This section discusses applications in the automotive industry, mechatronics, and information technology.

 

 

Automotive Industry

The automotive industry is currently in a fast and spectacular evolution towards the intelligent, safe, environmental, interconnected, and economic car. Electronics is at the basis of most of this development. New features such as automatic intelligent parking assist, blind-spot information system, navigation computers with real-time traffic updates, not to mention electronically controlled brakes or electronic power steering, are out and running in most recent high-end cars. This development is going to continue with new functionality being adopted not only in premium cars but also in the mass-market. This, of course, comes at a cost. Automotive electronics, currently accounts for 22% of a vehicle’s cost and is predicted to increase to 40% by 2010 (www.altera.com).

This evolution brings a series of challenges in all steps of the development cycle. How to specify and model such a complex system? There is a need for a component based modelling, analysis, and synthesis approach in which independently designed hardware and software components can safely be combined into a working system. How to achieve the ever increasing demand on functionality and safety, at an affordable cost?

Modern automotive electronic systems are highly distributed networks with components interacting over various infrastructures. How to achieve a safe and predictable system at such a huge level of complexity and heterogeneity? A well defined methodology is needed for mapping the complex functionalities on predefined distributed automotive platforms. This assumes well defined standards, middleware layers, analysis tools, software generation tools, design exploration and optimisation approaches.

Due to ARTIST2 activities, (e.g. the ARTIST workshop “Beyon AUTOSAR” in Innsbruck) several technical meetings between TU Braunschweig and leading automotive suppliers in the AUTOSAR context held place. As a main topic it was discussed how compositional performance verification methods can be utilized in the automotive design process to facilitate the network integration problem. TU Braunschweig was invited to the SAE world congress 2007 in Detroit to present recent results in compositional performance verification.

Mechatronic Industry

Traditionally, the development of mechatronic systems was a rather sequential process. First the mechanical part was designed, next the hardware infrastructure was fixed, and finally the embedded software was developed. Typically, this lead to many proplems at systems engineering, because only then the interference of design decisions from the disciples became visible. To impriove this process and to shorten the time-to-market, there is a clear trend towards concurrent engineering. To be able to detect problems earlier in the development cycle, there a strong need for high-level models that allow early analysis of system-level design decisions. Moreover, there is an increasing interest in the use of models to improve the early testing process; for instance, one would like to test the embedded software before its environment is available.

Concerning the execution platforms used, one can observe the need for a flexible process where one can easily switch between various solutions, such as the amount of distribution, the topology used, the communication infrastructure, and the operating system. Often in a first release of a high-tech system the execution platform is overdimensioned. For instance, one might choose a highly distributed architecture to avoid scheduling problems. In a later version, a strong cost reduction has to be achieved by combining more functionality on a single node. One major problem is to foresee at an early stage of the design whether a particular hardware platform is feasible for a given software system. Hence there is a strong need for methods that can help engineers to make a well-founded choice for an execution platform. Another important problem companies have to deal with is incomprehensive, far too detailed and inconsistent documentation. This creates a gap between the application domain and the implementation. To address this problem, systematic approaches are needed to refine high-level models into working implementation in a predictable way.

Information Technology

Microelectronic technology is continuing to grow according to Moore’s law. However, the need for computation power in industry is growing even faster. This is the case with traditional areas such as technical/scientific computation, and, more recently, modern applications, for instance interactive multimedia, high bandwidth communication, or speech recognition. Many of these applications are running on mobile a computer, which makes issues even more complicated: an unprecedented amount of computation power has to be delivered with very low energy consumption. So, instead of just running after high performance, industry is out after a good performance - energy product. These unprecedented performance/energy requirements cannot be achieved by further pushing processor technology along the traditional Pentium lines. New architectures are needed in which several lower performance (and less energy hungry) computation nodes are cooperating in order to globally achieve the expected performance. Modern MPSoC and NoC architectures are developed along these lines.

Another clear trend is towards reconfigurable architectures, in general, and configurable processors, in particular. The generic goal is to achieve a high degree of flexibility (traditionally available only with software implementation) at a power consumption which is much lower than achievable with a traditional software implementation using general purpose processors.

The emerging trend for multimedia applications on mobile terminals, combined with a decreasing time-to-market and a multitude of standards have created the need for flexible and scalable computing platforms that are capable of providing considerable (application specific) computational performance at a low cost and a low energy budget.
Hence, in recent years, the first multiprocessor System-on-Chip components have emerged (like e.g. TI OMAP, ST Nomadik, Philips Nexperia, IBM/Toshiba/Sonys CELL). These platforms contain multiple heterogeneous, flexible processing elements, a memory hierarchy and I/O components. All these components are linked to each other by a flexible on-chip interconnect structure. These architectures meet the performance needs of multimedia applications, while limiting the power consumption.

(c) Artist Consortium, All Rights Reserved - 2006, 2007, 2008, 2009

Réalisation Axome - Création de sites Internet