Annotation:
The article describes a method for sum code formation with a minimum total number of
undetectable errors (optimum code), based on the idea of modifi cation of the code withweighted transition summation between the bits, adjacent in data vector. The proposed sum
code can be assigned to the class of modular codes with weighted transitions summation. It
has the same number of check bits as a classic Berger code, however, it also has the improved
performance of error detection in data vectors, especially in a low multiplicity of errors. The
article states the properties of a new sum code, particularly important is that when data vectors
have m ? 2t (t = 1, 2, …) length the optimum code can be built. When m = 2t (t = 1,
2, …), however, the codes are close to optimum. The article explains this feature of a modular
code with weighted transitions. Moreover, the experiment with reference LGSynth`89 combinational
circuits shows that modular code with weighted transitions allows to organize
concurrent error detection (CED) systems with improved detection of errors in comparison
with structures, based on Berger codes. In some cases, the problem of detection of 100 % of
errors at the outputs of combinational circuits under control might be solved. All that makes
the resulting code a promising for solving the problems of technical diagnostics.
Key words:
technical diagnostics; concurrent error detection (CED) system; sum code; Berger code;
weighted transition; code with summation of weighted transitions; undetectable error in
data vector; error detection properties; optimum code