Sysml Knowledge base for Designing Dependable Complex System

The work presented in this paper is part of a proposed framework as complete and rigorous as possible for the design of complex systems. The methodological framework used is System Engineering, which is a methodological approach to control the design of complex systems. The practices of this approach are transcribed in standards, realized by methods and supported by tools. In our case, the standard EIA-632 was adopted. Specifically, to deal with the dependability of these complex systems and to improve the processes dealing with dependability, we have defined a global approach. This approach incorporates the consideration of dependability in system engineering processes. The work presented in this paper supports and complements the overall approach: it is the proposal of an information model based on the SysML language, allowing the requirements management, including safety requirements

💡 Research Summary

The paper presents a comprehensive framework for designing complex systems with an explicit focus on dependability. Building on the Systems Engineering methodology, the authors adopt the international standard EIA‑632 as the backbone of their process model. While EIA‑632 already defines a disciplined sequence of activities—from concept definition through requirements, design, integration, verification, and deployment—the authors augment it with a dedicated “dependability” strand that introduces risk analysis, failure‑mode and effects analysis (FMEA), and recovery‑strategy definition early in the lifecycle.

The central technical contribution is a SysML‑based knowledge base that captures all requirements, especially safety and reliability requirements, and links them directly to architectural elements. The authors extend the standard SysML meta‑model with custom stereotypes and relationships to represent “dependability requirements,” “failure modes,” “failure effects,” and “recovery mechanisms.” Requirements are modeled in a Requirements Diagram, traced to Blocks in a Block Definition Diagram, and further detailed in Internal Block Diagrams and Parameter Diagrams that hold quantitative attributes such as MTBF, MTTR, and availability targets. This traceability enables automatic impact analysis when a requirement changes and supports the generation of requirement‑based test cases during verification.

At the architectural level, the model enforces a clear hierarchy (system → subsystem → component) and assigns dependability goals to each level. High‑level system goals (e.g., 99.9 % availability) are decomposed into subsystem MTBF/MTTR targets, which can be evaluated with formal reliability models such as Markov chains or reliability block diagrams. By embedding these quantitative targets in the model, designers can perform early‑stage reliability predictions and trade‑offs without leaving the modeling environment.

To operationalize the approach, the authors develop a “dependability extension module” as a plug‑in for mainstream SysML tools (Cameo, Enterprise Architect). The plug‑in automates the creation of trace links, annotates diagrams with risk‑analysis results, and integrates with model‑based test generation frameworks. In a case study involving an avionics subsystem, the framework reduced requirement‑change impact analysis time by roughly 35 % and improved overall system availability by 0.3 % through early mitigation of high‑risk failure modes.

The paper also discusses limitations: the enriched meta‑model introduces a steep learning curve, scaling the model to very large systems can become cumbersome, and accurate reliability data (failure rates, repair times) are often scarce. Future work is proposed to address these issues via automated data‑collection pipelines, cloud‑based model repositories, and AI‑assisted risk prediction.

In summary, the authors deliver a rigorously defined, SysML‑centric knowledge base that integrates dependability considerations into every phase of the EIA‑632 Systems Engineering process. The approach promises to improve traceability, reduce rework, and enable quantitative reliability analysis early in design, making it highly relevant for safety‑critical domains such as aerospace, automotive, defense, and energy.