page 1  (26 pages)
2to next section

Strategies for Consistency Checking, the Choice of

Unification ?

H. Bowman, E.A. Boiten, J. Derrick and M.W.A. Steen

Computing Laboratory, University of Kent, Canterbury, CT2 7NF, UK.

(Phone: + 44 1227 764000, Fax 44 1227 762811
Email: fH.Bowman,E.A.Boiten,J.Derrick,

Abstract. There is increasing interest in models of system development which use Multiple Viewpoints. Each viewpoint offers a different perspective on the target system and system development involves parallel refinement of the multiple views. Our work particularly focuses on the use of viewpoints in Open Distributed Processing (ODP) which is an ISO/ITU standardisation framework. Multiple viewpoints, though, prompt the issue of consistency between viewpoints. This paper describes an interpretation of consistency which is general enough to meet the requirements of consistency in ODP. Furthermore, the paper investigates strategies for checking this consistency definition. Particular emphasis is placed on mechanisms to obtain global consistency (between an arbitrary number of viewpoints) from a series of binary consistency checks. The consistency checking strategies we develop are illustrated using the formal description technique LOTOS. Keywords: Viewpoints, Consistency, ODP, Formal Description Techniques, LOTOS.

1 Introduction

There has been significant recent interest in using viewpoints in system development. In such modelling, each viewpoint offers a different perspective on the target system and system development involves parallel refinement of the multiple views. Notable proponents of viewpoints modelling include [10] [15] [1] [11]. All these approaches prompt the central issue of viewpoint consistency, i.e. how to check that multiple specifications of the system do not conflict with one another and are in some sense" consistent. Our perspective on consistency is tinged by the particular application of viewpoints that our work has been targetted at, viz. the viewpoints model defined in the ISO/ITU Open Distributed Processing (ODP) standardisation framework. ODP defines a generic framework to support the open interworking of distributed systems components. A central tenet of ODP is the use of viewpoints in order to decompose the task of specifying distributed systems. ODP supports five viewpoints, Enterprise, Information, Computational, Engineering and Technology. In contrast to many other viewpoint models ODP viewpoints are predefined and in this sense static, i.e. new viewpoints cannot be added. Each of the viewpoints has a specific purpose and is targetted at a particular class of specification. A complete ODP specification should contain a description of the system from each of the defined viewpoints. In addition, formal description techniques (FDTs) are variously applicable to the specification requirements of the different viewpoints. For example, Z [14] is being proposed for the information viewpoint and LOTOS [3] for the computational viewpoint.

Figure 1 [7] depicts the relationships that are involved in relating ODP viewpoints. Development yields a specification that defines the system being described more closely. The term development embraces many mechanisms for evolving descriptions towards implementations, one of which is refinement. Because all five viewpoint specifications will eventually be realized by one system, there must be a way to combine specifications from different viewpoints during development; this is known as unification. For specifications in different FDTs to be combined or unified, a translation mechanism is needed to transform a specification in one language to a specification in another language. Consistency is a relation between groups of specifications.

In our work on consistency we distinguish between intra and inter language consistency checking. Intra language consistency considers how multiple specifications in the same language can be shown to be consistent, while inter language consistency considers relations between specifications in different FDTs. The latter issue is a significantly more demanding topic than the former.

?This work was partially funded by British Telecom Research Labs., Martlesham, Ipswich, U.K. and the Engineering and Physical Sciences Research Council under grant number GR/K13035.