Finite State Automata Semantics in Communicating Sequential Processes

MOTIVATION Traditionally, distributed systems and protocols are described with finite state automata (finite state machines). As result of that, many tools based on finite state automata have been developed and used. Such an example is business state machines used in IBM WebSphere Integration Developer [1]. Later on, other more powerful mathematical tools for specification and analyses of distributed systems have been developed, such as Petri nets [2], CSP [3], and so on. Modern tools and notations for specification, development and implementation of distributed systems are based on them. For example, Petri nets concepts are broadly used for specification of business processes in notations like UML (activity diagrams) [4], WS-BPEL [5], BPMN [6], etc. The newer mathematical tools are more powerful that the older ones. For example, Petri nets have more expressive power than finite state automata, but are less expressive than CSP. Our intention, here in this paper, is not to compare them. In commercial tools, that use finite state automata, as a base for business process specification, the problem is the need to convert older specifications into new one without losing the semantics. Newly developed tools are usually based on Petri nets or CSP. They are more powerful in specification and analyses, but they have to support continuity with the older developments. Such an example is IBM WebSphere Integration Developer that nowadays is based on WS-BPEL, but has to support backward compatibility with the business state machines. Intention of this paper is formally to specify finite state automata in CSP. Finite state automata semantics is clear, but there are needs for conversion of business processes specified in them to new form without losing the semantics.