From simply rolling your windows up and down to controlling your powertrain, microcontroller-based electronic control units (ECUs) have become part and parcel of automotive electronics. In fact, in an average car today some 30 or more ECUs offer a variety of functions and features for safety, fuel efficiency, comfort, convenience, and infotainment. In high-end cars, this number is easily more than 70. And it continues to rise. With this widespread proliferation of ECU modules and related embedded software in automotive applications, designers are confronted with significant challenges. According to analysts, an average car has more than 35 million lines of code. Add to this the complexity of several data buses in use in modern vehicles over which these ECUs must communicate. Although the need for more processing power with greater code complexity is driving the trend toward 32-bit microcontroller units.
With the market demanding more features and higher fuel efficiency in a shorter design cycle, traditional approaches are becoming inefficient and time consuming. There isn't enough time to develop code and then build the hardware to test the algorithms and associated code. Furthermore, these ECU modules must be tested in a system. Any late changes in any one of these units could translate into delays and more dollars. Research data shows that automobile manufacturers spend more than $2 billion a year in fixing software problems. However, if the designers can evaluate and verify ECU's performance at early stages, before building the hardware, corrections can be implemented quickly in the design process. That translates in to tremendous savings in time and cost of developing automotive electronics.
Consequently, there is a paradigm shift in automotive electronics design. Simulations and modeling is being put at the center of system design to overcome the limitations of traditional hardware/software design. Working with advanced software tools, design engineers employ mathematical models to create high-level block diagrams and state machines, which in turn, represents the system and its components such as the powertrain, transmission, exhaust, cooling, fuel, braking, and the gear box to name a few. Using these models in myriad applications under various operating conditions, model-based design is providing an environment where software verification and validation can begin much before hardware building starts, allowing information to flow from top down and bottom up.
Several EDA tool vendors have announced their entry in to this fray. A new generation of software tools for model-based design has emerged. The MathWorks, for example, has extended its simulation, modeling and code generation products to automotive electronics. Some of the early adopters include carmakers like Toyota and its primary supplier DENSO. Toward that end, Toyota and DENSO have inked a long-term agreement with the MathWorks to use its MATLAB, Simulink and embedded code generation tools like in a number of production software development programs, including powertrain control and other ECUs. Other model-based design users include Nissan Motor Co. and DaimlerChrysler.
As indicated in the cover story of this issue, the growing complexity of software in automotive electronics has prompted General Motors to make a similar transition to simulation and testing at the front end of the design process. This transition requires a well-defined methodology and a complex testing capability, according to GM (see “The Newest Car Models — Software,” by Randy Frank, p. 12.).
Major EDA tools supplier Mentor Graphics is looking beyond cabling and wire harness design tools for automotive applications. The company is expanding its offerings to this market. Besides enhancing existing tools for this application, it is also eyeing embedded software development using models and simulations. Other big names in this race include Cadence Design Systems and Synopsys. In the past, the automotive market was not lucrative for most EDA companies. But, that thinking has changed as automotive designers begin to reap the benefits of model-based design.