Design and Analysis of Rule-based Systems with Adder Designer

Monitoring and control systems are an important class of embedded systems. They check sensors providing information about the system’s environment and take actions depending on the sensor reading. An important part of such a system is a control process that makes decisions based on collected data. The control process may be implemented to use a rule-based system to make decisions. This paper focuses on the design and analysis of such rule-based systems for embedded control systems.

The presented approach is being developed so as to facilitate designing of Petri nets’ models of embedded real-time systems. RTCP-nets (Real-Time Coloured Petri nets, see [10]) are used as the modelling language. They are a result of adaptation of timed coloured Petri nets (see [3]) to modelling and analysis of embedded systems. RTCP-nets enable modelling of embedded systems incorporating a rule-based system. The paper focuses on designing of rule-based systems that can be included into an RTCP-net model.

A rule-based system can be represented in various forms, e.g. decision tables, decision trees, extended tabular trees (XTT, [7]), Petri nets ([2]) etc. An interesting comparison of different forms of rule-based systems can be found in [5]. Rule-based systems can be also developed in various forms. First of all, a set of decision rules can be explicitly given by a designer. On the other hand, a set of decision rules can be generated from the acquired data automatically ([8, 9]). A more detailed presentation of the current state-of-art can be found in [4].

The goal of the approach presented in this paper is to prepare such a decision table to be included into an RTCP-net model. To achieve a reasonable level of a rule-based system quality the set of rules must be designed in an appropriate way. Moreover, it should satisfy some properties such as: completeness, consistency and optimality. The approach is supported by a computer tool called Adder Designer. The tool allows designing tables with both simple and generalised decision rules. Moreover, it is equipped with transformation algorithms that allow users to convert a decision table with generalised decision rules into a table with simple ones and to glue two or more simple rules into a generalised one. Finally, Adder Designer enables users to verify selected qualitative properties of decision tables.