Thomas Huining Feng

EECS Department, University of California, Berkeley

Technical Report No. UCB/EECS-2009-77

May 20, 2009

http://www2.eecs.berkeley.edu/Pubs/TechRpts/2009/EECS-2009-77.pdf

In this dissertation I present my work on a flexible and efficient model transformation technique, which includes a transformation engine and a formal and expressive control language.

I define a basic transformation with a transformation rule, which describes the relationship between input models and the result of the transformation. I invent a syntax for transformation rules, which is close to the modeling language that designers use to create the input models. This makes it convenient for the designers to use the technology. The semantics of transformation rules is defined based on graph transformation theory. To improve flexibility over traditional approaches, I introduce extensions to allow complex transformation rules with variable structures to be specified with compact higher-order descriptions.

A transformation workflow consists of basic transformations controlled in a control language. Existing control languages include finite state machines and dataflow diagrams. Realizing their limitations in expressiveness and efficiency, I create the Ptera (Ptolemy event relationship actor) model of computation based on event graphs. I show that Ptera is an appropriate control language for transformation workflows. It allows transformation tasks to be composed hierarchically. The event queue and the notion of model time enable the scheduling of future tasks. By using a shared variable to represent the model under transformation, the overhead caused by communication with transient data packages is eliminated.

The research has a wide range of practical applications. I demonstrate the power of the idea with examples including model optimization and automatic model construction.

Advisors: Edward A. Lee

---

BibTeX citation:

@phdthesis{Feng:EECS-2009-77,
    Author= {Feng, Thomas Huining},
    Title= {Model Transformation with Hierarchical Discrete-Event Control},
    School= {EECS Department, University of California, Berkeley},
    Year= {2009},
    Month= {May},
    Url= {http://www2.eecs.berkeley.edu/Pubs/TechRpts/2009/EECS-2009-77.html},
    Number= {UCB/EECS-2009-77},
    Abstract= {In this dissertation I present my work on a flexible and efficient model transformation technique, which includes a transformation engine and a formal and expressive control language.

I define a basic transformation with a transformation rule, which describes the relationship between input models and the result of the transformation. I invent a syntax for transformation rules, which is close to the modeling language that designers use to create the input models. This makes it convenient for the designers to use the technology. The semantics of transformation rules is defined based on graph transformation theory. To improve flexibility over traditional approaches, I introduce extensions to allow complex transformation rules with variable structures to be specified with compact higher-order descriptions.

A transformation workflow consists of basic transformations controlled in a control language. Existing control languages include finite state machines and dataflow diagrams. Realizing their limitations in expressiveness and efficiency, I create the Ptera (Ptolemy event relationship actor) model of computation based on event graphs. I show that Ptera is an appropriate control language for transformation workflows. It allows transformation tasks to be composed hierarchically. The event queue and the notion of model time enable the scheduling of future tasks. By using a shared variable to represent the model under transformation, the overhead caused by communication with transient data packages is eliminated.

The research has a wide range of practical applications. I demonstrate the power of the idea with examples including model optimization and automatic model construction.},
}

EndNote citation:

%0 Thesis
%A Feng, Thomas Huining 
%T Model Transformation with Hierarchical Discrete-Event Control
%I EECS Department, University of California, Berkeley
%D 2009
%8 May 20
%@ UCB/EECS-2009-77
%U http://www2.eecs.berkeley.edu/Pubs/TechRpts/2009/EECS-2009-77.html
%F Feng:EECS-2009-77