Making blind algorithm of multipath fading underwater acoustic channels suitable for high order QAM signals

This paper begins with a detailed analysis of the difficulties encountered in trying to make blind algorithm suitable for high order QAM (quadrature amplitude modulation) signals. These difficulties are: (1) the MSE (mean square error) of SEI (super-exponential iterative) blind equalization algorithm is too large; (2) switching from SEI blind equalization algorithm to DD (decision directed) blind equalization algorithm, which can give small MSE, is quite often made impossible due to the phase rotation in the equalizer output caused by the phase fluctuation when there exists phase residual in the carrier frequency. We aim to overcome these two difficulties, thus making blind algorithm suitable for high order QAM signals. For the blind equalization problem in multipath fading underwater acoustic channels, we proposed MSEI (modified SEI) blind equalization algorithm and it is capable of correcting phase rotation of the carrier frequency. At the beginning of the iteration process, the MSEI algorithm is used. When MSE level is reduced to a certain level, the iteration process switches to the decision directed (DD) algorithm, which can achieve less residual MSE for high order QAM signals. The MSEI algorithm is derived and the reason why it can correct the carrier phase rotation is explained. The base band model of the switch scheme and the exchange criterion are given and then the iteration process of the proposed algorithm is outlined. The computer simulation results for high order QAM signals show: after 10000 iterations, the MSE for SEI algorithm plus DD algorithm is about -6 dB and the MSE for MSEI algorithm is about -11 dB; but the MSE for MSEI algorithm plus DD algorithm reaches about -18 dB only after 4000 iterations.