Civil Engineering Reference
In-Depth Information
6.6.3
Simulating the CAN Network
A functional test for a CAN ECU requires the involvement of the other devices
on the bus for two reasons: Firstly, the ECU only works correctly when its infra-
structure, e.g. NM, is present and functioning correctly. Secondly, many errors in a
networked device can only be observed when it is operating under the same condi-
tions as would apply in a real network. This includes, e.g. the various signals sent
cyclically by other ECUs which the SUT must interpret in parallel to the test task. It
goes without saying that such loads can impact the SUT's behaviour. In the automo-
tive industry, different vehicle ECU's are typically developed by several suppliers
at the same time. For this reason, the ECUs of other manufacturers are generally not
available during the development phase. To carry out the necessary testing despite
this, the remaining CAN network is usually simulated during this phase.
It makes sense to generate the so-called remaining bus simulation from a suitable
network description. The Vector Data Base for CAN communication (DBC) format
is widely used for this purpose, as it retains not only messages and signals, but also
attributes such as cycle times. In other words, it provides a complete description of
the CAN communication, and is often referred to as a database or K-Matrix.
21
With a remaining bus simulation, it is not necessary to simulate the actual appli-
cation itself. This means, e.g. that it is sufficient for an ABS ECU in a simulated or
partially simulated CAN vehicle network to receive signals relating to wheel revo-
lutions, brake pressure, brake pedal position, etc. and to transmit defined control
signals with appropriate values and in compliance with the specified cycle times
and protocols. It is not necessary to implement the actual ABS braking algorithm
for this purpose.
Remaining bus simulations do not just serve as a simple replacement for CAN
network components that are not yet available. They also make it easier to simulate
error situations such as the absence or loss of individual messages, which could
not be as easily implemented with a real ECU. It is the flexibility of the software-
based simulation that is so helpful here, inasmuch as it is easily controlled by the
test program. This makes it easy to implement controls for the participating ECUs
needed for functional tests in the test program. This is an important prerequisite for
test automation.
A remaining bus simulation can also be used prior to running an actual test on a
real ECU. The simulation makes it possible to detect possible problems in a CAN
network early on and to analyse the impact of changes to the communication def-
inition. The communication definition, in combination with the very illustrative
remaining bus simulation, also serves as a specification document for the develop-
ment of individual CAN ECUs.
21
K-Matrix stands for communication matrix because the database also describes which signal is
sent and received by which node.
