FIELD: computer engineering. SUBSTANCE: invention is related to coders with error correction. Proposed decoder of maximum of a posteriori probability ( MAP decoder ) for lattice codes with correction of errors employing final bit train forms output data of soft decision with evaluation of probability of states of first stage of lattice code. These probabilities replace a priori information of initial state of traditional MAP decoder. Decoder provides for distribution of probability of initial state by any of two methods. First method relates decision to task of characteristic value for which resulting eigenvector presents desirable distribution of probability of initial state. Knowing initial state cyclic MAP decoder executes the rest part of decoding in correspondence with conventional decoding algorithm on basis of maximum of a posteriori probability. Second method is based on recursive processing to secure convergence of iterations to distribution of initial state. After sufficient number of iterations state of cyclic sequence of states becomes known with high degree of probability and cyclic MAP decoder fulfills the rest part of processing in correspondence with conventional algorithm of decoding by method of maximum of a posteriori probability. EFFECT: high precision of decoding. 25 cl, 8 dwg
Authors
Dates
2002-02-10—Published
1997-04-14—Filed