A Low-Complexity MIMO Detector Based on Fast Dual-Lattice Reduction Algorithm

Lattice reduction (LR) aided multiple-input multiple-output (MIMO) detectors have been considered as an option to obtain near-maximum likelihood (ML) performance. We first give the analysis to show that large signal-to-noise ratio (SNR) corresponds to the short length of the dual basis vectors. Then, in order to further alleviate the complexity of LR aided MIMO detectors while maintaining acceptable performance, we study the dual-lattice reduction methods and propose a fast dual-lattice reduction (FDLR) algorithm which minimizes the orthogonality deficiency of dual-basis. And a tree search method is presented to implement the FDLR algorithm, which enables a flexible trade-off between performance and complexity. Compared to the existing dual Lenstra-Lenstra-Lovasz (DLLL) algorithm, out proposed FDLR algorithm requires less iteration time and yields more orthogonal basis vectors. Simulation results show that FDLR aided detectors achieve better performance and lower complexity than DLLL aided detectors, especially for large MIMO system.