Sequential - System Factorization

The success of high-level synthesis methods in reducing design time and formal veri cation methods in reducing design errors in digital VLSI circuits have opened the way to system-level synthesis and veri cation. Derivation is a form of formal veri cation that deals with correct-by-construction reasoning. A set of equivalence preserving transformations are used to derive an implementation from a speci cation. A key step in derivation is to impose an architectural structure on a behavioral speci cation by factoring functional behavior into abstract components [40]. These system factorization transformations impose a naive model for synchronization and data communication between components in a system. The thesis of this work is that system factorization must be generalized to support arbitrary interaction protocols between components in a system for derivational methodology to be useful in system-level design. This dissertation develops a general transformation to decompose a sequential system description into interacting sequential components with non-trivial interaction protocols. Behavior fragments from a system description are encapsulated into a sequential component. Sequential decomposition is used to construct an implementation of interacting sequential components from a higher level of speci cation, using a description of the interaction of a component with its environment as a parameter. Interface speci cation language (ISL) is introduced to describe the synchronous interaction of a machine with its environment. The complement of a machine speci ed in ISL describes the behavior of the machine's environment. Decomposition of sequential components is accomplished by encapsulating parts of a system at the algorithm, process or operation level of granularity into an abstract component with a speci-