Full-Diversity Approximated Lattice Reduction Algorithm for Low-Complexity MIMO Detection

In this letter, we propose a new approximated basis vector reordering (ABVR) criterion for low-complexity lattice reduction aided (LRA) multiple-input multiple-output (MIMO) detection. Despite the approximation, the ABVR criterion is proved to collect the full receiving diversity for LRA linear detection. A variant of the well-known complex Lenstra Lenstra Lovász (CLLL) algorithm, i.e., LLL with deep insertion (DLLL), is employed to accommodate the ABVR criterion (DLLL-ABVR). Compared with the original CLLL and other approximated algorithms, the proposed DLLL-ABVR algorithm largely reduces the number of basis vector reordering (BVR) operations. Simulation results show that, on a practical MIMO scale, the proposed lattice reduction algorithm provides similar detection performance, especially for successive interference cancelation (SIC) detectors, while requiring significantly lower computational complexity.