How to engineer tool-chains for automotive electrical/electronic architectures -

How to engineer tool-chains for automotive electrical/electronic architectures

Nowadays, the major part of innovations in modern cars are introduced by Electrical/Electronic (E/E) architectures and the applications implemented upon them. Although notable developments and new technologies in terms of batteries and engines are strongly pursued in the context of hybrid and electric vehicles, it is still expected that E/E architectures here become the main driver of innovation.

But modern E/E architectures consist of more than 100 Electronic Control Units (ECUs) and various bussystems like Controller Area Network (CAN), Local Interconnect Network(LIN), and FlexRay, operating in different domains and connected via one or more gateways. This system complexity mainly stems from the increasing demand for driver assistance functions, providing both improved safety and more comfort for occupants.

In this context,the Anti-lock Braking System (ABS), Electronic Stability Control (ESC) or Adaptive Cruise Control (ACC) represent some of the most common functions. It is anticipated thatimplementation of novel drive-by-wire systems particularly in electric vehicles and enhanced Human Machine Interaction (HMI) applications will additionally increase thecomplexity of E/E architectures.

Hence, the design of such architectures cannot be covered by a manual workflow anymore but rather requires a sophisticated tool-chain, formodeling, early-stage verification and validation, and testing of the system.

As a result, automotive companies and suppliers are showing an increasing interest in system-level modeling tools that cope with a growing complexity and harder requirements and, at the same time, simplify and improve the design process. In the past few years, a largenumber of design tools have emerged which are progressively being adopted in industry.

However, with the growing requirements on E/E architectures on the one hand andthe increasing number of possible design tools on the other hand, the selection of suitable tools is becoming a major challenge for car manufacturers (both the OriginalEquipment Manufacturers (OEMs) and Tier 1 suppliers). It is not possible to choose one single tool for the entire design process since such a versatile tool simply does notexist.

The complexity of E/E architectures rather requires the combined useof diverse model-based tools from various vendors, covering different development stages.However, while each tool might work well within its own scope of functionality, a tool integration into a flexible and consistent tool-chain is still a major challenge. Among others,problems arise from insufficient interfaces and error-prone manual, rather than automatic, input of essential design data.

This paper provides an overview of available toolsand guidelines to help researchers, developers, and engineers in the automotive domain in the selection of a proper tool-chain. We describe how to narrow down from a highnumber of design tools, how to create a good summary of their most essential properties and, finally, how to choose the most appropriate ones in order to build a tool-chain. Asa case study, we have surveyed 22 established and emerging tools that are commonly encountered in the automotive domain.

Taking all our findings into account, we conclude that due to strong competition in the automotive domain, there is a lack of appropriate interfaces between tools from different vendors. In spite of a few, more or less, efficient exceptions, for example, provided by tools from dSpace or Mathworks, there is almost no native tool connectivity between the high-leveldesign and the synthesis or verification and validation tools.

And although AUTOSAR is finding its way into more and more automotive software products, there is still a need forstrong interconnection mechanisms beyond the established data exchange formats, like Data Base Container (DBC),Field Bus Exchange (FIBEX), Kabelbaumliste (KBL-Harness Description List), or AUTOSAR XML.

To read this content in full, download the complete paper at the author archives on line at the University of Pennsylvania.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.