The paper presents the AIDOaRt project, a 3 years long H2020-ECSEL European project involving 32 organizations, grouped in clusters from 7 different countries, focusing on AI-augmented automation supporting modeling, coding, testing, monitoring, and continuous development in Cyber-Physical Systems (CPS). To this end, the project proposes to combine Model Driven Engineering principles and techniques with AI-enhanced methods and tools for engineering more trustable and reliable CPSs. This paper introduces the AIDOaRt project, its overall objectives, and used requirement engineering methodology. Based on that, it also focuses on describing the current plan regarding a set of tools intended to cover the model-based capabilities requirements from the project.
Tools used in the development of safety related software applications need to be qualified as safe. That is, the tools cannot be allowed to introduce hazardous faults into the application, e.g., a compiler shall not generate dangerous code due to failure of the compiler. In many cases laws and regulations require the product development of safety related applications to comply with industry sector specific safety standards. Examples of such standards include EN50129/50128 for railway applications, ISO/EN13849 for machines with moving parts, DO-178B/C for avionics, or IS026262 for cars. These standards require the use of a rigorous development and maintenance process. The standards are also mainly intended to be used when developing systems from scratch. However, most development and test tools are not developed from scratch according to the rigorous processes of these standards. In order to address this issue, some of the standards provide means for qualifying existing tools as a more lightweight and pragmatic alternative to a regular certification process. In this paper we analyze the concept of these qualification approaches. The result of the analysis in our contribution includes a set of approaches that can be applied individually or as a combination in order to reduce the effort needed for qualifying tools. As a running example we use one of the most flexible but at the same time dangerous, even prohibited, maintenance techniques available: dynamic instrumentation of executing code. With this example, we describe how exceptions in these standards can be utilized in order to qualify a dynamic instrumentation tool with a minimal effort, without following the process of tool certification as defined by the standards.
In the railway domain, standards such as the EN5012x family prescribe processes to be followed for the management and certification of safety-critical systems. This results in a need to model processes and retrieve process-based arguments to prove that the system achieved the required safety level in order to reduce time and cost spent in the certification process. In this paper, we present the application of the MDSafeCer, i.e. a model-driven safety certification method, for railways. In particular, we model in SPEM 2.0 the safety requirements process according to what described in the safety plan, and we show how it is possible to extract safety evidence to prove the compliance of this process to the EN50128 standard.
Space systems often need to be engineered in compliance with standards such as ECSS and need to ensure a certain degree of dependability. Given the multi-faceted nature of dependability (characterized by a set of concerns), assuring dependability implies multi-concern assurance, which requires the modelling of various system characteristics and their co-assessment and co-analysis, in order to enable the management of trade-offs between them. CHESS is a systems engineering methodology and an open source toolset, which includes ConcertoFLA. ConcertoFLA allows users (system architects and dependability engineers) to decorate component-based architectural models with dependability-related information, execute Failure Logic Analysis (FLA) techniques, and get the results back-propagated onto the original model. In this paper, we present the customization of the CHESS methodology and ConcertoFLA in the context of the ECSS standards to enable architects and dependability engineers to define a system and perform dependability-centered co-analysis for assuring the required non-functional properties of the system according to ECSS requirements. The proposed customization is then applied in the context of spacecraft Attitude Control Systems engineering, which is a part of satellite on-board software.