Coverage-Based Testing on Embedded Systems

One major issue of code coverage testing is the overhead imposed by program instrumentation, which inserts probes into the program to monitor its execution. In real-time systems, the overhead may alter the program execution behavior or impact its performance due to its strict requirement on timing. Coverage testing is even harder on embedded systems because of their critical and limited memory and CPU resources. This paper describes a case study of a coverage-based testing method for embedded system software focusing on minimizing instrumentation overhead. We ported a code coverage-based test tool to an in-house embedded system, IP phone. In our initial experiments, we found that this tool didn ´t affect the behavior of the program under test.