A Framework for Fault Tolerance Techniques in the Analysis and Evaluation of Computing Systems

The Algorithm Based Fault Tolerance (ABFT) approach transforms a system that does not tolerate a specific type of faults, called the fault-intolerant system, to a system that provides a specific level of fault tolerance, namely safety and/or recovery. ABFT techniques are most effective when employing a systematic form. The error detection is employed based on a high-rate real convolution code. This paper addresses new methods for performing error correction when real number codes are involved. The parity values are determined according to a systematic real convolution code. Detection relies on two sets of parity values which are computed in two different ways, one set from the input data but with a simplified combined processing subsystem, and the other set directly from the output processed data, employing the parity definitions directly. The ABFT philosophy leads directly to a model from which error correction can be developed. By employing an ABFT scheme with effective real convolution code, the design allows high throughput as well as high fault coverage. The simulations show that the great difference between the round-off error and the computer-induced error is large enough to be distinguished.