Bounded Block Parallel Lattice Reduction algorithm for MIMO-OFDM and its application in LTE MIMO receiver

Lattice reduction is a promising technique to enhance the performance of sub-optimal MIMO detectors. This paper presents the Bounded Block Parallel Lattice Reduction (BBP-LR) algorithm, which is an implementation friendly low complexity algorithm specifically optimized for practical MIMO-OFDM systems. The optimisation of the BBP-LR algorithm is based on the following: 1) exploiting the frequency coherence in MIMO-OFDM to enable parallel processing that suits SIMD/vector architectures; 2) bounding the execution behaviour to reduce complexity and enable hard real-time task scheduling; 3) performing a frame based preprocessing to significantly reduce complexity. The proposed BBP-LR algorithm has been evaluated in the context of LTE MIMO receivers. When comparing to linear MIMO detectors, the proposed algorithm brings performance improvement at a small complexity increase. Specifically, when compared to ZF, BBP-LR based preprocessing brings an impressive 11dB SNR gain at the cost of only 23% complexity increase.