Abstract: For the slow fading time-varying multi-path channel with low Rician factor, the signal energy cannot be effectively used, and the system error performance is poor. The purely equalization based on the minimum mean squared error (MMSE) technology cannot effectively reduce the error. On the basis of constructing the channel equivalent model, a two-round MMSE equalization algorithm is proposed that uses hard decision symbols to reconstruct the channel and compares the time-frequency joint judgement of the error information of the modified signal within two channel length. The simulation results show that the low Rice factor channel after equalization can approach the bit error characteristics of the high Rice factor. When the signal-to-noise ratio is 20dB, the bit error rate of QPSK signals under different Rice factors is reduced to 10^(-4) below and within an order of magnitude. When the signal-to-noise ratio is 2~6dB, the system bit error rate is lower than the bit error rate of the signal propagating in free space range from 10^(-1) to 10^(-3), indicating that MMSE equalization makes full use of the energy of the multipath component and obtains the signal-to-noise ratio gain.