Fault detection in combinational circuits based on self-dual complement to constant-weight code

Abstract

A new method of combinational circuits concurrent checking is considered based on Boolean complement of the working functions of the diagnosis object to the constant-weight code words with parallel checkout of the bits functions belonging to the self-dual functions class of the Boolean logic. The described approach to the organization of concurrent error-detection (CED) systems allows to increase the detection ability in comparison with the traditional checkout by constant-weight code. In this case, the undetectable error can be a such error, which will lead to the preservation of the code word weight generated at the outputs of the Boolean complement block, if the distortions of all the functions bits on the opposite sets is coincide. This condition is more stringent than the checkout of the vector belonging a pre-selected code, it means that it gives a greater probability of detecting distortion arising from errors. The described method allows checkout of combinational circuit based on constant-weight codes with the same number of single and zero bits. In this case, the priority is in the widely used «2-out-of-4» code, which has a simple checker structure. A detailed description of the new CED system structure is given, and algorithms for the extension of definition of the values of the checkout functions that provide testability of technical diagnostic tools are given.

Cite this paper

@article{anon2019fault,
  title = {{Fault detection in combinational circuits based on self-dual complement to constant-weight code}},
  author = {Ефанов, Д.В. and Sapozhnikov, Vladimir and Сапожников, Вл. В. and Pivovarov, D.V.},
  journal = {Proceedings of the Institute for System Programming of RAS},
  volume = {31},
  number = {1},
  pages = {115--132},
  year = {2019},
  doi = {10.15514/ispras-2018-31(1)-8}
}

Made from the index record. Check names, capitalisation and page numbers against the paper. Need another style? AI for Research has over 10,000.

Related papers