Diagnosis: Problem Statement
Fault diagnosis and testing play a fundamental role for dependable systems in order to identify faulty components. In particular
automotive applications have severe cost constraints but demand a high level of safety and performance, hence timely diagnosis of faulty
elements under deadline and resource constraints is of utmost importance. An approach for fault diagnosis and testing is already
successfully employed in the semiconductor industry, namely Built-in Self-Test (BIST). With a tester located inside the circuit under test
the controllability/observability problem of the nodes within a circuit is quite easily solved, which keeps test time short and hence
tester cost low. In the area of embedded computers in automobiles the testing problem is being faced as well in a very similar way:
Electronic systems in cars form a complex structure of distributed computing nodes. While accessibility and observability are quite
sufficient during development, it is very hard to verify the correct interaction of a given set of (even though pre-tested) components in a
specific type of car, even with a time-triggered communication approach. At this point testing again becomes very ineffective and
expensive, and again a consequent hierarchical test approach (node/cluster/system) is desirable. A BIST-like strategy on the node level has
already been successfully employed in avionic applications, therefore it is reasonable to translate this approach to the automotive area as
well. The technical challenge in such a translation, however, lies in the fact that automotive systems must be extremely cost efficient and
cannot afford a degree of redundancy comparable to that in avionic systems. Besides facilitating system integration a BIST strategy can
substantially save cost of maintenance and even allow on-line diagnosis, which may become relevant to cope with the even higher complexity
of embedded automotive systems of the future.