A dependence graph-based approach to the design of algorithm-based fault tolerant systems

A two-stage approach to the design of algorithm-based fault-tolerant (ABFT) systems is proposed. In the first stage a code is chosen to encode the data used in the algorithm. In the second stage the optimal architecture for implementing the scheme is chosen through the use of dependence graphs. Dependence graphs are a graph-theoretic form of algorithm representation. It is demonstrated that not all architectures are ideal for the implementation of a particular ABFT scheme. The authors propose new measures for characterizing the fault-tolerance capability of a system in order to better exploit the proposed design method. Dependence graphs can also be used for the synthesis of ABFT schemes for nonlinear problems. An example of a fault-tolerant median filter is provided to illustrate the usefulness of the dependence graph as a design tool for nonlinear system synthesis.<>