An Industrial Case Study for Adopting Software Product Lines in Automotive Industry

Abstract

Software Product Lines (SPLs) seek to achieve gains in productivity and time to market. Many companies in several domains are constantly adopting SPLs. Dealing with SPLs begin after companies find themselves with successful variants of a product in a particular domain. The adoption of an SPL-based approach in the automotive industry may provide a significant return on investment. To switch to an SPL-based approach, practitioners lack a reengineering approach that supports SPL migration and evolution in a systematic fashion. This paper presents a practical evolution-based approach to migrate and evolve a set of variants of a given product into an SPL and describes a case study from the automotive domain. The case study considers the need to handle the classical sensor variants family (CSVF) at Bosch Company. Using this study, we performed a contributed step toward future switch of the CSVF into the SPL. We investigated the applicability of the proposed evolution-based approach with a real variants family (using the textual requirements of the CSVF) and we evaluated our approach using several data collection methods. The results reveal that our approach can be suitable for the automotive domain in the case study. CCS CONCEPTS • Software and its engineering à Software development techniques à Reusability

A feature-based Modeling Strategy for Managing Stakeholders

“…Feature models are commonly used to represent the commonality and variability of a system in terms of mandatory, optional, and exclusive features, as well as propositional constraints over the features [17] [18] [19]. Engineers typically use feature models to represent features and their connections in a tree-like structure.…”

A concrete product derivation in software product line engineering: a practical approach

“…The product derivation process proposed by our approach entails selecting, eliminating, extending, and, in some cases, even modifying the current feature model and traceability links (from the features to the code). Both have been described previously (Ignaim and Fernandes, 2019;Ignaim, 2021) and are used in the derived products.…”

“…It is assumed that a new customer requests a product called ProductDerived, and the Classical Sensor Development Team (CSDT) writes the requirement document for this product according to the sensor requirements and the client's needs. Using our approach, we perform a product derivation from the CSVF resulting software product line (see Table 1) by going through the following three steps: algorithm_2,value_3,signal_13,signal_9,signal_10 Product 4 identification,flag_1,layoutR_1,layoutT_1,algorithm_1,value_3,signal_10 In step 1, the project technical manager selects the required ProductDerived features from the current feature model (derived from our previous work (Ignaim and Fernandes, 2019)). 'diagnosis', 'monitoring', 'calculation', 'message', 'transmit' with the layoutT_1 feature, and 'receive' with the layoutR_1 feature are all supported by the ProductDerived.…”

“…Step 2 uses the feature model refactoring technique (Ignaim and Fernandes, 2019) to eliminate (remove), add or modify features of the current feature model, which propagates features change from the feature modelling level to the software product line level. Finally, in step 3, in order to acquire the (ProductDerived), which is the resulting product implementation, we have used a tool named friendlyMapper tool 1 (Ignaim, 2021). The derivation process is based on mapping every feature of the feature combination of the ProductDerived to feature-related-code routines in the implementation code.…”

“…We present in this paper a practical approach and research tool for assisting automated product derivation in software product lines. The specific technique is a subset of a larger one described in our previous work, which supports feature modelling and traceability of software product lines (Ignaim, 2021). Our approach is based on the feature model refactoring notations presented in Alves et al (2006), which allow us to automatically derive a given product from a set of all possible products and then specify its code.…”