Image Processing Reference
In-Depth Information
Original bit stream
(TX node)
0
0
1
0
000 11111
00
1
00000
1111
00
00 1111
0 0
Level on the bus
0
0
1
0
000
11111
0
00
1
00000
1
1
111 0
0
0
0 0 1
1
1
1
1
0
0
0
Decoded bit stream
(RX node)
0
0
1
0
000
11111
00
1
00000
1111
00
00 1111
00
FIGURE .
Bit stuffing technique.
the complementary value, as depicted in Figure .. Stuff bits can be easily and safely removed by
receiving nodes, so as to obtain the original stream of bits back.
From a theoretical point of view, the maximum number of stuff bits that may be added is one every
bits in the original frame, so the encoding efficiency can be as low as % (see, for example, the
rightmostpartofFigure.,wheretheoriginalbitstreamalternatessequencesofconsecutivebits
at the dominant level followed by bits at the recessive level). However, the influence of bit stuffing in
real operating conditions is noticeably lower than the theoretical value mentioned above. Simulations
show that, on average, only to stuff bits are effectively added to each frame, depending on the size
of the identifier and data ields. Despite it is quite efficient, the bit stuffing technique has a drawback.
In particular, the time taken to send a message over the bus is not fixed; instead it depends on the
content of the message itself. his might cause unwanted jitters.
Not all the fields in a CAN frame are encoded according to the bit stuffing mechanism: in par-
ticular, it applies only to the initial part of the frames, from the start of frame (SOF) bit up to the
cyclic redundancy check (CRC) sequence. Remaining fields, instead, are of fixed form and are never
stuffed.
15.2.2 Frame Format
The CAN specification [ISO] defines both a standard and an extended frame format. hese formats
mainly differ for the size of the identifier field and for some other bits in the arbitration field. In
particular, the standard frame format (that is also known as CAN .A format) defines an bit
identifier field, which means that up to different identifiers are available to applications running
in the same network (many older CAN controllers, however, only support identifiers in the range
from to ). ).The extended frame format (identified as CAN .B), instead, assigns bits to the
identifier, so that up to half a billion different objects could exist (in theory) in the same network.
This is a fairly high value, which is virtually sufficient for any kind of application.
Using extended identifiers in a network that also includes .A compliant CAN controllers, usually
leads to unmanageable transmission errors, which effectively make the network unstable. Thus, a
third category of CAN controllers was developed, known as .B passive: they manage in a correct
way the transmission and the reception of CAN .A frames, whilst CAN .B frames are simply
ignoredsothattheydonothangthenetwork.
It is worth noting that, in most practical cases, the number of different objects allowed by the
standard frame format is more than adequate. Since standard CAN frames are shorter than extended
frames (because of the shorter arbitration field), they enable higher communication efficiency (unless
part of the payload is moved into the arbitration field). As a consequence they are adopted in most of
the existing CAN systems, and also most of the CAN-based higher-layer protocols, such as CANopen
and DeviceNet, basically rely on this format.
The CAN protocol foresees only four kinds of frames, namely data, remote, error and overload
frames. heir format is described in detail in Sections ... through ....
