Image Processing Reference
In-Depth Information
Sync
Arbitration loss
Node 1
Node 2
Node 3
t
Bus
FIGURE .
BA method with CAN.
remaining node is the one whose message has the lowest identifier (Figure .). The BA method
brings with it a major disadvantage. The propagation time of the signals on the line must be short
compared to the bit time to yield quasi-simultaneity for all nodes. With the highest bit rate of Mbit/s,
this means a maximum bus length of only m. In motor vehicles (the original application field of
CAN), this restriction is of lesser importance but in industrial automation technology it can lead to
a reduction in the data transfer rate.
As with decentralized TDMA, faulty nodes may infinitely block the bus. To prevent this situation,
CAN nodes contain error counters which are incremented whenever the node detects transmit or
receive errors and are decremented after successful transmit or receive procedures. he counter status
reflects the reliability of the node and determines if the node may fully participate in bus traffic or
only with certain restrictions. In the extreme case, it is completely excluded from the bus.
BA was so successful that it was used in several other fieldbus systems in similar form. Exam-
ples are building automation networks like EIB, BATIBUS, or EHS, as well as in other automotive
networks like VAN and FlexRay (for aperiodic traffic). But also CAN was used as a basis for fur-
ther extension. CAN as such originally only defined layers and in the OSI model. Although
this was sufficient for the exchange of short messages within a closed network, it was insufficient
for automation technology. For this reason, the CAN-in-automation user group (CiA) defined
the CAN application layer (CAL) and then the CANopen protocol. Other protocols for automa-
tion technology, also based on CAN, are DeviceNet and SDS. The CAN Kingdom protocol has
been specially developed for machine controls and safety-critical applications. These higher-level
protocols offer the possibility of exchanging larger volumes of data and of synchronizing end
devices. Network management functions solve the problems of node configuration and identifier
specification.
CSMA-CA has one inherent problem. Even in the absence of babbling idiots, the highest priority
object can practically block the bus and messages with lower priorities seldom get through. herefore,
QoS guarantees can only be given in a strict sense for the message with the highest priority if no addi-
tional measures are taken. A number of solutions have been proposed to overcome this problem. One
of them is to limit the frequency of the messages such that after a successful transmission, a node has
to wait a certain amount of time before being able to send again. Another possibility is to introduce
mechanisms in the upper protocol layers to exchange identifiers cyclically within specific priority
classes in addition to restrictions in access frequencies. This is a network management function
used, e.g., in CANopen. The third method of improving fairness (and thus real-time capabilities)
is to superimpose time-slot mechanisms at least for some message and traffic classes. his strategy is
being employed by TT-CAN and FTT-CAN.
