Design of Lattice Reduction Algorithms for Linear-Precoded MIMO System

In this letter, the design of lattice reduction (LR) algorithms specialized for improving the performance of a linear-precoded MIMO system is investigated. Conventionally, the lattice reduction algorithm required in such system is performed mostly by either the Lenstra-Lenstra-Lovasz (LLL) or the Seysen's algorithm (SA), which is designed to search for shorter or nearly-orthogonal bases of a lattice. In this letter, we show that additional performance gain can be further obtained by taking the mean-square-error (MSE) of the considered MIMO system into account. The computational complexity and the error rate achieved by the proposed algorithm is quantified and compared with the LLL and SA via numerical simulations.