An efficient output function partitioning algorithm for reducing hardware overhead in self-checking circuits and systems

Either m-out-of-n or Berger code checkers are generally used in a self-checking circuit or system to detect unidirectional errors. Results have shown that the hardware overhead of such a checker increases nearly exponentially as the code length increases. The use of many smaller checkers may require much less hardware overhead than the use of an equivalently larger checker. An efficient partitioning scheme is presented so that the output functions can be partitioned into a smaller size so that a smaller checker can be used. The development of the proposed partitioning algorithm is presented. Results of this study show that the proposed algorithm reduces both hardware overhead and performance degradation significantly.<
>