Description

Software development for the automotive domain has become the enabling technology for almost all safety-critical and comfort functions offered to the customer. 90 % of all innovations in automotive systems are directly or indirectly enabled by embedded software. However, tremendous complexities are introduced due to the large number of functions, their interactions, and their supporting infrastructure. Today's luxury cars contain up to 80 electronic control units (ECUs) and five different, inter-connected network platforms, over which thousands of software-enabled functions are distributed. This complexity has started to become the limiting factor for automotive software development; this was aptly demonstrated by the presentations, panels and discussions at ASWSD 2004.

Adequate management of this complexity is particularly important. Major challenges are, for instance, the rapidly increasing dependencies between safety-critical and comfort functions, advances in wired and wireless networking infrastructure that enable interconnection between cars and backend service providers and user interfaces that need to be easy to use, but also give priority to information necessary for safe vehicle operation without distracting the driver.

These challenges are aggravated by demanding time-to-market requirements, short development cycles, rapid change of technological infrastructures, customer demands, and product lines. The silent revolution currently underway in the automotive domain thus consists of a shift of focus from hardware to software infrastructures and from ECUs to software services as the center of concern in the development process. This puts the software architecture for future generation automotive systems in the spotlight as a critical element both for enabling advanced services supporting drivers and passengers, and for managing the complexity of these functions amidst the high safety demands they are subject to.

Advanced development methods such as tailored development processes, structured systems and software architectures, model-driven development techniques and notations a well as formalized techniques of quality assurance have emerged as an approach to dealing with the mentioned demands and complexities, in particular during the analysis, specification and design phases of the development process. Such advanced development approaches have numerous benefits and advantages, including:

They provide a basis for traceability from requirements specifications to implementation artifacts. This enables model-based requirements tracing, verification and validation approaches, and addresses systematic changes to models model during the development process.
They hold the promise of reduced turn-around times in iterative and incremental software and systems development. They enable engineers to explore model changes before changing the actual system.
They support product-line software development by separating different functions for product line alternatives in an otherwise common, integrated model.
Models, description techniques and associated development processes can be coordinated to provide contiguous, gap-free refinement and transformation steps from requirements to code. Models are connected and integrated to span all abstraction levels. This enables design tools, test and verification tools and code generators to work from the same sets of models to provide improved software quality.

On the infrastructure side, service-oriented architectures and automotive middleware platforms, such as Autosar, are emerging as a means to manage the complex dependencies between vehicular functions, to provide standardized, scalable and validated infrastructures.