|Advisor:||Edward A. Lee|
The modeling and design of complex embedded systems require tools that support multiple models of computation (MoCs). Timed MoCs, including discrete event (DE), continuous time, mixed signal, and hybrid systems, are among the most widely used. The goal of this research is to create a framework that supports the design of tools for these timed MoCs.
The tagged signal model (TSM) is the foundation of this framework. From studying the timed MoCs with TSM, a common tag system is identified, which gives structure to the signals in these MoCs. A generic simulation scheme that is applicable to the timed MoCs is proposed. At each time tag in the simulation, a set of equations on the current values of all signals are derived from the process functions and solved. This common scheme provides valuable guidance in implementing correct simulations of heterogeneous timed models. Another constituent of the framework is a component model for operational specification of processes. It is designed to minimize the interface complexity between components and their operating environment. A version of the DE domain in Ptolemy II is implemented using the proposed framework and component model. Current work includes implementing other timed MoCs with the framework, and developing a behavioral type system for the component model.
 E. A. Lee and A. Sangiovanni-Vincentelli, "A Framework for Comparing Models of Computation," IEEE Trans. CAD, Vol. 17, No. 12, December 1998.
Last updated 11/01/04