Safety-critical aerospace applications have become increasingly dependent on a software component. However, conventional approaches to increasing the software reliability of these systems have focused on purely programming aspects. While these methods can all lead to techniques which should minimize the number of flaws in the logic of programs, none claim to result in generation of code with zero errors. Recognition of the fact that systems for improved code reuse, code testing, or automated code generation will remain imperfect, has inspired the present innovative proposal for a hybrid hardware and software approach for improving software reliability. The Phase I effort is therefore aimed at researching how to create a hardware "watchdog" which is pre-programmed with characteristics of normal execution and flow of the program. Any deviations from normal program execution would be sensed and corrected by the new hardware device. This research is encouraged by preliminary results with similar approaches for improving the fault tolerance of machine controllers. Success with this proposed effort will anticipate demand by NASA and the aerospace industry for improved reliability of critical systems. It will also lead to creation, in Phase II, of a product for the commercial market which provides both soft and hard fault tolerance in the software running on personal computers.Results will be integrated into a complete approach towards making it possible to run any conventional software with increased reliability on personal computer systems. The whole aerospace industry could also be a large potential market for incorporation of this technology in flight control and flight monitoring systems. It is further possible that any algorithms and hardware developed could be modified to assist in the improvement of embedded software systems for the automotive industry. With successful completion of a product which is user friendly and inexpensive, all companies dependent on their computer avoiding unrecoverable crashes will be a likely customer for adoption of this system.fault avoidance, fault tolerance, software reliability, software validation, recovery block, watchdog
