10. CONCLUSION

DCharts are a new formalism that combines the benefits of statecharts and DEVS for the design of complex physical systems and software systems. It has the following advantages:

• A visual syntax is designed for the DCharts formalism. There is graphical representation for every entity or feature of DCharts.

• DCharts are powerful. They support statecharts-like hierarchical model design with variables. Variables help keep infinite and innumerable states. Recursive DCharts models are much more expressive than statecharts and DEVS in that they are able to specify infinite states and transitions.

• DCharts are modular. Importation is also known as tight coupling. Submodels are copied to the inside of a state of the importing model. The behavior of the submodels may not be modified by the importing model, except that macros can be redefined as parameters.

  Connections between multiple models via ports are also known as loose coupling. In that case, a model may affect other models only by means of messages sent via the established connections.

• DCharts are independent of simulation strategies. Though its definition only addresses real-time simulation, it is shown that virtual-time simulation can be easily accomplished by means of a clock component.

• DCharts are highly practical. SVM is a simulator for DCharts, which supports a complete semantics of DCharts 1.0. Many of the algorithms implemented in SVM can be reused by other simulators or applications. SVM itself is reusable (for example, by AToM$^3$ and SCC).

  SCC is a code synthesizer for DCharts capable of generating source code in multiple target languages. The synthesized code is efficient and suitable for practical purposes.

• Besides these, non-recursive DCharts can be transformed into statecharts with variables or DEVS models. Statecharts and DEVS models can also be transformed into DCharts. This property is useful for model simulation and model checking.

Three types of syntaxes are discussed: abstract syntax, graphical/visual syntax and textual syntax. The mathematical syntax provides a means by which DCharts models can be formally specified. The graphical syntax represents DCharts models visually, which is much more easily understood by human beings. The textual syntax is accepted and processed by computer programs, while at the same time designers can still easily write DCharts models with the textual syntax. A few extensions to the basic syntax are proposed by the textual syntax. Those extensions allow designers to specify their models with more flexibility. They are supported by SVM and SCC.

The future work on DCharts includes:

• Do more research on model checking and verification of DCharts. There are two possible approaches:
  • Build tools that directly check DCharts models, or verify them by means of simulations.

  • Transform DCharts models into models in other well-studied formalisms, and check/verify the new models with the tools available for those formalisms.

• The performance of some DCharts features (such as history) in the code generated by SCC must be improved. Above this, an important hurdle to cross is the need for a target-language-independent action language.

• Extend the concept of ports and connections to tight coupling, so that an importing model may only send events to its submodels via ports and connections established between them. This mechanism further protects the internal behavior of submodels. It makes DCharts more modular.

• Implement the support of more target languages in SCC. Users will be able to integrate the code generated by SCC with the code generated by other code generators for other formalisms. This integration allows the users to model a system with different formalisms and tools, and finally combine different parts to get a complete system.