Performance measures of data compression techniques in fault detection
Date
1986
Authors
Aitken, Robert Campbell
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The testing of digital circuits comprises a significant and growing percentage of their total production cost. Data compression is a set of testing techniques which can be shown to substantially reduce the expense of fault detection. Further, data compression is amenable to design for testability methodologies, as well as built- in self- test designs.
New performance measures are derived for two major data compression schemes: LFSR signature analysis and testing based on Rademacher-Walsh spectral coefficients. Results are based on both a standard theoretical framework and empirical values from a set of sample circuits. The empirical evidence obtained allows analysis of both the techniques and the validity of the theoretical framework.
The relative merits of various methods are given, based on the theoretical and empirical evidence presented. It is shown that those which include the syndrome of a function are particularly useful because of the deterministic bounds on fault coverage which can be obtained for them.