The main factors driving the development of communication networks in distributed-controller systems are increased functionality, legislation and better system performance in vehicle applications. To fulfill requirements such as increasing data rate, vehicle life and reliability, vehicle manufacturers need efficient coupling of subsystems by an error-tolerant bus system. The standardized linear bus architecture of the controller-area network (CAN) fulfills high-end automotive demands. It can be used as an embedded or open system for all industrial applications focusing on high configuration flexibility, real-time functionality at high transmission rates and error correction. Due to its hierarchical organization, the control module can be located as close as possible to the mechanical unit that it controls. This provides for considerable expansion of the data exchange between control units without involving a similar increase in hardware outlay.

High-end requirements are defined by closed-loop, real-time communication of automotive applications such as vehicle dynamics control (VDC), engine and transmission management.

• High transfer rates and network expansion:
  Typical process cycle times for engine management or VDC are in the range of a few milliseconds. In this time slot the data has to be sensed, conditioned, transmitted and processed by all receivers. Corresponding transfer rates up to 1 Mbit/second for powertrain functions have been realized in CAN for a maximum network expansion of 40 meters (approximately 120 feet).
  
  
• Availability:
  CAN operates according to the multimaster principle to ensure a high level of overall system availability. This principle is given maximum support by the linear bus topology, which is represented with only one logical communication line. A passive-star structure represents a worthwhile alternative if optical transmission media are used.
  
  
• Communications in the millisecond range under real-time conditions:
  The transmission of at least one message for time-critical controls must be guaranteed even if the message load in the network is high. Timing demands for body electronics applications or mobile communication control data (10 to 100 milliseconds) are typically smaller in magnitude when compared with the engine- management or VDC signal processing (1 to 3 ms).

  This reduced demand is caused by human reaction times but has to be evaluated carefully in case of closing modulation loops via the communication channel. Domain-specific demands, such as high data rates in mobile communications (audio/video), have to be distinguished from continuous vehicle-control data streams. An interdomain exchange by gateways is actually being discussed.

• Configuration flexibility:
  Various levels of vehicle equipment and expansion of existing electronic systems demand a flexible communication network. As a result, the nodes must be independent from each other (multimaster principle) and have to be addressed depending on the message content. By adding a content-related address with message acceptance check in each node, an efficient message filtering is realized. The bus access priority is user-defined by an identifier coded within the message.
  
  
• CAN not only recognizes errors-it also corrects them:
  Locally disturbed messages (for example, by electromagnetic radiation) are marked for all nodes in the system for data consistency reasons. A system-wide unique error flag permits every node to abort faulty messages during transmission. A localization of faulty nodes is performed by statistical analysis of permanent or temporary errors. In extreme cases, a faulty node switches itself off. In communication systems with message-oriented bus access, a message collision may occur. These message collisions, due to asynchronous transfer mode, must not disturb the message data. Rather, they must select and transmit the message with higher priority. This handling is performed in CAN by a nondestructive bit-wise arbitration method. Messages with recognized failures are repeated automatically.

  CAN offers considerable advantages with regard to the optimization of a distributed system. Application-specific hardware implementations from different chip providers can take over the communication part of system design and offer the systems engineer full advantage of the CAN possibilities at reasonable cost. CAN is increasingly being used in industrial field-bus systems. Examples of nonautomotive installations where CAN has proved its capabilities are medical electronics, machine tools (CNC), robotics, building-services management systems, textile machines, long-distance and local transport systems, elevators and electric stairways, testbenches and car-wash systems.

• CAN will support time-triggered operating systems:
  Classic event-triggered CAN peak loads may occur when the transmission of several messages is requested at the same time.

Increasing real-time interaction of different subnet functions (driver assistance, engine management, navigation, mobile communication) demands a detailed scheduling analysis of the whole system.

Some real-time operating systems (RTOSes) are based on static cyclic scheduling of all tasks in the application system (control unit). They build a schedule of time slots and place each task in at least one slot. For distributed-controller systems, system integration and composition of such an RTOS is served when the communication on the CAN follows a synchronized schedule.

See related chart