?
Оценка экспоненты вероятности ошибки для регулярных, основанных на графах МПП-кодов с конечной длиной
This paper deals with the error exponent of the regular graph-based binary low-density parity-check (LDPC) codes under the maximum likelihood (ML) decoding algorithm in the binary symmetric channel (BSC). Unlike other papers where error exponents are considered for the case when the length of LDPC codes tends to infinity (asymptotic analysis) we considered the finite length case (finite length analysis). In this paper we describe the method of deriving the lower bound on the error exponent for regular graph- based LDPC code with finite length under ML decoding and analyze the dependency of the error exponent on various LDPC code parameters. The numerical results, obtained for the considered lower bound, are represented and analyzed at the end of the paper.