Work Area: Theories for Concurrency and Real Time: Specification and Verification
Keywords concurrency, real-time, hybrid systems, temporal logic, process algebra
Start Date: 1 July 92 / Duration: 36 months / Status: running
[ participants / contact ]
Abstract REACT-P addresses the problem of developing correct concurrent and reactive systems. Two complementary approaches are being considered: refinement from specification, and verification. The technical emphasis is on verification tools, the handling of realistic features of systems such as real time, and the structuring of the development process by compositionality or layering. The work builds on results achieved in SPEC (3096).
Reactive systems continuously interact with their environment. Examples are concurrent, distributed or embedded systems. Such systems have become very widespread (eg as various forms of control systems or in computer networks), and their correctness is often critical but notably difficult to ensure. REACT-P aims to provide frameworks to help developers build correct reactive systems.
The project is based on the symbiosis between two complementary approaches of ensuring the correctness of reactive systems. The first is refinement from a specification through correctness-preserving transformations; the second is verification. The refinement approach provides a guide to the construction of the system. It is usually insufficient to guarantee correctness, since there is rarely a complete and accurate specification of the system. The system is thus specified as it is being developed. Verification provides the means to check, at various stages of the design, that the system description is actually compatible with its expected properties. These properties can range from several forms of consistency to intricate requirements specified in a logical language. Verification is thus the tool that enables the designer to be confident that the formal description of the system obtained does indeeed satisfy the requirements.
The technical emphasis of the project is threefold. First, a heavy emphasis is placed upon the verification methods that can automatically and efficiently handle real life examples. Second, a framework that incorporates the key relevant features of reactive systems is sought. For instance, real-time constraints, the behaviour of hybrid systems where the program actually controls a continuous physical device, and the probabilistic behaviour of processes, are all fully taken into account. Finally, the problem of structuring the development process by compositionality or layering is addressed. Such approaches are stepping-stones towards the use of formal design methods in large systems.
The first year of REACT-P has been marked by a number of achievements of which a noteworthy selection follows.
New insights into the development of correct reactive systems, and especially into the development of real-time and hybrid systems, are likely to be gained from this project. Verification tools that can cope with industrial examples can be expected as a downstream result.
All REACT reports and other information about the project is available by anonymous ftp from the pub/react directory on montefiore.ulg.ac.be (126.96.36.199).
Université de Liège - B
Grande Traverse 10
B - 4000 LIEGE
Universität Kiel - D
IMAG - F
ICS - FORTH - GR
University of Uppsala - S
University of Oxford Computing Laboratory - UK
VTT Computer Technology Laboratory - SF
Prof. P. Wolper
REACT-P - 6021, August 1994
please address enquiries to the ESPRIT Information Desk
html version of synopsis by Nick Cook