Image Processing Reference
In-Depth Information
Cycle time
Bus master
Slave 1
Slave 2
Slave 3
Slave 4
Slave 5
Synchronization, start of cycle
Slots for periodic traffic
Aperiodic traffic
FIGURE .
Centralized TDMA scheme (e.g., SERCOS).
Segmented
data
Parameter data
Process data
Slave 6
FCS
LBW
Slave 1
Slave 2
Slave 3
Slave 4
Slave 5
Slave 6
FCS
LBW
Slave 1
Summation frame
LBW: Loop-back word, start of frame
FCS: Frame checking sequence
FIGURE .
Summation frame of INTERBUS.
accordancewithitspositioninthering.Outputdatafromthemastertotheslaveandinputdata
fromtheslavebackareexchangedintherespectiveslotastheframeisshitedthroughthering.hus
one data frame is sufficient for the cyclic updating of all end devices. A clever arrangement of buffers
furthermore ensures that despite the ring structure with its inherent delays, all I/Os at the slaves are
updated at the same time, so that synchronous operation with respect to the process variables is guar-
anteed. Aperiodic traffic can be introduced in this rather rigid scheme through a so-called parameter
channel. Larger amounts of data are transferred in small packets in this channel without affecting the
exchange of cyclic process data.
The second method of managing the time slots is a decentralized one. In this case, there is no
dedicated node to initiate the cycle, rather all devices synchronize themselves either by explicit clock
synchronization mechanisms or by a set of timers that settle bus operation down to a stable steady
state. Fieldbus systems relying on such distributed mechanisms have a high degree of fault tolerance
because there is no single point of failure as in centralized approaches. herefore, they are well suited
for safety-critical applications (such as TTP/C, FlexRay, or ARINC ), provided that they are
designed to have real-time capabilities. Nevertheless, the underlying algorithms are relatively com-
plex, which is a why such systems have attracted a lot of scientific interest. The distributed nature
