FIELD: information technology.
SUBSTANCE: invention relates to a method of decoding low-density generator matrix codes (LDGC). The method of decoding involves adding L-K known filling bits into a received sequence R of code words and deleting code word symbols deleted in the channel from that sequence R to obtain Re. Rows corresponding to code word symbols deleted in the channel are also deleted from the matrix Gidgct obtained by transposing the LDGC generator matrix to obtain a matrix Ge. Transposition of columns of the matrix Ge is then performed in order to generate a matrix 
where A is a lower triangular square matrix of the order M, and D B record the transposition of columns corresponding to the relationship between Ge and Ga. Gaussian elimination is performed on matrix Ga to generate a matrix G* in which the first L rows make up an identity matrix, and transposition and summation operations are simultaneously performed on corresponding elements Re in accordance with operations for transposing and summation of rows in the Gaussian elimination in order to generate Re. I't is then obtained from the relationship Gb ∗I't = R'e and inversion is carried out in I't to obtain O't; st is obtained from GIdgct(0:L-1.0; L-1)O't=st and L-K known filling bits are deleted from st to obtain K information sequence bits.
EFFECT: easier decoding and considerably faster rate of decoding.
10 cl, 7 dwg
