Browse Prior Art Database

Dual Processing Approaches to Fault Detection and Isolation

IP.com Disclosure Number: IPCOM000080937D
Original Publication Date: 1974-Mar-01
Included in the Prior Art Database: 2005-Feb-27
Document File: 4 page(s) / 138K

Publishing Venue

IBM

Related People

Donovan, GF: AUTHOR [+3]

Abstract

A goal of 100% fault detection and isolation for each of two failures in two of three redundant computer sets is approached by a "dual process" (which does not allow exchange of data between computer sets) or a "modified dual process" (which does allow such exchange).

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 59% of the total text.

Page 1 of 4

Dual Processing Approaches to Fault Detection and Isolation

A goal of 100% fault detection and isolation for each of two failures in two of three redundant computer sets is approached by a "dual process" (which does not allow exchange of data between computer sets) or a "modified dual process" (which does allow such exchange).

Three computer sets 1, 2, 3 (Fig. 1) all are normally operative and execute a prescribed program in approximate synchronism. Set 1 is designated as the "master" and is in control; but remaining sets 2, 3 concurrently perform similar functions and are available to take over control, one at a time, whenever the master relinquishes control due to a failure.

In the dual-process approach (Fig. 2), sensor data is stored directly in the main and extended memories by the I/0 Unit upon command from the computer program. Parity checks are performed on memory addresses and data, and echo checks are performed on the command. Two sets of computations are made sequentially, first from main memory data, then from extended memory data, and each is followed by a computer self-test operation. While the program statements are being executed, a dynamic sum check is performed by a computer microprogram at branch points in the program. The sum check constants will have previously been generated by the compiler and inserted in main line code following store and branching instructions.

The final results, which are commands to the system, are contained in two locat...