State-based encoding of large asynchronous controllers
State encoding is one of the fundamental problems in the synthesis of asynchronous controllers. The requirement for a correct hazard-free implementation imposes severe constraints on the way encoding signals can be inserted in the specification of a controller. Even though some specification formali...
| Autores: | , |
|---|---|
| Tipo de recurso: | artículo |
| Fecha de publicación: | 2018 |
| País: | España |
| Institución: | Universitat Politècnica de Catalunya (UPC) |
| Repositorio: | UPCommons. Portal del coneixement obert de la UPC |
| Idioma: | inglés |
| OAI Identifier: | oai:upcommons.upc.edu:2117/123738 |
| Acceso en línea: | https://hdl.handle.net/2117/123738 https://dx.doi.org/10.1109/ACCESS.2018.2872678 |
| Access Level: | acceso abierto |
| Palabra clave: | Logic circuits Asynchronous circuits Circuit synthesis State encoding Circuits lògics Circuits asíncrons Àrees temàtiques de la UPC::Informàtica::Informàtica teòrica |
| Sumario: | State encoding is one of the fundamental problems in the synthesis of asynchronous controllers. The requirement for a correct hazard-free implementation imposes severe constraints on the way encoding signals can be inserted in the specification of a controller. Even though some specification formalisms, such as Burst-mode machines or Signal Transition Graphs, enable to specify behaviors at the event level, the state encoding methods that provide the best good-quality solutions work at the state level. This imposes a severe limitation on the size of the controllers that can be handled by these methods. This paper proposes a method to solve the encoding problem for large asynchronous controllers using statebased methods. It is based on an iterative process of projection and re-composition that reduces the size specification by hiding signals, partially solves the encoding problem at the state level and re-composes the original specification using a synchronous product. The process iterates until all encoding conflicts have been solved. The method is proved to preserve the behavior of the specification (branching bisimilarity) and shown to be capable of providing good-quality solutions for controllers of more than 100 signals and 106 states. |
|---|