Can We Eliminate Fault Escape in Self-Testing by Polynomial Division (Signature Analysis) ?

This paper is based on a search for an ideal signature analysis scheme that will ensure that the detected faults do not escape due to error masking. We present a number of general techniques based on linear feedback shift registers and characterize them for their impact on error escape and fault escape. We show that, under certain conditions of uniformity, the error escape probability of any general data compression scheme is governed by a fundamental information theoretic lower bound. We then demonstrate that the techniques suggested in this paper conform to the established lower bound, but elimination or even reduction of fault escape by them cannot be guaranteed.

[1]  Edward J. McCluskey,et al.  Design for Autonomous Test , 1981, IEEE Transactions on Computers.

[2]  Jacob Savir,et al.  On Random Pattern Test Length , 1984, IEEE Transactions on Computers.

[3]  Theo J. Powell,et al.  Analysis and Simulation of Parallel Signature Analyzers , 1987, ITC.

[4]  James E. Smith,et al.  Measures of the Effectiveness of Fault Signature Analysis , 1980, IEEE Transactions on Computers.

[5]  Dilip K. Bhavsar,et al.  Self-Testing by Polynomial Division , 1981, ITC.