The existing distributed iterative Interference Alignment(IA) algorithms do not take the processing capacities of the base stations and the user equipment into account, which makes the base stations and the user equipment have both high computational complexity and high system overheads. A non-iterative IA optimization algorithm based on maximum capacity is proposed for the downlink and uplink, respectively. With the proposed algorithm, the base stations could obtain the transmitting pre-coding matrix or the receiving suppressing matrix by the optimal method based on maximum capacity, and the user equipment could obtain the receiving suppressing matrix by zero force method and randomly generate the transmitting pre-coding matrix. Analysis and simulation results show that, with the proposed algorithm, the computational complexity and system overheads at the base station are the same as the existing algorithms, but at the user equipment, they can be greatly decreased. Furthermore, the proposed algorith can achieve a good performance similar with that of the Maximum Signal-to-Interference-and- Noise Ratio(Max-SINR) algorithm.