CSE4317 Individual Presentation Topics

Computers in Safety-Critical Systems [Bowyer, Chapter 6]

  1. ADA, exception handling,and the Ariane 5 failure. The Ariane 5 failure is described in two reprinted articles (see references below). Both mention the role of exception handling in the language ADA. One of the articles observes that "Ada's exception mechanism has been criticized in the literature, but in this case is could have been used to catch the exception." (italics added) Read Tony Hoare's ACM Turing Award Lecture paper in the February 1981 issue of the Communications of the ACM. Was the problem with the use of exception handling in the Ariane 5 accident the type of problem that Hoare had predicted would happen? [Additional Assignment #1, page 173, Bowyer, 2nd edition.]

    Jean-Marc Jezequel and Bertrand Meyer, "Design by Contract: The Lessons of Ariane," IEEE Computer, Vol. 30, No. 1,pp. 129-130, January 1997.

    Bashar Nuseibeh, "Ariane 5: Who Dunnit?" IEEE Software, Vol. 1, No. 3, pp. 15-16, May/June 1997.

  2. The Patriot antimissile system failure. The Patriot antimissile system received an enormous amount of favorable publicity during the Gulf War with Iraq. Postwar accounts of the system's performance were not nearly so favorable. One spectacular incident of failure was traced to a timing-related problem in the system's control software. Report on this failure, its consequences, and how it could have been prevented. A popular press account of the problems with the Patriot appears in L. Lee's The Day The Phones Stopped, Primus Books, 1992. [Additional Assignment #2, page 173, Bowyer, 2nd edition.]

  3. The Aegis/Vincennes defensive weapons system incident. The Aegis system was developed by the U.S. Navy to allow ships to monitor the airspace around them. An overview of the original hopes for the system appears in J. A. Adam's "Pinning defense hopes on the Aegis," IEEE Spectrum, pp. 24-27, June 1998. In July 1988, the USS Vincennes, equipped with the Aegis system, accidentally shot down an Iranian passenger airliner, killing 290 people. A contributing element to this accident was the poor design of the user interface. Look into this incident to find out exactly what flaws were later identified in the user interface and how they can be (or have been) corrected. A popular press account of the Vincennes incident also appears in L. Lee's The Day The Phones Stopped, Primus Books, 1992. [Additional Assignment #4, page 173, Bowyer, 2nd edition.]

  4. The North Staffordshire Hospital radiation-therapy incident. The North Staffordshire Hospital is located near London, England. From 1982 through 1991, a computer program used in calculating doses of radiation used in therapy contained an error that caused patients to receive 35 percent less radiation than prescribed (see "Computer errors blamed in cancer deaths," The Tampa Tribune, September 30, 1993). below). Look into this incident to find out exactly what the software error was, how it was eventually detected, and what was done to prevent such an error from occurring again. [Additional Assignment #5, page 173, Bowyer, 2nd edition.]

  5. Software error that caused telephone system outages. The cause of this incident involved three faulty computer instructions that were hidden in software changes that were sent out without major testing, because the company judged that the changes were "too small to require it." It's hard to understand how any company could judge any modification to any safety-critical software to be too small to require testing. But amazing as it seems, this actually happened in 1991 (see J. Burgess' "Tiny bug caused phone blackouts," The Washington Post, July 10, 1991). Phone service was affected in Washington, Pittsburgh, Los Angeles, and San Francisco. Report on exactly what caused the problem and on the measures that have since been taken in the telephone industry to prevent such problems from recurring. [Additional Assignment #7, page 173, Bowyer, 2nd edition.]

  6. Computer error blamed for plane crash. In June 2000 a federal jury found an onboard computer maker and a software maker guilty of making defective equipment. Honeywell Air Transport Systems of Phoenix, the onboard computer maker, and Jeppesen Sanderson of Englewood, Co., the software company, made the equipment aboard flight 995 of American Airlines. The plane crashed in Colombia back in 1995, killing 159 people. Report on the details of this incident and the outcome of the resulting legal battles.

  7. Software error contributes to plane crash. National Transportation Safety Board (NTSB) investigators say that a software error may have been a contributing factor in the 1997 crash of the Korean Air 747, Flight 801, in Guam. Supposedly, the bug did not cause the crash; however, if it were not for the bug, the crash might have been averted. Report on the details of this incident and the findings of the NTSB.