Sohag Kabir
Sohag Kabir is an Assistant Professor of Computer Science at the University of Bradford. He is also the programme leader for the MSc Big Data Science and Technology and MSc Internet of Things programmes. He received his PhD in Computer Science and MSc degree in Embedded Systems from the University of Hull, UK in 2016 and 2012, respectively. His research interests include model-based safety assessment, probabilistic risk and safety analysis, dynamic safety and reliability analysis, and stochastic modelling and analysis.
Address: Bradford, UK
Address: Bradford, UK
less
Related Authors
InterestsView All (32)
Uploads
Papers by Sohag Kabir
MBSA techniques usually combine different classical safety analysis approaches to allow the analysts to perform safety analyses automatically or semi-automatically. For example, HiP-HOPS is a state-of-the-art MBSA approach which enhances an architectural model of a system with logical failure annotations to allow safety studies such as Fault Tree Analysis (FTA) and Failure Modes and Effects Analysis (FMEA). In this way it shows how the failure of a single component or combinations of failures of different components can lead to system failure. As systems are getting more complex and their behaviour becomes more dynamic, capturing this dynamic behaviour and the many possible interactions between the components is necessary to develop an accurate failure model.
One of the ways of modelling this dynamic behaviour is with a state-transition diagram. Introducing a dynamic model compatible with the existing architectural information of systems can provide significant benefits in terms of accurate representation and expressiveness when analysing the dynamic behaviour of modern large-scale and complex safety-critical systems. Thus the first key contribution of this thesis is a methodology to enable MBSA techniques to model dynamic behaviour of systems. This thesis demonstrates the use of this methodology using the HiP-HOPS tool as an example, and thus extends HiP-HOPS with state-transition annotations. This extension allows HiP-HOPS to model more complex dynamic scenarios and perform compositional dynamic dependability analysis of complex systems by generating Pandora temporal fault trees (TFTs). As TFTs capture state, the techniques used for solving classical FTs are not suitable to solve them. They require a state space solution for quantification of probability. This thesis therefore proposes two methodologies based on Petri Nets and Bayesian Networks to provide state space solutions to Pandora TFTs.
Uncertainty is another important (yet incomplete) area of MBSA: typical MBSA approaches are not capable of performing quantitative analysis under uncertainty. Therefore, in addition to the above contributions, this thesis proposes a fuzzy set theory based methodology to quantify Pandora temporal fault trees with uncertainty in failure data of components.
The proposed methodologies are applied to a case study to demonstrate how they can be used in practice. Finally, the overall contributions of the thesis are evaluated by discussing the results produced and from these conclusions about the potential benefits of the new techniques are drawn.