基于大规模MIMO网络的低复杂度块对角化算法

Low-Complexity Block Diagonalization Precoder for Massive MIMO Networks

在多用户MIMO系统下行链路中,块对角化(Block diagonalization,BD)预编码算法的和速率性能要优于匹配滤波算法(Matched filter,MF)和迫零算法(Zero-forcing,ZF)。然而,传统的BD算法利用矩阵分解来构造除当前用户的其他所有用户信道的零空间,需要O(N2)浮点运算次数(Float point operations,FLOPs)。当基站的天线数N趋向于大规模时,BD算法计算复杂度巨大。本文提出一种基于投影子方法构造其他用户合成信道的零空间的BD算法,该算法仅需O(N)FLOPs。仿真表明:同传统的BD算法相比,本文所提出的低复杂度BD算法显著地降低了实现复杂度,而和速率性能损失微小,仍然优于MF和ZF,并且当N趋向于大规模时,它的和速率性能趋向于传统的BD算法和SVD算法。

In multi-user MIMO downlink with multi-antennas at user terminals,the block diagonalization（BD）precoder obtains a better sum-rate performance than matched filter（MF）and zero-forcing（ZF）,thus becomeing an attractive precoder.However,the conventional BD generates the null space for all channel matrices excluding the current user by using singular-value decomposition（SVD）and requires float point operations（FLOPs）of order O（N2）.As the antennas number Nat base station（BS）tends to be a large-scale one,computational complexity increases sharply on BS.Thus,a novel low-complexity implementation for BD with a complexity of order O（N）FLOPs is proposed,where the null space of other users′channel matrices is constructed by aprojection matrix.Simulation and analysis exhibit that the proposed low-complexity BD significantly reduce the complexity,compared with the conventional BD and regularized block diagonalization（RBD）at the cost of a slight sum-rate loss.The algorithm performs better than MF and ZF.Meanwhile,the sum-rate performance is similar to the conventional BD and SVD as Napproaches a large scale number.