News and New Products

US, Europe diverge on automotive-infotainment architectures

SMSC's INIC eLITE promises North American auto engineers the advantages of distributed systems, without making the jump to full-blown MOST.

By Ron Wilson, Executive Editor -- EDN, 4/4/2007

The architectures of automotive infotainment systems differ significantly between US and European design teams, according to Jim Westphal, senior director of strategic marketing at interface-chip supplier SMSC.

Many European manufacturers have enthusiastically embraced the MOST (Media Oriented Systems Transport) interconnect architecture, not only as a way of moving media data between storage and head units, but also as an enabling technology for highly distributed infotainment systems, Westphal says. In the US, however, Westphal still sees auto designers developing big, monolithic head units that contain receivers, storage, processing, amplification, and a display in a single box that can now weigh upwards of five pounds.

So while European engineers are struggling with the design and debug of network-distributed, real-time systems, US designers face a dead end: systems that simply can no longer be implemented by putting all the active circuitry into one big box and then stringing analog cabling around the vehicle. "New designs are adding more audio channels—which means more amplifiers—at the same time that they are adding new media types, new features, and more signal processing," Westphal says. "It just won't fit in the head unit any longer. There isn't space, and there isn't the cooling capacity."

This is causing the North American designers to tentatively begin distributing their systems, Westphal says. But it's not practical for them to take the first steps with new physical-layer media and new, unapproved cable and connector technology. So the designers are falling back on technologies that Detroit has already deployed in volume: copper twisted-pair analog connections and the CAN (controller-area-network) bus for control signals. This eliminates the need to evaluate new technology, such as the low-cost optical interconnect used in the MOST network. But it places further stress on the cable plant and EMI budget of the vehicle.

With its INIC eLITE product, introduced this week at ESC, SMSC is offering a migration path that the company believes can solve the US designers' short-term problem while putting them on the road to full MOST-based distributed systems. The idea is to use a previously developed intelligent MOST NIC (network-interface-controller) chip, but with a twisted-pair physical layer instead of optical fiber, and without the multiple-access MOST protocol stack. Thus the NIC provides point-to-point interconnect for both media and control signals, but with existing electrical connectors and unshielded twisted-pair wiring.

The scheme handles signaling and data transport according to MOST standards: purely digital data in 128-byte frames. The data within a frame may be organized as either streaming-media data or as packet data, allowing free mixing of virtual media and control channels. Without the multiple-access protocol, the intelligence within the NIC chip can handle most or all of the overhead of framing and unframing, so the only significant change to the infotainment-electronics modules is that they now communicate digital data without performing digital-to-analog conversion.

The changes actually may go a little deeper than this. The system is essentially a synchronous broadcast bus, bursting out segmented data streams in real time. This means there is no time for error correction or retries if data arrive damaged. So whereas in an analog interconnect system, interference could result in audible or visible noise, in the digital system lost data could result in disruption to a much more interrelated digital data stream, depending on the type of data. Error-recovery mechanisms in the infotainment modules may have to be strengthened.

But the opportunity to move the underlying architecture from a monolith surrounded by a web of analog connections to at least a digital web could have strong appeal for manufacturers. And the fact that the approach is partway along the road to a full-blown MOST architecture should be appealing as well.

Westphal suggests that the idea may also have its appeal for the aftermarket infotainment vendors, who today have to struggle with installing de-facto distributed systems into a vehicle that was not designed to accommodate them. "Today the aftermarket vendors and installers are trying to drop digital sources and displays into a context of purely analog interconnect," Westphal says. "An approach like ours might be a simplification for them."