Engineering topics

Blough, Douglas M.

Publications and source records attributed to Blough, Douglas M..

4 records

NASA NTRS1990

Performance analysis of a generalized concurrent error detection procedure

A general procedure for error detection in complex systems, called the data block capture and analysis monitoring process, is described and analyzed. It is assumed that, in addition to being exposed to potential external fault sources, a complex system will in general always contain embedded hardware and software fault mechanisms which can cause the system to perform incorrect computations and/or produce incorrect output. Thus, in operation, the system continuously moves back and forth between error and no-error states. These external fault sources or internal fault mechanisms are extremely difficult to detect. The data block capture and analysis monitoring process is concerned with detecting deviations from the normal performance of the system, known as errors, which are symptomatic of fault conditions. The process consists of repeatedly recording a fixed amount of data from a set of predetermined observation lines of the system being monitored (i.e., capturing a block of data) and then analyzing the captured block in an attempt to determine whether the system is functioning correctly. The performances of linear, quadratic, and logarithmic data analysis algorithms are rigorously characterized in terms of the probability of correctly detecting an error, the expectation and variance of the number of false alarms per error, and the expectation and variance of the latency in detection of errors. Insight into the nature of the general problem of error detection is obtained.

Blough, Douglas M.↗

NASA NTRS1989

Efficient diagnosis of multiprocessor systems under probabilistic models

The problem of fault diagnosis in multiprocessor systems is considered under a probabilistic fault model. The focus is on minimizing the number of tests that must be conducted in order to correctly diagnose the state of every processor in the system with high probability. A diagnosis algorithm that can correctly diagnose the state of every processor with probability approaching one in a class of systems performing slightly greater than a linear number of tests is presented. A nearly matching lower bound on the number of tests required to achieve correct diagnosis in arbitrary systems is also proven. Lower and upper bounds on the number of tests required for regular systems are also presented. A class of regular systems which includes hypercubes is shown to be correctly diagnosable with high probability. In all cases, the number of tests required under this probabilistic model is shown to be significantly less than under a bounded-size fault set model. Because the number of tests that must be conducted is a measure of the diagnosis overhead, these results represent a dramatic improvement in the performance of system-level diagnosis techniques.

Blough, Douglas M.↗

NASA NTRS1988

Fault diagnosis in sparse multiprocessor systems

The problem of fault diagnosis in multiprocessor systems is considered under a uniformly probabilistic model in which processors are faulty with probability p. This work focuses on minimizing the number of tests that must be conducted in order to correctly diagnose the state of every processor in the system with high probability. A diagnosis algorithm that can correctly diagnose the state of every processor with probability approaching one in a class of systems performing slightly greater than a linear number of tests is presented. A nearly matching lower bound on the number of tests required to achieve correct diagnosis in arbitrary systems is also proven. The number of tests required under this probabilistic model is shown to be significantly less than under a bounded-size fault set model. Because the number of tests that must be conducted is a measure of the diagnosis overhead, these results represent a dramatic improvement in the performance of system-level diagnosis technique.

Blough, Douglas M.↗

NASA NTRS1987

Performance analysis of a generalized upset detection procedure

A general procedure for upset detection in complex systems, called the data block capture and analysis upset monitoring process is described and analyzed. The process consists of repeatedly recording a fixed amount of data from a set of predetermined observation lines of the system being monitored (i.e., capturing a block of data), and then analyzing the captured block in an attempt to determine whether the system is functioning correctly. The algorithm which analyzes the data blocks can be characterized in terms of the amount of time it requires to examine a given length data block to ascertain the existence of features/conditions that have been predetermined to characterize the upset-free behavior of the system. The performance of linear, quadratic, and logarithmic data analysis algorithms is rigorously characterized in terms of three performance measures: (1) the probability of correctly detecting an upset; (2) the expected number of false alarms; and (3) the expected latency in detecting upsets.

Blough, Douglas M.↗