Leszek Holenderski
Axel Poigné
In 6th NWPT, pages 192-193
In the computational model for synchronous reactive systems, time is assumed to be discrete, i.e. divided into enumerable number of non-overlapping segments, called instances of time. In every instance of time, a synchronous reactive system performs a so-called reaction step, which consists in fetching inputs, computing, and emitting outputs. The fetch-compute-emit sequence is considered to be an atomic action which takes exactly one instance of time. A system may consist of several reactive components, running in parallel and communicating with each other. The components perform their reaction steps simultaneously, i.e. in the same instance of time, and the outputs emitted in one component are immediately available to all other components. Thus, communication is based on instantaneous broadcasting, used in the synchronous programming languages Esterel, Lustre and Argos, as opposed to instantaneous hand-shaking, used in synchronous CCS.
In fact, we only consider pure synchronous reactive systems, i.e. those in which all inputs and outputs are pure signals. A pure signal does not carry a value. It is either present or absent, in every instance of time.
It is convenient to represent a pure synchronous reactive system (with signals ), by a Mealy automaton whose transitions are labelled by pairs , where . A transition , where and are states, describes a possible reaction step of the system. It has the following intuitive reading: provided the system starts an instance in state and is the set of all signals present during the instance, the system emits signals and finishes the instance in state . In such a setting, parallel composition of reactive systems is represented by a product of Mealy automata, which leads to the well-known ``state explosion'' problem.
It turns out that the transition relation of a pure synchronous reactive system can be coded by two sets of Boolean functions. The first one represents a set of (possibly recursive) Boolean equations whose solutions represent pairs . The second one represents a set of simultaneous Boolean assignments which represent changes of states.
In the paper, we show how to compose such Boolean automata using typical operations, like sequential and parallel composition, choice, iteration and preemption. The main result is that a Boolean automaton grows linearly with the growth of a reactive synchronous system it represents.
Comments
GMD-SET, Schloss Birlinghoven, D-53757 Sankt Augustin,
Germany.
Available as PostScript,
DVI.