Abstract:  A parallel multi-channel blind equalization algorithm is proposed to overcome the problems such as long tap coefficients and the incompetence for deep fading channels. By utilizing the independently optimized characteristics of forward and feedback filters in the feedback structure, the new algorithm adds the weighted decision value into the following iterative procedure, which is likely to eliminate the smearing effect. Meanwhile, the length of tap coefficients is effectively reduced by means of the high sampling rate of the fractionally spaced equalizer, the small steady-state error of decision-directed algorithm and the parallel processing of the received signal. For CDL signal, experimental results show that the new blind equalization algorithm is capable of simultaneously accelerating the convergence and reducing BER.