Keywords: certification, safety, dependability, reliability, software, railway.

The current geo-political and social factors are increasing the demand for a simple, effective and energy saving transportation means for the great numbers of people that need to move daily from one place to another. One of the solutions for this problem is based on the railway lines through the major urban centers, which require the railway industry to improve the necessity for more integrated systems to satisfy this growth in passenger transit. Associated with this, there is increased demand for the assurance of the necessary safety levels for each integrated railway system.

In this paper, the authors analyse the certification issues in railway safety, comparing them to other critical domains, such as aeronautics, space, and automotive, while highlighting the main commonalities and differences between them.

The IEC 61508 is the tentative response of the community to address the system and functional safety in the same way across the different application domains. The meaning of safety integrity level may be applicable to all the mission-critical domains and, if using a similar basis, it is even possible to analyse hazards from one domain based on similar hazards from another domain. This information flow will enable the dependability and safety engineers to better understand and quantify the risks associated with each hazard, making a more quantitative analysis (based in historic data) instead of a simple qualitative analysis using their past experience.

A thorough analysis of the current methods for assessing a systems' dependability, both qualitatively and quantitatively, is given, with particular emphasis on software related issues. The familiar failure modes, effects (and criticality) analysis and hardware software interfaces analysis are briefly presented as qualitative methods that shall be used to analyse the system or software dependability. Fault tree analysis and reliability block diagrams are the possibility presented to make the quantitative analysis of the failure rates and the reliability predictions for a system.

The quantitative assessment of the system dependability requires credible evidence of a sufficiently low failure rate, in accordance with the particular system requirements.

Quantifying the dependability and probability of failure of the software is clearly a sensitive and controversial task.

In this paper, we debate on the possible alternatives for analyzing a system safety and dependability, their limitations and shortcomings, and report on the industry feasible approaches. The approach taken for certification of a railway interlocking system based in the methods used by the industry is also described but with past knowledge of other domains, such as aeronautics, space and defense.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £110 +P&P)