| System level design of an embedded computing system involves a multi-step process to refine the system from an abstract specification to an actual implementation by defining and modeling the system at various levels of abstraction. System level design supports evaluating and optimizing the system early in design exploration.;Embedded computing systems may consist of multiple processing elements, memories, I/O devices, sensors, and actors. The selection of processing elements includes instruction-set processors and custom hardware units, such as application specific integrated circuit (ASIC) and field programmable gate array (FPGA). Real-time operating systems (RTOS) have been used in embedded systems as an industry standard for years and can offer embedded systems the characteristics such as concurrency and time constraints. Some of the existing system level design languages, such as SpecC, provide the capability to model an embedded system including an RTOS for a single processor. However, there is a need to develop a distributed RTOS modeling mechanism as part of the system level design methodology due to the increasing number of processing elements in systems and to embedded platforms having multiple processors. A distributed RTOS (DRTOS) provides services such as multiprocessor tasks scheduling, interprocess communication, synchronization, and distributed mutual exclusion, etc.;In this thesis, we develop a DRTOS model as the extension of the existing SpecC single RTOS model to provide basic functionalities of a DRTOS implementation, and present the refinement methodology for using our DRTOS model during system level synthesis. The DRTOS model and refinement process are demonstrated in the SpecC SCE environment. The capabilities and limitations of the DRTOS modeling approach are presented. |
