Efficient protection against data errors in embedded control software

This paper presents a novel approach to tolerating transient data faults that may affect the software executing on commercial-of-the-shelf (COTS) embedded processors. The main concept of the approach is the use of information redundancy, in which the program user data and user stack areas are duplicated byte-forbyte in areas of RAM known as mirror arrays. We also present a novel approach to implementing the management of this data duplication and fault detection/correction in software which results in highly readable, portable code. Preliminary results are reported for a matrix multiplication program which indicates the effectiveness of the technique. Key-Words: Software Fault Tolerance, Embedded Systems.