Browse Prior Art Database

Mixing Hamming Code With TMR

IP.com Disclosure Number: IPCOM000075671D
Original Publication Date: 1971-Oct-01
Included in the Prior Art Database: 2005-Feb-24
Document File: 2 page(s) / 77K

Publishing Venue

IBM

Related People

Maley, GA: AUTHOR

Abstract

Error-correcting codes such as those devised by Hamming are efficient when storing data in memory or transmitting data over lines. However, codes of this type do not undergo nonlinear transformation without losing their ability to correct errors. Therefore two words that are Hamming encoded cannot be subjected to either And or Or operations and still retain their error-correcting capabilities. In order to realize a fault tolerant computer, it is necessary to convert from a Hamming code to triple modular redundancy TMR and back to Hamming. This circuitry performs conversion to and from TMR in such a way that at no time is the data exposed to errors caused by faulty components. From Hamming to TMR.

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 53% of the total text.

Page 1 of 2

Mixing Hamming Code With TMR

Error-correcting codes such as those devised by Hamming are efficient when storing data in memory or transmitting data over lines. However, codes of this type do not undergo nonlinear transformation without losing their ability to correct errors. Therefore two words that are Hamming encoded cannot be subjected to either And or Or operations and still retain their error-correcting capabilities. In order to realize a fault tolerant computer, it is necessary to convert from a Hamming code to triple modular redundancy TMR and back to Hamming. This circuitry performs conversion to and from TMR in such a way that at no time is the data exposed to errors caused by faulty components. From Hamming to TMR.

The usual Hamming corrector is constructed from Exclusive-Or trees 10...12, And's 20...23, and Exclusive-Or's 24...27. Trees 10, 12, and 14 consist of a plurality of Exclusive-Or's. One of these trees 10...12 exists for each check bit. Each tree produces a signal if that check bit does not agree with its associated data bits. Each Exclusive-Or tree provides true and inverted outputs which are inputs to And's 20...23. The output of each And 20...23 is one of the inputs to each of Exclusive Or's 24...27. The second input to each of these Exclusive-Or's is a data bit. The outputs of Exclusive-Or's 24...27 are corrected data bits. For example, if data bit 7 is in error, each input to And 20 is up providing an up level input to Exclusive-Or 24, thus changing the polarity of erroneous data bit 7.

In order to obtain these data bits in TMR, two additional sets of And's and Exclusive-Or's are used. The first additional set consists of And's 30...33 and Exclusive-Or's 40...43. The second additional set consists of And's 34.....