Selected Papers
Below are just a few examples of the papers I have produced over the years. Bibliographical details of the publications can be found in my list of Selected Publications.
The Importance of Data in Safety-Critical Systems
All computer-based systems make use of data in one form or another and it is common to consider this data as an integral part of the system’s software. However, an increasing number of systems make use of data to configure the system or to describe its environment, and in such cases data often forms a distinct element. Data in these ‘data-driven’ systems is often generated and maintained quite independently from the executable software, and there is some evidence to suggest that in many cases it is not receiving the attention it deserves.
An Internet-Based Searchable Database of Air Accidents
Despite improvements in the safety of individual aircraft, a steady growth in the volume of air travel is producing an overall increase in serious aircraft accidents. One of the weapons in the fight against this trend, is the use of accident analysis to detect patterns within these accidents and to identify areas where improvements are required. This paper describes the Warwick Air Accident Database (WAAD), which was established specifically to facilitate this process of accident analysis.
The Role of Data in Safety-Related Railway Control Systems
To illustrate the role and importance of data in safety-related systems, this paper looks at the data associated with a railway command and control system. Such a system has a range of safety-related functions, and must also operate in the context of other safety, protection and business planning systems. The paper considers typical data errors associated with the railway environment and proposes the early definition of a system data architecture, which will allow the application of safety analysis techniques such as HAZOP.
Data-driven Systems - the State of the Ark?
When configuring a microprocessor for a given application by the addition of software, it is clear that the characteristics, and therefore the safety, of the resultant system are determined by the nature of the software as well as that of the processor. One would therefore not attempt to argue that a computer system was safe simply because the processor has been used in other applications that have been shown to be safe. Similarly, when configuring a data-driven system by the addition of data it would seem obvious that the safety of the system is determined partly by the nature of the data, and that experience gained using systems configured with one set of data is not sufficient to deduce that the system will be safe using any other data set. However, despite the seemly obvious truth of this conclusion, experience shows that many system developers are placing great faith in tests performed using a single set of test data during system development, and inferring from this a great deal about the overall performance of the system.
