Software technology safety modeling

Panesar-Walawege, R.K., Sabetzadeh, M., Briand, L. Supporting the verification of compliance to safety standards via model-driven engineering Approach, tool-support and empirical validation. Information and Software Technology 55(5), 836-864 (2013)... 

Instrumentation and Control (I C) systems are very often subject of probabilistic examination either within separate structural reliability analysis or Probabilistic Safety Assessment of a whole technological complex (e.g. Nuclear Power Plant). Use of programmable components in the design of these systems represents a challenge and utilizes the methods, which have been developed for components with a different behaviour. The typical method used for above mentioned examination is Fault Tree Analysis (FTA) (Vesely et al., 1981). The way of software faults modelling within Fault Trees vary a lot between particular models and there is no generally accepted modelling technique. 

SASSUR is targeted at bringing together experts, researchers, and practitioners from diverse communities, such as safety and security engineering, certification processes, model-based technologies, software and hardware design, safety-critical systems, and applications communities (railway, aerospace, automotive, health, industrial manufacturing, etc.). 

According to the model of 5 safety classes recommended by Hblscher and Rader /2/, class 4 corresponds to the "normal safety standard" in technology, this is no dangerous failure may occur in the presence of an undetected dangerous fault. The requirements of the safety classes 1, 2 and 3 are already so high that corresponding circuits can in almost every case only be realised with redundant or partly even diversity hardware und software. 

As previously mentioned, software safety mechanisms implement software safety requirements. Our classification of mechanisms and their formalization has taken into account their attributes, semantics, and also the technology platform in which they are realized (i.e. AUTOSAR). This resulted in patterns that can be applied by safety engineers in order to satisfy technical safety requirements. These patterns are closely related to safety mechanisms. This is consistent with traditional engineering methods, since engineers tend to specify implementation requirements very close to the solution domain. We use these patterns to define our specification language and meta-model relations from the identified patterns to AUTOSAR elements. 

Basically all models in safety technology are System models . The entire ISO 26262 is based on a stmcture in which software and hardware components are also described through a systemic approach. Therefore, a combination of system elements is chosen that facilitates the implementation of the intended functionality. 

Generally by the very nature of their inductive development process, systems developed asing AI techniques tend to show excellent performance against the particular data model used during the problem definition activity. Moreover, this performance can often be achieved for the expenditure of very low levels of effort n compared with conventional software systems [4]. For example, some estimates have placed the cost of development of AI type systems at perhaps one tenth to one hundredth of that associated with conventional systems to achieve the same purpose, moreover the maintenance effort assodated with the deployment of AI technology can also be very low, indicating a considerable level of user satisfaction with such systems once deployed. Where there are significant safety risks associated... 

---