Browse Prior Art Database

Improved Detection of Unique Sensitization with Application to Redundancy Identification

IP.com Disclosure Number: IPCOM000103606D
Original Publication Date: 1993-Jan-01
Included in the Prior Art Database: 2005-Mar-18
Document File: 3 page(s) / 159K

Publishing Venue

IBM

Related People

Malik, A: AUTHOR

Abstract

A process for identifying uniquely sensitized nets is disclosed. The process improves Automatic Test Generation (ATG) for the circuit and and expedites identification of redundant faults.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 44% of the total text.

Improved Detection of Unique Sensitization with Application to Redundancy Identification

       A process for identifying uniquely sensitized nets is
disclosed.  The process improves Automatic Test Generation (ATG) for
the circuit and and expedites identification of redundant faults.

      Since pioneering work on the D-algorithm [1,2], many
improvements have been made to ATG [3-7].  The state of the art
algorithms provide much better test coverage with much less
computation.  To a large extent, this is accomplished by an early
determination of the unique values  that must be assigned to some
signals in order to propagate the fault effect to a primary output.
Such signals are known as uniquely sensitized signals.  Early
detection of uniquely sensitized signals avoids assignment of
incorrect values to them which would otherwise require recomputation
of the logic values for those signals that relied on the incorrect
assignments.  Such recomputation is called backtracking.

      In addition to their application in ATG, unique sensitization
has been found to be useful for logic synthesis: they provide a part
of observabilitY don't care set which is useful for the minimization
of the logic network [8].

      The notion of unique sensitization was first introduced in FAN
algorithm [3] as a preprocessing step before the test generation.
The technique was improved [5,6] and a method was provided later to
use it during the test generation [6].  The application during the
test generation as opposed to preprocessing is called the dynamic
application.  However, all of these methods use structure of the
logic network to obtain the uniquely sensitized signals.

      A method is provided that takes advantage of the Boolean
properties of the network to deduce the uniquely sensitized signals.
It can be shown that the uniquely sensitized signals obtained by the
topographical information of the network are a subset of those
obtained by our method.  Our technique uses learning (implications)
already a part of Schultz's algorithm.  Implications are also an
integral part of Global Flow [9] and Dual Global Flow [10] algorithms
for logic optimization.

      A technique can be applied as a preprocessing step as well as
dynamically to cut down on the number of backtrackings.  When applied
statically, it can be used to identify certain redundant faults.  The
removal of the redundancies is one  way to reduce the size of the
network [11].  When applied dynamically, it can indicate the
necessity of a backtrack.  As mentioned above, backtracks are
undesirable and should be avoided.  However, if a backtrack is
inevitable, it should begin as soon as possible in order to terminate
any unnecessary computation in progress.

      As a byproduct of our approach, we also identify signals that
cannot take on certain values during the test generation in order for
the test to exist.  These signals are not necessarily  uniquely
sen...