Ingredients for successful system level automation & design methodology. Support for multiple models of computation, heterogeneous behavioral hierarchy, model-driven validation, and service-oriented tool integration environment

This dissertation addresses the problem of making system level design (SLD) methodology based on SystemC more useful to the complex embedded system design community by presenting a number of ingredients currently absent in the existing SLD methodologies and frameworks. The complexity of embedded systems have been increasing at a rapid rate due to proliferation of desired functionality of such systems (e.g., cell phones, game consoles, hand-held devices, etc., are providing more features every few months), and the device technology still riding the curve predicted by Moore's law. Design methodology is shifting slowly towards system level design (also called electronic system level (ESL)). A number of SLD languages and supporting frameworks are being proposed. SystemC is positioned as being one of the dominant SLD languages. The various design automation tool vendors are proposing frameworks for supporting SystemC-based design methodologies. We believe that compared to the necessity and potential of ESL, the success of the frameworks have been limited due to lack of support for a number of facilities and features in the languages and tool environments. This dissertation proposes, formulates, and provides proof of concept demonstrations of a number of ingredients that we have identified as essential for efficient and productive use of SystemC-based tools and techniques. These are heterogeneity in the form of multiple models of computation, behavioral hierarchy in addition to structural hierarchy, model-driven validation for SystemC designs and a service-oriented tool integration environment. In particular, we define syntactic extensions to the SystemC language, semantic modifications, and simulation algorithms, precise semantics for model based validation etc. For each of these we provide reference implementation for further experimentation on the utility of these extensions.;This project received support from NSF, SRC, and Bluespec.