Result-Oriented System-Level Modeling
for Efficient Design of Embedded Systems

NSF CAREER Award #0747523

Principal Investigator: Rainer Dömer
 
Abstract
Activities
Findings
Publications
Abstract

We are surrounded by embedded computing systems, ranging from video-enabled mobile phones over real-time automotive applications to reliable medical devices. Just as the quality of an architectural blue-print determines the quality of the resulting building, the model of an embedded system is the key to its successful implementation.

This project moves research and education on embedded system design forward in the area of system-level specification and modeling. While traditional work largely has focused on simulation and synthesis from a given system model, this project addresses the creation and optimization of system models for effective use in existing design processes. The results of this project are directly applicable to established system design flows in industry and fit well into existing and new courses in computer engineering education.

This project optimizes the modeling of embedded systems by use of four novel techniques. First, it advances a new model of computation, named ConcurrenC, which refines the generic capabilities of common C-based system-level description languages. Second, the creation of the system model is automated by computer-aided re-coding that derives an executable model directly from reference code. Third, the efficiency of the model is optimized using Result-Oriented Modeling (ROM), which, in contrast to traditional Transaction-Level Modeling (TLM), offers gains in simulation speed of multiple orders of magnitude and highest accuracy at the same time. Fourth, this project investigates TLM of computation, an area where TLM has not been applied before.

This NSF CAREER project has been proven highly successful. Significant results include (A) the study of concurrent models of computation and in particular break-through advances in parallel system simulation, and (B) the development of computer-aided recoding techniques for model optimization and refinement.

Activities


Findings

The major findings in this project concentrate on (a) ConcurrenC model of computation and (b) Recoding of embedded system models.

(a) In the context of ConcurrenC, we have created a novel model of computation which features fast simulation algorithms and parallel, multi-core, and distributed simulators and corresponding compilers. Together, these results contribute to significantly faster validation of embedded systems.

(b) In the context of Recoding, we have designed novel methods and tools for computer-aided modeling and transformation of design models. Our experimental results on video codec applications demonstrate feasibility and promise significant reduction in design specification and modeling time.

For more details on our findings, please refer to the reports and publications below.


Publications

Overall, this NSF CAREER project has contributed to a total of 16 technical reports, 3 presentations and tutorials, 1 book chapter, 15 conference papers, 4 journal articles, and 1 book.

Moreover, this work has been recognized with a DATE Best Paper Award 2014 and an EDAA Outstanding Dissertation Award 2013.

  • W. Chen, X. Han, C. Chang, G. Liu, R. Dömer:
    "Out-of-Order Parallel Discrete Event Simulation for Transaction Level Models",
    IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, accepted for publication, July 15, 2014.


Acknowledgement and Disclaimer:
This material is based upon work supported by the National Science Foundation Grant No. #0747523.
Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.

08/07/14 R. Doemer (doemer@uci.edu).