Automotive embedded systems have evolved enormously over the past decades. The use of electronics and software in automotive products has grown exponentially. For example, today vehicles in series production contain the same electronics as an aircraft did two decades ago. To satisfy customer demands and competitiveness between OEMs, innovation will drive the increase in electronics parts to 40% of a vehicle’s value by 2015.
The inclusion of embedded system technology in vehicles has had - and is still having - a radical impact on their development, production and maintenance. Over the past decades, automotive embedded systems have evolved from single standalone computer systems, simple enough to be designed and maintained with a minimum of engineering, to distributed systems including several networks, large numbers of sensors, electric motors and points of interactions with humans. These distributed systems provide enormous opportunities for the future, but at the same time require new skills, methodologies, processes and tools.
A holistic approach to automotive embedded systems modelling needs to address several concerns, from features to implementation over structure and behaviour, environment modelling and requirements to verification and validation information. Such an approach also needs to consider mapping and interoperability with existing tools and moreover, for industrial acceptance, to provide tools and be standardised. By developing EAST-ADL2, the ATESST research project aimed to provide a basis for such a holistic approach.
Improvement of automotive safety, efficiency and environmental-friendliness relies heavily on the usage of electronics and software. Recent systems provide optimisation and coordination of functions within the vehicle but also externally between vehicles and road infrastructure (environment).
The complexity induced by the electronic architecture and increasingly by the applications, needs to be managed adequately. The goal of the ATESST project was to deliver an Architecture Description Language suitable for these systems. This provided a means to handle the complexity and improve safety, reliability, cost and development efficiency of automotive electronic systems.
An Architecture Description Language (ADL) for automotive, software-intensive systems – why does it matter? Experience has shown that the comprehensive development of automotive systems needs more attention.
How can the automotive-specific language EAST-ADL2 help? Cornerstones of high-quality embedded systems are mature engineering techniques using model-based development and standardized, reliable architectures. Industrial experience shows that model-based development techniques are required in the automotive domain to improve the quality and dependability of embedded systems. The focus of EAST-ADL2 developed in ATESST was to adequately meet the engineer’s needs regarding practical methods and engineering information management. EAST-ADL2 focused mainly on the architecture-induced complexity of automotive embedded systems. It is therefore the ontology for automotive electronics and software, making system models unambiguous, consistent and exchangeable.
The ATESST results were based on the EAST-ADL (Architecture description language developed in the ITEA EAST-EEA project). A completely revised version of EAST ADL was delivered, EAST-ADL2.0, where existing constructs were validated and adjusted, and additions were made where needed. In particular, environment modelling and behaviour specification were refined. These areas were both important aspects of handling application complexity. Also, the support for variability and re-use were amended.
Variability of automotive system increased the complexity that engineers were facing and was a major threat to safety and reliability. EAST-ADL2.0 was formalized through the definition of a UML2 (Unified Modelling Language) profile. This profile took into account the new Object Management Group standard for real-time and embedded systems called MARTE (Modelling and Analysis of Real-Time and Embedded systems) as well as the Society of Automotive Engineers' Architecture Analysis & Design Language, AADL.
The ATESST results were validated through the implementation of a prototype tool based on the Eclipse framework and an automotive demonstrator. The demonstrator would contain safety functions of different character, such as adaptive cruise control, remote speed limits and engine management
Within the ATESST project, an open source UML2 modelling and profiling environment based on Eclipse, called Papyrus, has been developed. A number of Eclipse Plug-ins have also been developed. Dedicated enhanced support of the language was developed for the Papyrus UML modelling tool, which is freely available (www.papyrusuml.org).
The tool was customized to facilitate its use for UML2 and to show the capabilities offered by the use of EAST-ADL2 concepts, e.g. support for feature and variability modelling or initial support for import and export to the Simulink platform.
The main achievement of the ATESST project is the standardization of the architectural description language EAST-ADL2. If EAST-ADL2 becomes a de-facto standard for the automotive industry, complementing the AUTOSAR standard, it will introduce a common terminology and serve as a basis for improved tool interaction and information exchange between/within companies. Being a UML2 solution, it allows open usage and standardization of results.
The ATESST project developed not only the language and a backbone for the accompanying set of practical methods, but also a set of Eclipse-based plug-in tools and a comprehensive, realistic example of the language’s use, demonstrating its suitability for down-to-earth industrial scenarios.
In short, with EAST-ADL2 the ATESST team has provided a highly relevant model-based development technique for those who are seeking to improve industrial automotive system development.