Talking to the car

Automotive networks are in the throws of change. Gabriel Leen and Donal Hefferman of PEI Technologies and the University of Limerick give a running commentary of the industry’s movements.

Digital networks are revolutionising automotive control system design, from safety critical drive-by-wire to high bandwidth multi-media entertainment systems.

Traditionally, in-vehicle signalling between simple devices such as switches and actuators was achieved using point-to-point wiring; resulting in bulky, expensive, and complicated wiring harnesses. The amount of wiring grew to a stage where its bulk became a problem. Figure 1 shows the growth of vehicle wiring in Volvo passenger cars over nearly eight decades.

The problems of bulk wiring include; reduced layout space, assembly difficulties, reduced serviceability, costs failing to outweigh benefits of extra wiring, fuel efficiency demands weight reduction, and numerous connectors cause unreliable operation with each link, reducing mean time between failure.

During the 1980’s in-vehicle control networking, based on serial multiplexed buses, emerged and started to replace bulk wiring. Many of the technical concepts for vehicle networking were borrowed from developments in computer data networks, but vehicle communications requirements are driven by control strategies rather than by classical data transfer strategies.

In the US the SAE (Society of Automotive Engineers) has formally classified vehicle networks based on their bit transfer rates (Table 1). Class D networks are not exactly defined, but networks exceeding a data rate of 1Mb/s are considered Class D networks.A vehicle is an extremely hostile environment for electronic equipment, providing vibration, temperature swings from –40°C to +80°C, oil splashing, petrol and water infiltration, electromagnetic fields (automotive can be >200V/m, domestic is 3V/m and industrial 10V/m), electrical spikes/transients of both polarities(>±100V), load dumps, jump starts, high humidity, dust, potential mis-wiring of electrical systems like short circuits to ground or positive, reverse battery, and, yes, sand storms.

Automotive solutions also need special design considerations such as:

High Integrity: The probability of an undetected error must be negligible for the life of the vehicle.

Bounded determinism: A guaranteed upper level on message latency time for control problems.

EMC compliance: Both emitted radiation levels and absorption levels must be met.

Low interconnection count: Each additional connector increases the probability of a fault.

Compact connectors: The connector is often the largest component in an automotive electronic module.

Low cost: A saving of a few pence per component amounts in high volume production.

Network evolution: Variations across models require a network which is easily expanded and modified.

Fault tolerance: Communication must be restored when faults are removed and redundancy is also important.

Control networks must be reliable, responsive networks, which can be applied to critical real-time control applications such as powertrain control and vehicle dynamics. Control networks are also used in somewhat less critical applications such as comfort electronic systems concerned with control of power windows, adjustable seats, cabin temperature, etc.

Drive-by-wire networks are a special class of control network where the network design needs to be ultra reliable.

High bandwidth networks are used for vehicle multimedia applications, where cost is not excessively critical.

Control networks

Automotive control networks operate at moderate data rates. CAN (Controller Area Network) and J1850 are currently two of the most successful standards for vehicle control networks. The de facto standard for vehicle control networks is CAN.

CAN was developed by Robert Bosch GmbH in the mid 1980’s and was first implemented in a Mercedes Benz S-class car, in 1991. CAN has since been adopted by most major European automotive manufacturers and a growing number of US companies are now using it.

Car manufacturers Audi AG, BMW AG, DaimlerChrysler AG, Volvo Car Corporation AB, and Volkswagen AG, the communications specialist Volcano Communications Technologies AB, and the semiconductor manufacturer Motorola recently formed of an industry consortium to implement an open standard for class-A serial buses.

The standard has been named Local Interconnect Network (LIN). Typical applications for the LIN bus are assembly units such as doors, steering wheel, seats, climate regulation, lighting, rain sensor, or alternator. The commonly used analog coding of signals will be replaced by digital signals, leading to an optimised wiring harness.

Drive-by-wire networks

Electronic modules and electric motors are replacing the equivalent mechanical systems in critical control systems, eliminating power steering pumps, hoses, environmentally unfriendly hydraulic fluid, inefficient drive belts, pulleys, and brake servos. Systems like E-Steer (Steer-by-wire) from Delphi Automotive Systems will be seen, soon, in high volume vehicles. A brake-by-wire solution will be available in a German commercial passenger car in the year 2002.

In Europe the X-By-Wire Consortium is an EU-funded BRITE-EURAM research project, which has resulted in the design of a novel Time-Triggered Architecture (TTA) network, where channel access control is based on a TDMA (Time Division Multiple Access) scheme, derived from a fault tolerant common time base. Fundamental to this strategy is the communications protocol: the TTP (Time-Triggered Protocol). The project is now being commercialised with the help of the Austrian company TTTech.

The CAN community is currently working on a layer standard which will allow CAN to compete in drive-by-wire solutions. This TTCAN (Time Triggered CAN) solution will include clock synchronisation for true time triggered behaviour.

BMW, in conjunction with some semiconductor companies, has developed its ’byteflight’ system (formerly called SI bus) as a drive-by-wire network solution, which is ready for commercial implementation.

A number of companies such as Microsoft, Saab, Mecel, Intel, Clarion and others are working on their vision for entertainment, navigation and business applications, and have demonstrated their work in the Personal Productivity Vehicle. Meanwhile, IBM, Delco, Netscape and Sun Microsystems are developing the Network Vehicle which uses Java as its operating environment. The development of such in-car personal computers will support features such as: voice activated control, internet access from the car, text to speech e-mail reading; voicemail, and auto-route planning with real-time updates using traffic reports from car radio RDS or the Web.

A new class of vehicle network is emerging to connect the forthcoming in-car personal computers and their peripherals. The Optical Chip Consortium has specified a network called Domestic Digital Bus which is a fibre optic solution offering approximately 12 Mb/s of bandwidth, currently used in the new Mercedes S-Class. Oasis Silicon Systems has developed the Media Orientated Systems Transport solution, again with a fibre optic physical layer, giving a transfer rate of 25 Mb/s. And Delphi Automotive Systems provides a solution in the form of MML with an impressive transfer rate of 100 Mb/s.

Toyota, GM, Ford, Daimler, Chrysler and Renault founded Automotive Multimedia Interface Collaboration to standardise the vehicle multimedia architecture for the 21st century.

There is a clear incentive for companies to have their technologies incorporated into the vehicle multimedia standards and Microsoft is making a significant effort with the Windows CE based Auto PC.

From a network perspective it is probable that an IDB-C (Intelligent Data Bus-CAN) solution will become one of the adopted standards. IDB-C is based on CAN’s physical and data link layers, but will be complemented by a higher speed multimedia bus, almost certainly based on fibre optic media.