Image Processing Reference
In-Depth Information
where
B
MAX
is the transmission time of the longest message in the considered network and is, in its
turn, upper bounded by the duration of the largest CAN frame (-byte payload). In this case, the
obtained results are pessimistic.
15.3.3 Usage
Schedulabilityanalysishasbeen(andstillis)widelyusedfordesigningnetworkedembeddedsystem
based on CAN, to check whether or not the network is correctly configured and/or dimensioned.
Itcanbeappliedinallsystemswherecommunicationsarebasedonasynchronousdataexchanges
(either cyclic or sporadic). One of the most interesting (and popular) real-life examples of this kind
of systems are in-vehicle networks used in the automotive domain. In the case of production cars, in
fact, the system is made up of several ECUs (from about up to slightly less than ), each one of
which produces several messages according to either a periodic or a sporadic schedule.
It is worth noting that, from the point of view of communications, data exchanges have been
assumed uncorrelated (indeed, they can be correlated at the application level). In this case, the
previous analysis is perfectly suitable to assess whether or not timing constraints are met.
Besides schedulability analysis, several concepts introduced above have to be considered carefully
in the design phase, particularly when assigning identifiers to different messages that have to be
exchanged in a real-time distributed control application. From an intuitive point of view, the most
urgent messages (i.e., those characterized by either tightest deadlines or higher generation rates)
should be assigned the lowest identifiers (e.g., identifier labels the message which has the highest
priority in any CAN network).
If the period of cyclic data exchanges (or the minimum interarrival time of acyclic exchanges) is
knowninadvance,anumberoftechniquesbasedoneithertheratemonotonicordeadlinemonotonic
approaches,thathaveappearedintheliterature[TINa],canbeproitablyusedtoind(ifitexists)an
optimal assignment of identifiers to messages, so that the resulting schedule is feasible (i.e., deadlines
of all the messages are always respected).
15.4 Considerations about CAN
The medium access technique, which CAN relies on, basically implements a nonpreemptive dis-
tributed priority-based communication system, where each node is enabled to compete directly for
thebusownershipsothatitcansendmessagesonitsown(thismeansthatCANisatruemulti-master
network). his can be advantageous for use in event-driven systems.
15.4.1 Advantages
The CAN protocol is particularly suitable for networked embedded systems, as many (inexpensive)
microcontrollers currently exist that include one (or more) CAN controllers, which can be read-
ily embedded both in existing or new projects. his aspect, together with the fact that the protocol
ismature,stableandverywell-known,makesCANaviablesolutionforseveralyears,inspiteof
newer real-time protocols that feature transmission speeds one-to-two orders of magnitude higher
than CAN.
Compared to traditional field networks, CAN is by far simpler and more robust than the token-
based access schemes (such as, for example, PROFIBUS when used in multi-master configurations).
In fact, there is no need to build or maintain the logical ring, nor to manage the circulation of the
token around the master stations. In the same way, it is noticeably more flexible than solutions based
on the time division multiple access (TDMA) or combined-message approaches—two techniques
adopted by SERCOS and INTERBUS, respectively. This is because message exchanges do not have
