The application of lattice-reduction to the K-Best algorithm for near-optimal MIMO detection

An efficient lattice-reduction (LR) aided implementation of the K-Best algorithm is proposed for the general infinite lattice detection problem, which is realized with about 80% less complexity than currently reported architectures. The saving in complexity is achieved by the introduction of an on-demand candidate generation scheme along with a distributed sorting scheme. The proposed scheme does not require any a priori knowledge of the candidate displacement as it expands the candidates using the Schnorr-Euchner method. It is scalable in terms of the number of transmit antennas and its complexity grows sub-linearly with the constellation order. The parallelism intrinsic to the algorithm makes it suitable for the pipelined VLSI implementation.