Data: An often-ignored component of safety-related systems

Safety-related systems are being constructed from hardware, software and data. The data often describes the real world environment in which the system will operate and plays a vital role in ensuring its correct operation. Logic as well as good engineering practice dictates that data is produced to the same integrity requirements as the other system elements. Unfortunately, experience and anecdotal evidence suggest that this is all too commonly not the case. Data-driven systems use data from a number of sources including data extracted (and possibly processed) from existing external information systems and data produced specifically for the required system. This data is used to describe the system environment using configuration data (which is largely static or slowly moving) and status data (which is dynamic and will sometimes change rapidly). In addition, a minority of systems may use data to describe a changing use of the system with time. This additional data may be thought of as a schedule or timetable identifying control requirements as sequences or combinations of control actions. Typically, information is supplied to these systems through a ‘data supply chain’ that may involve transformations and adaptations by external information systems and human processes. The management of the data supply chain can introduce significant errors to the development and operation of safety-related systems. The work described in this paper sets out to provide much needed guidance on appropriate methods of dealing with data, which is a largely ignored system component.