DIAGNOSABILITY AND DIAGNOSIS OF ALGORITHM-BASED FAULT-TOLERANT SYSTEMS

Parallel processing architectures are now in common use for signal pro cessing and other computation-intensive applications. These applicatio ns are characterized by high throughput and long processing periods. S uch characteristics decrease the reliability of high-performance archi tectures. The erroneous data produced by faulty processors could have damaging consequences, particularly in critical real-time applications . It is therefore desirable that any erroneous data produced by the sy stem be detected and located as quickly as possible. Algorithm-based f ault tolerance (ABFT) is a low-cost system-level concurrent error dete ction and fault location scheme. We apply methods used in the analysis of multiprocessor systems employing system-level diagnosis to the ana lysis of ABFT systems. A new algorithm to analyze an ABFT system for i ts fault diagnosability is developed using these methods. Based on thi s work, a fault diagnosis algorithm is developed for ABFT systems. No such algorithm has been presented previously.