Abstract
In this paper, we present a transition journey of automotive software architecture design from using legacy approaches and toolchains to employing new modeling capabilities in the recent releases of Matlab/Simulink (M/S). We present the seamless approach that we have employed for the software architecture modeling of a mixed-critical electric powertrain controller which runs on a multi-core hardware platform. With our approach, we can achieve bidirectional traceability along with a powerful authoring process, implement a detailed model-based software architecture design of AUTOSAR system including a detailed data dictionary, and carry out umpteen number of proof-of-concept studies, what-if scenario simulations and performance tuning of safety software. In this context, we discuss an industrial case study employing valuable lessons learned, our experience reports providing novel insights and best practices followed.
Highlights
Employing AUTOSAR specific block set in the latest releases of M/S, we were able to implement a detailed model-based software architecture design of AUTOSAR system including a detailed data dictionary; Making use of the Simulink System Composer toolbox in the recent releases of M/S, we have created custom-defined profiles to support trade-off analysis of safety software; Using powerful simulations supported by M/S toolboxes, we were able to carry out umpteen proof-of-concept studies and what-if scenario simulations; With the help of a state-of-the-art requirements management tool plugin (Polarion for Simulink [9]) and custom-defined scripts, we achieved seamless bidirectional traceability between requirements and architecture design
M/S domains, the Rubus Component Model (RCM) and EAST-ADL [13] are among other solutions used by a smaller community in the vehicular domain [14]
A descriptive model of such a system may provide a high-level overview of the system in an intuitive and understood way
Summary
The global Electric-Vehicle (EV) industry continues to expand rapidly, bolstered by the increase in demand for fuel efficient, high performance and low emission vehicles. AUTOSAR employs component-based software architecture, for the design and implementation of automotive software systems In this context, a majority of the state-of-the-practice in the automotive Original Equipment Manufacturers (OEMs) for AUTOSAR-based, model-driven Embedded Software Engineering (ESE) is split between two main domains. A majority of the state-of-the-practice in the automotive Original Equipment Manufacturers (OEMs) for AUTOSAR-based, model-driven Embedded Software Engineering (ESE) is split between two main domains They are the Unified Modeling Language (UML) [4]/Systems Modeling. (1) State-of-the-practice steps followed by automotive OEMs and (2) Our approach (in the last twelve months) in a real-life automotive project of a multi-core, mixed-critical electric powertrain controller Even though such models are at a higher abstraction level, the document containing these models is massive. In our real-life project experience involving development of electric powertrain software (cf. Section 1.3) for an OEM.Initially, we have employed the legacy approach in Figure 1(1) and faced the challenges listed above
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
