Item Type Article Authors Papadopoulos, Y.; Walker, M.; Parker, D.; Sharvia, S.; Bottaci, L.; Kabir, Sohag; Azevedo, L.; Sorokos, I. Citation Papadopoulos Y, Walker M, Parker D et al (2016) A synthesis of logic and bio-inspired techniques in the design of dependable systems. Annual Reviews in Control. 41: 170-182. RightsAbstract: The technologies of model-based design and dependability analysis in the design of dependable systems, including software intensive systems, have advanced in recent years. Much of this development can be attributed to the application of advances in formal logic and its application to fault forecasting and verification of systems. In parallel, work on bio-inspired technologies has shown potential for the evolutionary design of engineering systems via automated exploration of potentially large design spaces. We have not yet seen the emergence of a design paradigm that effectively combines these two techniques, schematically founded on the two pillars of formal logic and biology, throughout the design lifecycle. Such a design paradigm would apply these techniques synergistically and systematically from the early stages of design to enable optimal refinement of new designs which can be driven effectively by dependability requirements. The paper sketches such a model-centric paradigm for the design of dependable systems that brings these technologies together to realise their combined potential benefits.

Safety-critical systems are becoming larger and more complex to obtain a higher level of functionality. Hence, modeling and evaluation of these systems can be a difficult and error-prone task. Among existing safety models, Fault Tree Analysis (FTA) is one of the well-known methods in terms of easily understandable graphical structure. This study proposes a novel approach by using Machine Learning (ML) and real-time operational data to learn about the normal behavior of the system. Afterwards, if any abnormal situation arises with reference to the normal behavior model, the approach tries to find the explanation of the abnormality on the fault tree and then share the knowledge with the operator. If the fault tree fails to explain the situation, a number of different recommendations, including the potential repair of the fault tree, are provided based on the nature of the situation. A decision tree is utilized for this purpose. The effectiveness of the proposed approach is shown through a hypothetical example of an Aircraft Fuel Distribution System (AFDS).

Over the years, several approaches have been developed for the quantitative analysis of dynamic fault trees (DFTs). These approaches have strong theoretical and mathematical foundations; however, they appear to suffer from the state-space explosion and high computational requirements, compromising their efficacy. Modularisation techniques have been developed to address these issues by identifying and quantifying static and dynamic modules of the fault tree separately by using binary decision diagrams and Markov models. Although these approaches appear effective in reducing computational effort and avoiding state-space explosion, the reliance of the Markov chain on exponentially distributed data of system components can limit their widespread industrial applications. In this paper, we propose a hybrid modularisation scheme where independent sub-trees of a DFT are identified and quantified in a hierarchical order. A hybrid framework with the combination of algebraic solution, Petri Nets, and Monte Carlo simulation is used to increase the efficiency of the solution. The proposed approach uses the advantages of each existing approach in the right place (independent module). We have experimented the proposed approach on five independent hypothetical and industrial examples in which the experiments show the capabilities of the proposed approach facing repeated basic events and non-exponential failure distributions. The proposed approach could provide an approximate solution to DFTs without unacceptable loss of accuracy. Moreover, the use of modularised or hierarchical Petri nets makes this approach more generally applicable by allowing quantitative evaluation of DFTs with a wide range of failure rate distributions for basic events of the tree.INDEX TERMS Reliability analysis, fault tree analysis, dynamic fault trees, modularisation, petri nets.

No abstract

Ensuring safety and explainability of machine learning (ML) is a topic of increasing relevance as data-driven applications venture into safety-critical application domains, traditionally committed to high safety standards that are not satisfied with an exclusive testing approach of otherwise inaccessible black-box systems. Especially the interaction between safety and security is a central challenge, as security violations can lead to compromised safety. The contribution of this paper to addressing both safety and security within a single concept of protection applicable during the operation of ML systems is active monitoring of the behavior and the operational context of the data-driven system based on distance measures of the Empirical Cumulative Distribution Function (ECDF). We investigate abstract datasets (XOR, Spiral, Circle) and current security-specific datasets for intrusion detection (CICIDS2017) of simulated network traffic, using statistical distance measures including the Kolmogorov-Smirnov, Kuiper, Anderson-Darling, Wasserstein and mixed Wasserstein-Anderson-Darling measures. Our preliminary findings indicate that there is a meaningful correlation between ML decisions and the ECDF-based distances measures of the input features. Thus, they can provide a confidence level that can be used for a) analyzing the applicability of the ML system in a given field (safety/security) and b) analyzing if the field data was maliciously manipulated 1 .