An adaptive fault tolerance scheme for applications on real-time embedded system

An adaptive fault tolerance scheme aiming at those real-time embedded systems that have single processor and limited resources is presented in this paper. As the fault behavior exhibited by system changes with time and environment, it can decide a suitable strategy among various reliable techniques that have been introduced into application software. Prediction on fault emergence is made by strategy decider according to its memory of fault emerging history. The application module is assigned to contact with fault tolerance controller through its stubs. Performance of the scheme is measured and evaluated in experiment. Some parameters are also discussed.