Implementation of Specification Conjunction and Domain Interaction in Rosetta

System Level Design is nowadays a complex process due to heterogeinity of domains of components in a single system. The language standard, Rosetta, was proposed as a means to help in such designs. An important feature in Rosetta is known as domain interaction and involves the analysis of interaction between models from domain to domain. The resulting interation models can then be used to ensure correctness of the system at the design level. The goal of my doctoral dissertation is therefore to implement the semantics of domain interactions in Rosetta and to apply it in the specification of real world examples. One of the challenges in systems engineering is the ability to combine semantically different models in a predictive way. Many failures of systems have been attributed to the semantic gap between elements of different domains. This problem of semantic gap between domains becomes more and more obvious as more systems include heterogeneous components. Modern systems combine electronic and software control to mechanical elements, or combine discrete with continuous as in hybrid systems. In some cases, the heterogeneity of a system does not call for domain interaction analysis as domain models may be orthogonal to one another. However, many cases exist where information from one domain does influence a model in another domain are more and more frequent. Therefore, interest in modeling of domain interactions will obviously increase. The most common approach to modeling multiple domains has been to find a single common semantics for the domains comprising a system. The main drawback of this approach is that the common semantics may be too general or favor one domain over the others. Another approach, as proposed with Ptolemy, is to use a coordination framework to provide an open architecture with a well-defined interface that allows a designer to use any model of computation. A third approach as used in Viewpoints is to provide a framework that allows structuring, organizing and managing of the different perspectives of designers. Although Ptolemy and Viewpoints allow hetereogeneous domain modeling as does Rosetta, these two approaches do not allow analysis of how one domain-specific model can affect another model in a different domain. Furthermore, they do not allow reuse of a model by projection into different domains. Modeling of interactions is based on the ability to move information from one domain to another. The interaction describes how to derive properties that hold in one domain from properties that hold in another domain. This concept of modeling truth or properties from domain to domain is derived from Institution Theory. Since the semantics of domains is well defined in Rosetta, my work in implementing the domains and their interactions consists of first developing an appropriate syntax and semantics, and then increasing the functionality of the Rosetta tool suite to take care of the additional semantics. The definition of a syntax is not as simple as it may seem due to the fact that the syntax must be explicit enough to define the interaction, without being too complex so as to discourage designers from using it. To achieve the second part of my research goal, I must model real world heterogeneous systems. The feasibility of such modeling will help evaluate the approach of using domain interaction models. I hope to be able to show that through modeling domain interaction, a better knowledge will be available on how an overall system is going to work once all the components are put together. This should help in avoiding failures of system because of a misinterpretation of information by components of different domains.