Engineering topics

Knight, J. C.

Publications and source records attributed to Knight, J. C..

At least 19 records

NASA NTRS1997

Development of a Software Safety Process and a Case Study of Its Use

Research in the year covered by this reporting period has been primarily directed toward the following areas: (1) Formal specification of user interfaces; (2) Fault-tree analysis including software; (3) Evaluation of formal specification notations; (4) Evaluation of formal verification techniques; (5) Expanded analysis of the shell architecture concept; (6) Development of techniques to address the problem of information survivability; and (7) Development of a sophisticated tool for the manipulation of formal specifications written in Z. This report summarizes activities under the grant. The technical results relating to this grant and the remainder of the principal investigator's research program are contained in various reports and papers. The remainder of this report is organized as follows. In the next section, an overview of the project is given. This is followed by a summary of accomplishments during the reporting period and details of students funded. Seminars presented describing work under this grant are listed in the following section, and the final section lists publications resulting from this grant.

Knight, J. C.↗

NASA NTRS1996

Development of a Software Safety Process and a Case Study of Its Use

Research in the year covered by this reporting period has been primarily directed toward: continued development of mock-ups of computer screens for operator of a digital reactor control system; development of a reactor simulation to permit testing of various elements of the control system; formal specification of user interfaces; fault-tree analysis including software; evaluation of formal verification techniques; and continued development of a software documentation system. Technical results relating to this grant and the remainder of the principal investigator's research program are contained in various reports and papers.

Knight, J. C.↗

NASA NTRS1991

A research program in empirical computer science

During the grant reporting period our primary activities have been to begin preparation for the establishment of a research program in experimental computer science. The focus of research in this program will be safety-critical systems. Many questions that arise in the effort to improve software dependability can only be addressed empirically. For example, there is no way to predict the performance of the various proposed approaches to building fault-tolerant software. Performance models, though valuable, are parameterized and cannot be used to make quantitative predictions without experimental determination of underlying distributions. In the past, experimentation has been able to shed some light on the practical benefits and limitations of software fault tolerance. It is common, also, for experimentation to reveal new questions or new aspects of problems that were previously unknown. A good example is the Consistent Comparison Problem that was revealed by experimentation and subsequently studied in depth. The result was a clear understanding of a previously unknown problem with software fault tolerance. The purpose of a research program in empirical computer science is to perform controlled experiments in the area of real-time, embedded control systems. The goal of the various experiments will be to determine better approaches to the construction of the software for computing systems that have to be relied upon. As such it will validate research concepts from other sources, provide new research results, and facilitate the transition of research results from concepts to practical procedures that can be applied with low risk to NASA flight projects. The target of experimentation will be the production software development activities undertaken by any organization prepared to contribute to the research program. Experimental goals, procedures, data analysis and result reporting will be performed for the most part by the University of Virginia.

Knight, J. C.↗

NASA NTRS1988

The implementation and use of Ada on distributed systems with high reliability requirements

The use and implementation of Ada were investigated in distributed environments in which reliability is the primary concern. In particular, the focus was on the possibility that a distributed system may be programmed entirely in Ada so that the individual tasks of the system are unconcerned with which processors are being executed, and that failures may occur in the software and underlying hardware. A secondary interest is in the performance of Ada systems and how that performance can be gauged reliably. Primary activities included: analysis of the original approach to recovery in distributed Ada programs using the Advanced Transport Operating System (ATOPS) example; review and assessment of the original approach which was found to be capable of improvement; development of a refined approach to recovery that was applied to the ATOPS example; and design and development of a performance assessment scheme for Ada programs based on a flexible user-driven benchmarking system.

Knight, J. C.↗

NASA NTRS1987

The implementation and use of Ada on distributed systems with high reliability requirements

Performance analysis was begin on the Ada implementations. The goal is to supply the system designer with tools that will allow a rational decision to be made about whether a particular implementation can support a given application early in the design cycle. Primary activities were: analysis of the original approach to recovery in distributed Ada programs using the Advanced Transport Operating System (ATOPS) example; review and assessment of the original approach which was found to be capable of improvement; preparation and presentation of a paper at the 1987 Washington DC Ada Symposium; development of a refined approach to recovery that is presently being applied to the ATOPS example; and design and development of a performance assessment scheme for Ada programs based on a flexible user-driven benchmarking system.

Knight, J. C.↗

NASA NTRS1987

Detection of faults and software reliability analysis

Specific topics briefly addressed include: the consistent comparison problem in N-version system; analytic models of comparison testing; fault tolerance through data diversity; and the relationship between failures caused by automatically seeded faults.

Knight, J. C.↗

NASA NTRS1987

The implementation and use of Ada on distributed systems with high reliability requirements

A preliminary analysis of the Ada implementation of the Advanced Transport Operating System (ATOPS), an experimental computer control system developed at NASA Langley for a modified Boeing 737 aircraft, is presented. The criteria that was determined for the evaluation of this approach is described. A preliminary version of the requirements for the ATOPS is contained. This requirements specification is not a formal document, but rather a description of certain aspects of the ATOPS system at a level of detail that best suits the needs of the research. The survey of backward error recovery techniques is also presented.

Knight, J. C.↗

NASA NTRS1987

Second generation experiments in fault tolerant software

The purpose of the Multi-Version Software (MVS) experiment is to obtain empirical measurements of the performance of multi-version systems. Twenty version of a program were prepared under reasonably realistic development conditions from the same specifications. The overall structure of the testing environment for the MVS experiment and its status are described. A preliminary version of the control system is described that was implemented for the MVS experiment to allow the experimenter to have control over the details of the testing. The results of an empirical study of error detection using self checks are also presented. The analysis of the checks revealed that there are great differences in the ability of individual programmers to design effective checks.

Knight, J. C.↗

NASA NTRS1986

Detection of faults and software reliability analysis

Multiversion or N-version programming was proposed as a method of providing fault tolerance in software. The approach requires the separate, independent preparation of multiple versions of a piece of software for some application. Specific topics addressed are: failure probabilities in N-version systems, consistent comparison in N-version systems, descriptions of the faults found in the Knight and Leveson experiment, analytic models of comparison testing, characteristics of the input regions that trigger faults, fault tolerance through data diversity, and the relationship between failures caused by automatically seeded faults.

Knight, J. C.↗

NASA NTRS1986

A second generation experiment in fault-tolerant software

Information was collected on the efficacy of fault-tolerant software by conducting two large-scale controlled experiments. In the first, an empirical study of multi-version software (MVS) was conducted. The second experiment is an empirical evaluation of self testing as a method of error detection (STED). The purpose ot the MVS experiment was to obtain empirical measurement of the performance of multi-version systems. Twenty versions of a program were prepared at four different sites under reasonably realistic development conditions from the same specifications. The purpose of the STED experiment was to obtain empirical measurements of the performance of assertions in error detection. Eight versions of a program were modified to include assertions at two different sites under controlled conditions. The overall structure of the testing environment for the MVS experiment and its status are described. Work to date in the STED experiment is also presented.

Knight, J. C.↗

NASA NTRS1986

The implementation and use of Ada on distributed systems with high reliability requirements

The general inadequacy of Ada for programming systems that must survive processor loss was shown. A solution to the problem was proposed in which there are no syntatic changes to Ada. The approach was evaluated using a full-scale, realistic application. The application used was the Advanced Transport Operating System (ATOPS), an experimental computer control system developed for a modified Boeing 737 aircraft. The ATOPS system is a full authority, real-time avionics system providing a large variety of advanced features. Methods of building fault tolerance into concurrent systems were explored. A set of criteria by which the proposed method will be judged was examined. Extensive interaction with personnel from Computer Sciences Corporation and NASA Langley occurred to determine the requirements of the ATOPS software. Backward error recovery in concurrent systems was assessed.

Knight, J. C.↗

NASA NTRS1986

A second generation experiment in fault-tolerant software

The primary goal was to determine whether the application of fault tolerance to software increases its reliability if the cost of production is the same as for an equivalent nonfault tolerance version derived from the same requirements specification. Software development protocols are discussed. The feasibility of adapting to software design fault tolerance the technique of N-fold Modular Redundancy with majority voting was studied.

Knight, J. C.↗

NASA NTRS1986

The implementation and use of Ada on distributed systems with high reliability requirements

The use and implementation of Ada in distributed environments in which reliability is the primary concern were investigted. A distributed system, programmed entirely in Ada, was studied to assess the use of individual tasks without concern for the processor used. Continued development and testing of the fault tolerant Ada testbed; development of suggested changes to Ada to cope with the failures of interest; design of approaches to fault tolerant software in real time systems, and the integration of these ideas into Ada; and the preparation of various papers and presentations were discussed.

Knight, J. C.↗

NASA NTRS1985

An experimental evaluation of error seeding as a program validation technique

A previously reported experiment in error seeding as a program validation technique is summarized. The experiment was designed to test the validity of three assumptions on which the alleged effectiveness of error seeding is based. Errors were seeded into 17 functionally identical but independently programmed Pascal programs in such a way as to produce 408 programs, each with one seeded error. Using mean time to failure as a metric, results indicated that it is possible to generate seeded errors that are arbitrarily but not equally difficult to locate. Examination of indigenous errors demonstrated that these are also arbitrarily difficult to locate. These two results support the assumption that seeded and indigenous errors are approximately equally difficult to locate. However, the assumption that, for each type of error, all errors are equally difficult to locate was not borne out. Finally, since a seeded error occasionally corrected an indigenous error, the assumption that errors do not interfere with each other was proven wrong. Error seeding can be made useful by taking these results into account in modifying the underlying model.

Knight, J. C.↗

NASA NTRS1985

The implementation and use of Ada on distributed systems with high reliability requirements

The use and implementation of Ada in distributed environments in which reliability is the primary concern were investigated. In particular, the concept that a distributed system may be programmed entirely in Ada so that the individual tasks of the system are unconcerned with which processors they are executing on, and that failures may occur in the software or underlying hardware was examined. Progress is discussed for the following areas: continued development and testing of the fault-tolerant Ada testbed; development of suggested changes to Ada so that it might more easily cope with the failure of interest; and design of new approaches to fault-tolerant software in real-time systems, and integration of these ideas into Ada.

Knight, J. C.↗

NASA NTRS1985

The implementation and use of ADA on distributed systems with high reliability requirements

The use and implementation of Ada in distributed environments in which reliability is the primary concern is investigated. Emphasis is placed on the possibility that a distributed system may be programmed entirely in Ada so that the individual tasks of the system are unconcerned with which processors they are executing on, and that failures may occur in the software or underlying hardware. A new linguistic construct, the colloquy, is introduced which solves the problems identified in an earlier proposal, the conversation. It was shown that the colloquy is at least as powerful as recovery blocks, but it is also as powerful as all the other language facilities proposed for other situations requiring backward error recovery: recovery blocks, deadlines, generalized exception handlers, traditional conversations, s-conversations, and exchanges. The major features that distinguish the colloquy are described. Sample programs that were written, but not executed, using the colloquy show that extensive backward error recovery can be included in these programs simply and elegantly. These ideas are being implemented in an experimental Ada test bed.

Knight, J. C.↗

NASA NTRS1984

The implementation and use of Ada on distributed systems with high reliability requirements

Knight, J. C.↗

NASA NTRS1984

The implementation and use of Ada on distributed systems with high reliability requirements

The use and implementation of Ada (a trade mark of the US Dept. of Defense) in distributed environments in which the hardware are assumed to be unreliable were investigated. The possibility that a distributed system is programmed entirely in Ada so that the individual tasks of the system are unconcerned with which processors they are executing on and failures occurring in the underlying hardware were examined.

Knight, J. C.↗