Archive of the Technology Trends Category

Vendors are jockeying for position in a potentially huge market for in-vehicle TV technology. At the National Associated of Broadcasters show in Las Vegas, Qualcomm‘s MediaFLO Technologies unit broadcast live streaming television to an SUV, just to prove they could. The MediaFLO platform employs a dedicated mobile broadcast network that, unlike satellite-based TV transmissions, does not require direct line-of-sight, so video quality and signal strength shouldn’t be compromised in crowded metropolitan areas. As its name implies, the platform is based on FLO air interface technology standardized by the Telecommunications Industry Association (TIA).

Meanwhile, ICO Global Communications‘ launched its ICO G1 geosynchronous satellite to cover North America. IMS Research analyst Helena Perslow calls it the first deployment of the DVB-SH (digital video broadcast - satellite service to handhelds) mobile video standard in the U.S.

“DVB-SH is expected to be one of the dominant mobile TV standards, and the one showing the most rapid growth over the next few years,” Perslow said. “In a large and spread-out country such as the U.S., satellite broadcasting may be the key to such mobile video applications as in-car TV.”

ICO expects to start its alpha trials later this year and to launch commercially in 2009. According to Perslow’s report, although in-car broadcast TV till now has not been a significant market in the U.S., such new mobile digital video applications as ICO’s will stimulate tremendous growth in the coming years. “Beginning in 2009, sales of in-car satellite TV receivers will almost double in one year and treble over two years.”

The MathWorks says automotive engineers can now verify compliance with IEC-61508 (safety systems) and The MathWorks Automotive Advisory Board (MAAB) modeling standards within Simulink. At SAE 2008 in Detroit, the company also announced version 2.0 of the MAAB‘s Control Algorithm Modeling Guidelines for MATLAB, Simulink and Stateflow, and earlier it noted that the R2008a releases of Simulink and Real-Time Workshop Embedded Coder include support for import and export of AUTOSAR Software component descriptions and generation of AUTOSAR compliant production code.

Simulink Verification and Validation software (www.mathworks.com/products/simverification/) version 2.3 automatically evaluates and verifies system models for compliance with those standards as well as with DO-178B. New model-checking features also extend the existing support available within Simulink Verification and Validation for customer-developed modeling guidelines.

Engineering organizations use modeling guidelines as a best practice to improve readability, increase maintainability, and encourage reuse, while guarding against modeling errors. Typically, modeling guidelines are verified through visual inspection of models and related documentation during design reviews. According to The MathWorks, organizations can reduce or eliminate time-consuming and error-prone manual verification by using Simulink Verification and Validation to capture their internal guidelines and automatically check their models.

The Model Advisor also checks for other attributes, such as model consistency and code-generation compatibility. Engineers can use the customization API in Simulink Verification and Validation to develop their own modeling checks, using MATLAB scripts, and register them with Model Advisor for automatic execution.

“Modeling-standards compliance is crucial to the overall development process,” said Jon Friedman, aerospace, defense, and automotive marketing manager at The MathWorks. “Many users already employ Simulink Verification and Validation to check their designs against their own modeling guidelines as a core part of their development process. Now, engineers can use the same tools to ensure that their models meet the DO-178B, IEC-61508, and MAAB standards for data types, diagnostics, code generation, and safety, without having to spend critical staff-hours on repetitive manual review work.”

Information on version 2.0 of the MAAB‘s Control Algorithm Modeling Guidelines for MATLAB, Simulink and Stateflow is available at www.mathworks.com/industries/auto/maab.html/. “By leveraging these guidelines and extending them with company- or project-specific rules, automotive engineering teams can build and manage models for specification exchange, simulation, automatic code generation, documentation, and test definition,“ Friedman explained. “Teams can generate designs that are reusable, easy to integrate, and consistent with proprietary or industry-standard guidelines.”

AUTOSAR enhancements in the R2008a releases of Simulink and Real-Time Workshop Embedded Coder eliminate the need for engineers to introduce new blocksets or tools in order to support AUTOSAR. Software Component descriptions generated by AUTOSAR authoring tools, such as DaVinci Tool Suite from Vector Informatik, can be automatically imported into Simulink to design and implement the functionality of the component. No structural changes to the model are needed. As a result, a single Simulink model can be used as a golden reference throughout the development process for simulation, rapid prototyping, and production code generation in both AUTOSAR and non-AUTOSAR environments.

After the component is modeled and designed, AUTOSAR compliant code can be automatically generated using the newest release of Real-Time Workshop Embedded Coder. To validate the generated code and test it against the reference model behavior, software-in-the-loop testing is supported by automatically configuring and generating an interface to the code that routes the simulation data using the AUTOSAR Runtime Environment API calls.

Friedman said that in addition to the component, an updated AUTOSAR Software Component description is automatically generated, enabling the component to be integrated with the AUTOSAR Runtime Environment (RTE) and AUTOSAR Basic Software.

As the number of electronic functions in a vehicle increase, so do the number of electronic control units (ECUs) required. That‘s unavoidable in some cases, but in others, it‘s possible and beneficial to reduce the total number of ECUs by controlling multiple functions from a single location. To that end, Robert Bosch LLC is developing what it calls a domain control unit (DCU). Kay Stepper, marketing manager for Bosch‘s Chassis Systems Control Division, says the DCU is based on a 32-bit, 130nm multi-core microcontroller that‘s optimized for advanced safety applications. In its current iteration the DCU will scale to 180MHz and come with up to 4MB of flash with error correction code. It will include a floating point unit, implement AUTOSAR, and include a FlexRay interface as well as CAN and LIN interfaces.

Omron is poised to enter the active safety market with a system that demonstrates sensor fusion. It combines LIDAR-based adaptive cruise control system with camera-based lane departure warning. The firm has configured a demo vehicle - a Mercedes C320 - with a black and white camera and a dashboard display. The CMOS camera displays the road ahead, highlighting the lane markers in red. The dashboard displays two broken lines converging, plus an icon representing a vehicle ahead, when one is present. The Omron system, which operates on a CAN bus, links up with the Mercedes cruise control system, the speed of which can be set in one mile per-hour increments, so a driver can set a comfortable trailing distance. What‘s especially cool about the system, however, is the way it nudges a driver with a clear but gentle tug on the wheel in the proper direction - toward the center of the lane - when the vehicle drifts over a lane marker. The haptic signal may not be firm enough to awaken a truly drowsy driver, but Omron also has an internal camera in development.

It‘s axiomatic that the sooner a design error is discovered the cheaper it is to fix it, so it makes sense to verify, validate and test early and often.

Jon Friedman, automotive industry marketing manager at The MathWorks, says that many errors originate somewhere between requirements and specifications in the design process, so it follows that verification tools should link requirements documents to specifications and verify that the specification meets the requirements. That‘s one of the fundamental principles that underlie The MathWorks new Simulink Design Verifier. Another is the importance of formal - mathematically provable - verification to complement simulation and functional verification in assuring that the requirements on which a model is based are met.

Based on a formal methods engine - Prover Technology‘s Prover Plug-in - the Simulink Design Verifier automates test generation and property proving for Simulink and Stateflow models. It generates test cases and counterexamples automatically and performs proofs by using automated mathematical reasoning to explore model execution paths.

Friedman says the Simulink Design Verifier removes the monotony and the heavy lifting from the process of creating test cases to cover all possible usage scenarios and workflows, after which production code can be generated from the verified model using Real-Time Workshop Embedded Coder.

For applications like human machine interface and body electronics, the problem facing test engineers is the sheer number of potential inputs and values. There may be a smaller number of variables in safety and powertrain applications, but the margin for error is also less.

What‘s been your experience with testing and verification?

STMicroelectronics and Freescale say that since announcing their joint design effort last year they‘ve accelerated automotive design activity in the areas of IP development, flash technology alignment and new product definition. They‘re applying Power Architecture technology for powertrain, chassis, motor control and body systems and they‘ve produced test chips in 90nm embedded flash technology.

Freescale VP Mike McCourt says the pair plan to deliver four new products early next year, and in the next couple of years they expect a “comprehensive roadmap” of scalable MCU designs. Mike reports that customers are interested in the MCU designs that are coming, as well as in the combined product and process capabilities and the dual-sourcing potential. The companies have set up four design centers employing a total of 130 design engineers.

ST VP Marco Maria Monti says that the common MCU architectural platform design facilitates the simultaneous design of multiple products with peripheral sets optimized for specific target applications, translating to faster time-to-market.

As Dr. Phil might ask, how is all of that working for you? Are you using Power Architecture technology and are you among the automotive customers anticipating delivery of samples in Q1? What‘s your take on the collaboration?

Like every other product development organization, BMW Group (www.bmwgroup.com) wants to squeeze excess time out of its process and lower its manufacturing costs. As part of that effort, it‘s decided to standardize on MSC Software‘s (www.mscsoftware.com) MD Nastran, which is multi-discipline (hence the MD) enterprise software for simulation and engineering analysis, including computational fluid dynamics/fluid structure interaction, strength and durability, shock and vibration, and thermal behavior (convection and conduction).

The software, which integrates MSC Nastran, Marc, Dytran and LS-Dyna, is intended to help BMW reduce the number of simulation tools it has to manage as well as to reduce its need for physical prototypes. MD Nastran shares a common data model among various environments - fluid, thermal, structural, etc. - so no time is lost in translation from one to the next. The tighter coupling of simulation environments should also reveal design flaws earlier.

Electronic stability control (ESC) may cost an OEM $300 or more per-vehicle, so how much should the OEM charge the customer for the feature? Nothing, says John Krafcik, vice president of product development and strategic planning at Hyundai Motor America. Hoping to seize the ESC high ground, Hyundai has decided to make the feature standard on models that account for about 70% of its sales in the ‘07 model year - upwards of 350,000 vehicles.

Citing research conducted by the Insurance Institute for Highway Safety, Krafcik noted that ESC can reduce the risk of fatal single-vehicle crashes by 56% and fatal multi-vehicle crashes by 32%. According to IIHS, as many as 10,000 fatal accidents could be avoided each year if all vehicles were equipped with ESC. Susan Ferguson, IIHS senior vice president for research, says the Institute‘s findings “indicate that ESC should be standard on all vehicles,” and adds, “Very few safety technologies show this kind of large effect in reducing crash deaths.” A researcher at the University of Michigan concurs, suggesting that ESC systems “appear to be the most significant safety advance since seat belts.”

Hyundai is by no means alone in making ESC a standard feature. According to the IIHS, ESC is standard on 40 percent of 2006 passenger vehicle models, including every vehicle from Audi, BMW, Infiniti, Mercedes, and Porsche; however, nine automakers make the technology available on no more than 25% of their models.

ESC technology is available from a handful of suppliers, and there‘s not much differentiation among them, according to Krafcik. OEMs must decide where to set the threshold, and whether or not to provide an ESC on/off switch (Hyundai will), or include a buzzer or warning light when the system activates (Hyundai won‘t). “The reason that ESC works so well is that it helps a driver do what the driver would do intuitively in an ESC situation,” Krafcik says.

Hyundai‘s decision to make ESC technology standard gives the company great economies of scale, according to Krafcik. “If we‘d made ESC an option, we‘d double our build complexity, yet few dealers would order it because few customers are likely to ask for it,” he says. With ESC a standard feature, however, suppliers can anticipate high volume and price accordingly. Krafcik adds that ESC gives Hyundai‘s design engineers more freedom, since they can achieve good limit handling performance without having to compromise for non-ESC models.

Krafcik says that with ESC as a standard feature on its midsize Sonata sedan, he can claim safety leadership positioning against its primary competitors - Toyota Camry and Honda Accord - and still undercut them in price. “Anyone can do what we‘re doing, but since they‘re not, we see (ESC) as an opportunity for us to lead.” It‘s an opportunity he thinks can only last another couple of years by which time stability control is likely to be standard on every vehicle.

SMSC became a major player in the market for MOST (Media Oriented Systems Transport) components in March 2005 when it acquired Oasis Silicon Systems. Since then it‘s sold MOST chips and licensed a number of firms to use its MediaLB interface technology for connecting to MOST networks.

Now it‘s come out with a MOST50 family of products said to operate on MOST networks at 50Mb/s versus 25Mb/s and to run on unshielded twisted pair (UTP) copper wires versus plastic optical fiber (POF).

MOST is popular among European automakers as an effective network protocol for infotainment applications but it hasn‘t gained much traction in the U.S., where cost is a primary concern. It‘s facing competition from IDB-1394 - automotive grade FireWire.

Most if not all MOST networks in Europe use plastic optical fiber (POF), but U.S. automotive design engineers wanted to see it available on UTP, so MOST Cooperation developed a copper wire spec (ePhy) that U.S. OEMs are currently testing.

SMSC says that until now, UTP performance topped out at 1Mb/s because of the radiation-sensitive environment of cars in production today. With ePHY, OEMs can achieve 50Mb/s while continuing to use existing electrical cables and connectors.

Chipmakers are becoming slightly more granular in the way they position their automotive microcontrollers. Freescale and Renesas, to name two, each offer MCUs specifically for telematics. How do you select the one that’s right for your application, and what compromises, if any, do you have to make? To judge from the firms’ respective descriptions, what’s most important is the breadth and depth of application development/debug support, and the optimum combination of peripherals on-chip or readily accessible.

Freescale’s 885 MIPS MPC5200B, introduced at Telematics Detroit last year, is pin- and software-compatible with the MPC5200. It‘s based on Freescale‘s mobileGT architecture/infrastructure (Wind River, QNX, and Green Hills Software, etc.), and supported by the Media5200 Development System, the Lite5200B evaluation board, and various third-party firms.

Application development aids include PowerTAP run control tools; the OSEKturbo operating system; CodeWarrior Development Studio mobileGT Edition; and a Linux Board Support Package (BSP) optimized for the mobileGT architecture and Media5200 development platform.

The Media5200 includes 128MB DDR SDRAM and 64MB flash; an integrated graphics system with an 8.4-inch color LCD; a multi-channel audio subsystem with an AC‘97 sound card; camera input; GPS; integrated CAN, J1850 and MOST networking support; two PCI and one mini-PCI connectors, and ATA, USB, Ethernet, S/PDIF and multiple serial connections.

Anand Ramamoorthy, general manager of Freescale‘s infotainment, multimedia and telematics business, says the MPC5200B‘s single (PowerPC) core is more efficient and less expensive than multi-core alternatives. The chip features a double precision floating point unit; a memory management unit-based architecture with DDR memory support; integrated PCI, ATA and USB buses; multiple serial channels; up to three I2S interfaces, and Ethernet support.

Because all applications execute on a single core, Ramamoorthy says that, for example, features like front-seat navigation can work in tandem with rear-seat entertainment. He adds that the MPC5200B can support audio compression decode/encode, as well as video decode.

Renesas is targeting cost-sensitive telematics applications with its 540 MIPS/2.1 GFLOPS SH7397 telematics chip, the “Euclid.” Based on a 300MHz SH-4A superscale core, the SH7397 is upward code-compatible with the 200MHz, SH-4-based SH7760 (“Camelot”). The new chip is at the high end of Renesas‘ “compact solutions” line. The firm also offers “scalable” and “highly integrated” solutions.

The SH-4A core has two separate 32-Kbyte, 4-way set-associative cache memories, one for instructions and the other for data. Paul Sykes, product marketing manager for telematics, says the combination boosts throughput by improving the cache hit rate. On-chip, fast-access 16-Kbyte RAM also speeds processing.

The SH7397 has a built-in floating point unit, a dedicated bus for connection to external high-speed DDR SDRAM, a color LCD controller capable of approximately 64,000 colors on an 800×600-pixel LCD panel, and a USB interface with a v1.1 host and a v2.0 function controller. The controller enables connections to mobile phones, portable music players, and other consumer devices.

Other peripherals include a 4-channel 10-bit A/D converter, real-time clock (RTC), 6-channel timer (TMU), interrupt controller, and 6-channel direct memory access controller (DMAC) for high-speed data transfers to and from memory.

The SH7397 has a serial sound interface and an audio CODEC interface for transmission/reception of voice and audio data in hands-free applications. Memory card interfaces support MultiMediaCard, SD memory card, PC card, and smart card for exchange and storage of multimedia and other data. The chip includes a CAN interface, and an Ethernet controller that can be used as a general-purpose LAN interface and as a link for software debugging. It also has a 3-channel serial communication interface with FIFO (SCIF), a 3-channel serial I/O with FIFO (SIOF), and a 2-channel I2C bus interface.

Renesas‘ “Sequoia” reference platform provides external memory; debug ports; peripheral functions for audio, display, and CAN; USB, PCMCIA, MMC, smart card, SD memory/IO card and Ethernet interfaces, and support for real-time operating systems (VxWorks, WindowsCE and embedded Linux) and middleware. QNX Software plans to add support for Sequoia to the QNX Neutrino RTOS.

What other devices work well for telematics applications?