bus architecture. We present our DSE tool in which the DSE problem is modeled as a constraint satisfaction problem, and it is solved using a constraint programming solver. This approach provides a modular framework in which different constraints such as deadline, throughput and energy consumption can easily be plugged depending on the system being designed.
Año de publicación:
2016
Keywords:
- constraint programming
- Synchronous Dataflow
- Multiprocessors
- Real-time systems
- Design space exploration
- Periodic Tasks
Fuente:
scopusTipo de documento:
Conference Object
Estado:
Acceso restringido
Áreas de conocimiento:
- Software
Áreas temáticas de Dewey:
- Ciencias de la computación
- Programación informática, programas, datos, seguridad
- Física aplicada
Objetivos de Desarrollo Sostenible:
- ODS 9: Industria, innovación e infraestructura
- ODS 7: Energía asequible y no contaminante
- ODS 8: Trabajo decente y crecimiento económico