Image Processing Reference
In-Depth Information
19.2 Volcano Concepts
he Volcano concept is founded on the ability to guarantee the worst-case latencies of all frames sent
inamultiprotocolnetworksystem.hisisakeystepbecauseitgivesthefollowing:
•
A way of guaranteeing that there are no communications-related timing problems.
•
A way of maximizing the amount of information carried on the bus. The latter is
important for reduced production costs.
•
The possibility to develop highly automated tools for the design of optimal network
configurations.
The timing guarantee for CAN is provided by mathematical analysis developed from academic
research []. Other protocols like FlexRay are predictable by design. For this reason, some of the
subjects discussed below are CAN specific; others are independent of the protocol used.
The analysis is able to calculate the worst-case latency for each frame sent on the bus. his latency
is the longest time from placing a frame in a CAN controller at the sending side to the time the frame
is correctly received at all receivers.
The analysis needs to make several assumptions about how the bus is used.
Oneoftheseassumptionsisthatthereisalimitedsetofframesthatcanaccessthebus,andthat
time-related attributes of these frames are known (e.g., frame size, frame periodicity, queuing jitter,
and so on).
Another important assumption is that the CAN hardware can be driven correctly:
•
Internal message queue within any CAN controller in the system is organized (or can
be used) as such that the highest priority message will be sent out first if more than one
message is ready to be sent. (The hardware-slot position-based arbitration is OK as long
as the number of sent frames is less than the number of transmit-slots available in the
CAN controller.)
•
CAN controller should be able to send out a stream of scheduled messages without releas-
ing the bus in the interframe space between two messages. Such devices will arbitrate for
the bus right after sending the previous message and will only release the bus in case of
lost arbitration.
A third important assumption is the “error model”: the analysis can account for retransmissions due
to errors on the bus, but requires a model for the number of errors in a given time interval.
The Volcano software running in each ECU controls the CAN hardware and accesses the bus so
that all these assumptions are met, allowing application software to rely on all communications taking
place on time. This means that integration testing at the automotive manufacturer can concentrate
on functional testing of the application software.
Another important benefit is that a large amount of communications protocol overhead can be
avoided. Examples of how protocol overheads are reduced by obtaining timing guarantees are
•
There is no need to provide frame acknowledgment within the communications layer,
dramatically reducing bus traic. he only case where an ECU can fail to receive a frame
via CAN is if the ECU is “off the bus,” a serious fault that is detected and handled by
network management and on-board diagnostics.
•
Retransmissions are unnecessary. he system level timing analysis guarantees that a frame
will arrive on time. Timeouts only happen after a fault, which can be detected and handled
by network management and/or the on-board diagnostics.
A Volcano system never suffers from intermittent overruns during correct operation because of the
timing guarantees, and therefore achieves these efficiency gains.
