Image Processing Reference
In-Depth Information
The CAN specification foresees five different mechanisms to detect transmission errors:
. Cyclic redundancy check : When transmitting a frame, the originating node appends a -
bit-wide CRC to the end of the frame. Receiving nodes reevaluate the CRC to check if
it matches the transmitted value. Generally speaking, the CRC used in CAN is able to
discover up to  erroneous bits arbitrarily distributed in the frame and also errors bursts
including up to  bits.
. Frame check : Fixed-format fields in the received frames can be easily tested against their
expected values. For example, the CRC and ACK delimiters as well as the EOF field have
to be at the recessive level. If one or more illegal bits are detected, a form error is generated.
. Acknowledgment check : he transmitting node verifies whether the ACK bit in the frame
that is being sent has been set to the dominant value. If this does not occur, an acknowl-
edgment error is issued. Such a kind of check is useful to discover whether or not there are
other active nodes in the network (i.e., if the link of the transmitting node is not broken).
. Bit monitoring : Each transmitting node compares the level on the bus to the value of the bit
which is being written. Should a mismatch occur, an error is generated. It is worth noting
that the same does not hold for the arbitration field and the acknowledgment slot. Such an
error check is very effective to detect local errors that may occur in the transmitting nodes.
. Bit stuffing : Each node verifies whether the bit stuffing rules have been violated in the
portion of the frames from the SOF bit up to the CRC sequence. In the case  bits of
identical value are read from the bus, an error is generated.
The residual probability that a corrupted message goes undetected in a CAN network—under realistic
operating conditions—has been evaluated to be about .

 times the error rate or less.
×
15.2.5 Fault Confinement
So as to prevent a node that is not operating properly from sending corrupted frames repeatedly,
thus blocking the whole network, a fault confinement mechanism has been included in the CAN
speciication.hefaultconinementunitsupervisesthecorrectoperationoftherelatedMACsublayer
and, should the node become defective, it disconnects that node from the bus.
hefaultconinementmechanismhasbeenconceivedsoastodiscriminate,aslongasitispossible,
between permanent failures and short disturbances which may cause bursts of errors on the bus.
According to this mechanism, each node can be in one of the three following states:
Error active
Error passive
Bus-off
Erroractiveanderrorpassivenodestakepartinthecommunicationinthesameway.However,they
react to error conditions diferently. hey send active error flags in the former case and passive error
flags in the latter. his is because an error passive node has already experienced several errors, so it
should avoid interfering with the network operations (a passive error flag, in fact, cannot corrupt the
ongoing frame exchange).
hefaultconinementunitusestwocounterssoastotrackthebehaviorofthenodewithrespect
to the transmission errors: they are known as transmission error count (TEC) and receive error
count (REC), respectively. The rules by which TEC and REC are managed are actually quite com-
plex. However, they can be summarized as follows: each time an error is detected, the counters are
increased by a given amount, whereas successful exchanges decrease them by one. Furthermore, the
amount of the increase for the nodes which first detected the error is higher than the nodes that sim-
ply replied to the error flag. In this way it is very likely that counters of faulty nodes increase more
 
Search WWH ::




Custom Search