Browse Prior Art Database

Error Masking

IP.com Disclosure Number: IPCOM000091204D
Original Publication Date: 1969-Nov-01
Included in the Prior Art Database: 2005-Mar-05
Document File: 2 page(s) / 27K

Publishing Venue

IBM

Related People

Duke, KA: AUTHOR [+2]

Abstract

A computer or other digital processing system can be built to overcome component failures by duplicating each unit. When a particular unit gives an erroneous result, it is replaced by its duplicate and the process is repeated to obtain the correct result. While a high degree of reliability can be achieved by such a system, recovery from a failure involves temporarily suspending normal operation while the erroneous process is repeated. In many situations, such suspension is not allowable. This error masking technique allows recovery from error-producing failures without the suspension of normal operation.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 75% of the total text.

Page 1 of 2

Error Masking

A computer or other digital processing system can be built to overcome component failures by duplicating each unit. When a particular unit gives an erroneous result, it is replaced by its duplicate and the process is repeated to obtain the correct result. While a high degree of reliability can be achieved by such a system, recovery from a failure involves temporarily suspending normal operation while the erroneous process is repeated. In many situations, such suspension is not allowable. This error masking technique allows recovery from error-producing failures without the suspension of normal operation.

In drawing A, M1 and M2 are identical functional units, simultaneously performing the same operations on identical data. N1 is a unit whose input source is the output of either M1 or M2. These outputs are coded in an error- detecting code. M1 and M2 are so arranged that an internal failure, that affects an output, puts the output into the error space. S1 is a switching network which selects the source of input to N1. S1 is controlled by the error-detecting network D1. While both M1 and M2 are functioning correctly, either source can be selected and used as input to N1. If a failure occurs which produces an erroneous output from either M1 or M2, the error-detecting network causes S1 to select a correct output. If N1 is replicated, then S1 and all or part of D1 have to be replicated as in B.

M and N can be replicated any number of times to achieve the...