Skip to Main Content
The standard LDPC decoder is generalized to compute joint posterior conditional probabilities of arbitrary sets of codeword bits given the received data. This generalized decoder is applied to the problem of blind turbo equalization. Blind channel identification is formulated as an expectation maximization problem. The solution depends on first and second order statistics of the transmitted code bits. Marginal and pairwise joint probabilities computed by the generalized LDPC decoder are used to evaluate the first and second order statistics for channel identification. Channel estimates are used by a MAP equalizer. Information is exchanged between the MAP equalizer and the LDPC decoder in a turbo fashion. After each turbo iteration, the channel is re-estimated. Simulation results compare blind turbo equalization with channel informed turbo equalization.