System Level Design of Embedded Systems

University Of Kansas Center For Research Inc, Lawrence KS

Investigators

Abstract

A. Project Summary The objective of this research is to develop and demonstrate capabilities for analyzing complex interactions between heterogeneous embedded systems specification domains. Three research activities are being undertaken: (i) development of domain models for important embedded systems requirements domains; (ii) development of semantics for combining heterogeneous embedded systems specifications; and (iii) development of techniques for predictive analysis of composed specifications. The results of this research are being evaluated using problems from the Ambient Computing Environments (ACE) research center at The University of Kansas / ITTC. The facet-based approach being investigated involves modeling and integrating heterogeneous specifications using domain specific semantic systems. Each system model, or facet, is defined by extending a domain. A domain provides a vocabulary and model of computation for describing a system from one perspective. Facets use domains as modeling semantics and are composed to describe components and systems. Specifically, individual models are composed to define: (i) multiple aspects of the same component; and (ii) heterogeneous, multi-component systems. Projection functions and morphisms support specification transformation and composition using category and institution theoretic semantics. Special projection functions, called interactions, model how and when models from one domain interact with models from another. Interactions support integrated component analysis and modeling communication between heterogeneous components. Developing domain models, domain interaction models, and prototype analysis tools demonstrates the effectiveness of the facet-based approach for modeling embedded systems. The first and second research tasks encompass the development of domain and domain interaction models for defining embedded components. Both functional and performance constraint models are being supported. Domain models are being developed to represent basic models of computation and vocabularies for specification. Interaction models consisting of projection functions and specification morphisms are being defined to move information between domains. The third research task is the development of software systems for analyzing specifications. Two complimentary approaches are being explored: (i) formal analysis; and (ii) simulation. Formal analysis is achieved by mapping domain definitions to existing formal analysis environments. Specification simulation is achieved by mapping domain definitions to communicating, heterogeneous simulation environments. Challenge problems from the NSF and DARPA sponsored Ambient Computing Environments (ACE) research infrastructure at ITTC are being used to evaluate ongoing research. ACE provides an environment where computing moves seamlessly with individuals within a physical environment. ACE is implemented as a collection of highly mobile, heterogeneous embedded systems that perform tasks including personal data management, control of robotic devices, and various sensor related tasks The results of this research have widespread impact on the research and industrial communities. The Rosetta systems level specification language is being used for domain models, interaction definitions and system specifications for embedded systems. Rosetta is being standardized by the Accellera CAD standards organization as a next-generation systems level design language. Successful results from this research activity are being disseminated as a part of the base Rosetta domain system and are being integrated into the collection of standard Rosetta domain models.