Why feature dependencies challenge the requirements engineering of automotive systems: An empirical study

Vogelsang, Andreas; Fuhrmann, Steffen · 2013 · OpenAlex-citations

DOI: 10.1109/re.2013.6636728

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This empirical study investigates the extent, awareness, and impact of functional dependencies and feature interactions within automotive software systems. The research addresses a critical gap in requirements engineering: while many approaches exist to model feature dependencies, there is little empirical data on their prevalence in industrial contexts or how developers handle them. The authors argue that unmodeled dependencies increase system complexity and lead to errors discovered late in development, necessitating a rigorous assessment of these interactions in real-world automotive software. The study employed a two-phase mixed-methods design conducted at the BMW Group, focusing on the driving dynamics and driver assistance domain of a future sports utility vehicle. The quantitative phase analyzed the functional architecture of a system comprising 94 vehicle features and 325 leaf functions. Using a custom Java tool, the researchers extracted a vehicle feature graph based on data flow dependencies between leaf functions. The qualitative phase involved semi-structured interviews with four feature experts. These experts evaluated a sample of 89 dependencies identified by the tool, classifying them as plausible or implausible, and known or unknown. They also discussed the importance of these dependencies for development processes and impact analysis. The results revealed a high density of interdependencies: 85% of the analyzed vehicle features depended on at least one other feature, with only 9 features being completely independent. The analysis identified 1,451 dependencies across the system, with some features influencing up to 53 others. Expert evaluation confirmed the validity of the automated extraction, as 89% of the examined dependencies were deemed plausible. However, developer awareness was low; 58% of the plausible dependencies were unknown to the experts prior to the study. The interviews highlighted that dependencies often arise from architectural decisions and shared logical signals, which are frequently undocumented. Experts emphasized that comprehensive knowledge of these dependencies is crucial for accurate impact analysis, complexity assessment, and tracing requirements to architectural decisions. The study concludes that current development methods are insufficient because they rarely consider feature interactions in specifications, leading to increased effort in later integration and testing phases. The findings challenge the assumption that architectural-level modeling is sufficient for capturing dependencies, as nearly half of the dependencies were unknown to developers. The authors advocate for more extensive modeling techniques and rigorous documentation of feature interactions in requirements engineering. This work provides empirical evidence supporting the need for improved specification techniques that explicitly account for the complex, intermeshed nature of modern automotive software systems.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• mode awareness