In this dissertation, we have developed and investigated several data-mining based methodologies that reduce the test cost of an integrated, heterogeneous system by identifying its redundant tests. Our solutions address three problems, namely, to (1) identify and create correlation functions for the redundant tests based on continuous test measurements, (2) identify and create correlation functions for the redundant tests based on binary pass-fail test outcomes, and (3) adaptively re-learn the correlation functions for the redundant tests to guarantee their accuracies.Random errors in the process or test measurement equipment. ... control the DE of Ffc. But YL due to t\ can also be a source of concern, so it too must be monitored to avoid significant cost from rejecting passing chips. ... Note that, since test data used to learn the original F\. is typically collected over a short period of time as compared ... of t\. If these misprediction levels are acceptable, no action is taken.
|Title||:||Reducing Test Execution Cost of Integrated, Heterogeneous Systems Through Data Mining|
|Publisher||:||ProQuest - 2008|