Scr as an Iv&v Tool

This paper describes some preliminary work on the use of SCR-style speci cation as a tool for Independent Veri cation and Validation (IV&V). Our intention is to use formal methods not as a part of the development process itself, but as a `shadow' activity, performed by an independent team of experts. Our long-term expectation is that this approach will turn out to be a less painful way of introducing formal methods into well-established, large-scale software development processes. However, there are a number of problems that need to be solved before SCR-style methods can be used in this way. The most important of these is the need to model the relationship between informal and formal speci cations. The context for this work is the development of software for the International Space Station (ISS) project. Boeing Space and Defense Group Houston (Prime) is responsible for supervising the overall development and integration of International Space Station software. There are three Product Groups (PGs), McDonnell Douglas Aerospace, Rockwell Aerospace Rocketdyne and Boeing Space and Defense Group Huntsville, who are developing several key Computer Software Con guration Items (CSCIs), which Prime is responsible for integrating. There are also several International Partners (IPs) including Russia, Japan, Canada, and the European Space Agency, who are developing software that will need to be incorporated into ISS. With over 45 ight computers and an estimated 1.1 million source lines of ight code, the potential problems are considerable. Software IV&V is currently being performed by Intermetrics, under an interim contract. The Intermetrics team is based at Fairmont, W.Va., with personnel stationed in Houston and Huntsville in order to interact with the development teams. In this paper, we rstly describe the IV&V process, and discuss some of the aspects of this process that hinder e ective IV&V. With this as background, the remainder of the paper focuses on the use of methods and tools within this process. We describe how the use of a method in an IV&V process di ers from its use in development work, and list some of the concerns that arise when a new method is adopted in IV&V. We present a small experiment in the use of SCR-style tables. The experiment showed that the natural language used in the Software Requirements Speci cation (SRS) documents is inherently ambiguous, and that the