Iterative architecture of low complexity of equalization, estimation, and decoding for the multi-channels ways with fast evanescence

An initialization process with a new sequence of periodical learning for turbo equalization including the estimation and the decoding of channels were proposed for the multichannels ways with fast evanescence. In this article, the equalization is completed by a canceller of interferences and the estimation of channels is realized by using the method of Valenti and Woerner. As for the converter, this last one is of type LDPC in which the matrix of parity is generated from the circular permutations of a matrix of predefined identity. To insure the adaptation between the equalizer and the decoder, we determined analytically the expression of the extrinsic information has the entrance of the decoder. An adaptive equalizer using a function of not linear cost was proposed to offer a low rate of error for the initialization. The proposed architecture is of low complexity and allows to reduce considerably the effects of selectivity both on fr-equencies and temporal for channels of multi-Rayleigh ways for the normalized Doppler frequencies. f_m=f_D T_s=O.02 and 0.04.