K. Brink

Hybrid specification of control systems

An alternative approach to the development of control systems is presented. It differs from current ones in two aspects, because it is based on a formal software specification language, which allows formal verification and because it enables multidisciplinary development of a control system as existing control models can be easily incorporated in the discussed approach. Simulation of ASTRAL specifications is discussed. Simulation, as opposed to verification which is laborious, is useful to provide a quick impression of possible system behaviors. It is shown how existing (continuous) control models, like those constructed using software packages like MatLab, can be incorporated in an ASTRAL specification thus leading to a hybrid specification containing both continuous and discrete subsystems. This hybrid style of specification is illustrated by the example of a robot control system. The specification language and some aspects of control theory are discussed in the introduction. In the second section the example specification concerning robot control and incorporation of continuous system models is discussed. Simulation of the specification is addressed in the third section. The final section presents conclusions and suggestions for future work.

Analyzing Schedulability of Astral Specifications using Extended Timed Automata

This paper reports our experiences with using an extension of timed automata [1] for schedulability analysis of prototype implementations. The approach builds upon requirements specifications constructed using the formal real-time specification language Astral [7]. Astral specifications are translated into extended timed automata. The resulting automata are augmented with implementation details like assignment of processes to processors, priorities, worst-case execution times of operations, and scheduling policies. Schedulability analysis is then performed by (automated) formal verification of the extended automaton.

SEAL: a simple language for prototyping action-event specifications

Abstract According to e.g. Parnas [1], reactive systems can be modeled very conveniently using an action-event paradigm. According to this paradigm a system is assumed to be build around a state, events causing the system to modify this state. In this paper we describe a simple language, SEAL, (SEAL is an acronym for Seal: an Experimental Action-event Language) for prototyping systems which have been described in the form of action-event specifications. The language has been used for modelling some simple problems; our current activities encompass an improved implementation that allows some analysis on the statical program text.

Ada 95 as implementation vehicle for formal specifications

A main concern in the initial phases of the development of a system to be built is capturing the system requirements and expressing them as an adequate model, either formal or informal. In subsequent phases of the development of the system this model is used as reference for transformation steps. In this paper we describe a reasonably successful experimental strategy for the implementation of real-time systems, starting from a formal specification, resulting in an Ada 95 implementation of the system and we evaluate the approach based upon practical experiences.

Issues in real-time process controller realization

The first step in the software development process is the construction of a requirements specification. Ideally such a specification expresses all requirements to be posed upon the software to be developed. Current formal methods lack one desirable property: their use requires skills that are often not possessed by specialists from other disciplines. This lack of understandability complicates the realization of systems. In this paper we address problems that arise in the specification of timing requirements for a software controller. It turns out that such requirements are often implicitly present in the domain oriented specification. This process is illustrated with a case study. Finally we address solutions for the problems encountered, which emphasize the need for developing specification formalisms that are suited to specialists from various disciplines.<<ETX>>

Experiences with analysis of formal specifications in Astral

An important issue in the design and implementation of real-time software is the verification of (temporal) properties. Our research of the past few years focused on requirements specification of real-time software and the use of tools in modeling and design. The development of a correct and complete requirements specification is difficult, and therefore tool support is considered important. Our development approach, builds upon requirements specifications constructed using the formal real-time specification language Astral. The development framework offers support for end-users in the construction of a software requirements specifications in Astral. Such end-risers are, for example, control engineers in the area of embedded, real-time control software. Support in the construction of an Astral specification through the use of analysis tools is important because it provides the user with (i) evidence concerning the correctness of the specification and (ii) feasibility of its implementation. This paper describes our development framework focusing on four approaches that have been employed in (timing) analysis of Astral specifications. The first two approaches, simulation and prototyping, are informal, while the latter two are formal, based on a translation into an extension of timed automata.

Closed world specification of embedded real-time controllers

We propose integrated engineering of embedded controllers where process control engineering (PCE) and software engineering (SWE) are fully integrated into one process of development. The PCE products are transformed into a formal hybrid specification language, H-ASTRAL, an extension of ASTRAL currently being developed at TU Delft, which enables the specification of closed models, such that both the discrete controller and the continuous controlled system are modeled in H-ASTRAL. The H-ASTRAL specification can be simulated and is starting point for implementation. In our approach the abstract closed-world specification is transformed into a concrete open-world model, in which the continuous mathematical model of the controlled process is replaced with the interface to reality. The paper reports on work in progress, summarizes the approach and discusses a real-world example.

Formal Specification Applied to Control Systems Development

Abstract Formal software specification languages are becoming sufficiently powerful to consider their use in application domains as realistic. This paper reports on a project aiming at the application of formal specification notations and techniques in the development of real-time control applications. It puts emphasis on the development of reliable control systems by means of the integration of a formal specification notation in the controller development process. This paper discusses the application of ASTRAL (Ghezzi and Kemmerer, 1991a) for the requirements specification of a robot control system. It shows that formal specification provides a useful extension to current modelling capabilities in the application domain