AUTOSAR aims at facilitating reuse of standardized software components in automotive software development. For this, it defines three layers: the application components (with standardized interfaces between components), the (standardized) runtime environment, and the basic software (which abstracts the hardware and contains the driver modules). Despite this standardization, the combination of application components, runtime and basic software should still offer differentiating functionality for cars. In an exploratory study, we found this situation challenging for Requirements Engineering practice, where standard requirements implied by the AUTOSAR standard should be treated differently from OEM specific requirements targeted towards differentiation or innovation features.
We will discuss our preliminary findings at the 23rd IEEE International Requirements Conference 2015 in Ottawa, Canada during the poster and tool demo session as well as our method in the 5th IEEE International Workshop on Empirical Requirements Engineering.
This work is based on Mozhan‘s Master thesis. We would like to thank the company contacts for their keen collaborations in the study. This work has been partially funded by Software Center Project 11 ”Ecosystemability Assessment Method“ and the Swedish Vinnova/FFI project ”Next Generation Electronic Architecture“.
AUTOSAR has changed significantly how software is developed in the automotive sector. As a central standard, AUTOSAR enables reuse of software components as well as their interoperability. For AUTOSAR compliant ECU development, car manufacturers source Electronic Control Units (ECUs) from Tier-1 suppliers, but ask those Tier-1 suppliers to install AUTOSAR compliant basic software from a certified AUTOSAR- Tier-2 supplier. In this setup (to which we refer as the AUTOSAR ecosystem), the OEM has a direct business relationship with the Tier-1, but only an indirect relationship to the AUTOSAR- Tier-2 supplier, which leads to complex flows of requirements and related information between the organizations involved. In this extended abstract, we summarize preliminary results of a qualitative investigation of Requirements Engineering challenges in the AUTOSAR ecosystem. In particular, we interviewed 7 project managers from an AUTOSAR-Tier-2 supplier, and triangulated our results with 6 additional interviews with subjects from two Tier-1 suppliers and one OEM. We found that most of the requirements towards the AUTOSAR-Tier-2 supplier can be directly mapped to standard AUTOSAR components. However, a significant amount of requirements were new requirements and specific to the OEM or even a project. The well-known requirements engineering challenges we found to surface in the AUTOSAR ecosystem were mainly connected to these non- standard requirements. Standard and non-standard requirements are usually mixed, which makes it hard to fully leverage the potential benefits of reuse within the AUTOSAR standard. We argue that the holistic ecosystem perspective allows exploration of new strategies for mitigating this challenge.
Mozhan Soltani and Eric Knauss. Cross-Organizational Challenges of Requirements Engineering in the AUTOSAR Ecosystem: An Exploratory Case Study. In Proceedings of 5th IEEE International Workshop on Empirical Requirements Engineering at RE ’15, Ottawa, Canada, 2015 (PDF)
AUTomotive Open System ARchitecture (AUTOSAR) is a central standardization in the automotive indus- try which was developed to address the increasing complexity in the automotive embedded development. In this paper, we share results from an interview-based case study in which we explore the requirements engineering process of an AUTOSAR-Tier-2 supplier. Our results confirm well-known challenges re- garding requirements communication, and verification in the cross-organizational requirements engineering process. These challenges are related to those modules that handle complex sensors and actuators, where specific requirements from the car manufacturers are required beyond the AUTOSAR standard requirements. We report that using the AUTOSAR standard brings commodity advantages, however the presence of non-AUTOSAR requirements results in (partly well-known) requirements challenges in the automotive domain. We believe that approaching these challenges from a broad ecosystem perspective will allow future research to address these challenges with tailored requirements engineering processes that facilitate col- laboration between all AUTOSAR ecosystem parties.