Low-complexity precoding for MIMO channels with finite alphabet input

Low-complexity precoding design without iteration for multiple-input multiple-output channels with finite alphabet input is studied. Contrast to the iterative gradient method for finding the globally optimal precoder, which is time-consuming and computationally complicated, this work studies low-complexity precoding design with high performance. The problem of optimising the mutual information is identical to maximising the minimum Euclidean distance of the input vector at high total power. Hence, the precoding design can be formulated as semi-definite programming problem and thus can be optimally solved. Numerical examples show that the proposed scheme achieves good tradeoff between the computational complexity and high performance compared with the conventional counterparts.