Abstract: Interference alignment technique effectively improves the channel capacity and restrain the interference, whose question can be attributed to find a closed-form solution. According to the reciprocity of the channel, the traditional interference alignment scheme uses iteration method(distributed) to approximate the closed-form solution, but it bring tremendous computational overhead. In this letter, a low complexity distributed interference alignment is proposed for K user MIMO interference channel. By means of QR decomposition of the interference covariance matrix of the Kth user based on modified Gram-Schmidt, we will choose the last dk columns of the unitary matrix Q to approximate as the interference suppression filter matrix of the user K. Compared to the traditional distributed interference alignment scheme, it not only is the same basically in the system capacity and iterative convergence speed, but also obviously decreases the system computational complexity. The results of simulation show that the validity of the method.