Abstract
In this thesis, a fixed complexity lattice reduction algorithm with the Tomlinson-Harashima precoding (THP) [1] is proposed for MIMO broadcast channels. Fixed complexity LLL algorithm in deep insertion (LLL-deep) [2] do sorted-QR (SQR) ordering and size reduction on the primal basis. However, we observe that doing SQR ordering on the dual basis is better than doing SQR ordering on the primal basis in ordering THP. So we proposed a modified version of fixed complexity LLL-deep algorithm that do SQR ordering and size reduction on the dual basis, which is different from fixed complexity LLL-deep that do SQR ordering and size reduction on the primal basis. Also, the complexity of proposed algorithm is fixed, which is more sutiable for hardware implementation. Simulation results show that for a fixed number of reduction stages, the performance of proposed lattice reduction algorithm is better than the fixed LLL algorithm [3] and fixed complexity LLL-deep algorithm in THP, and has nearly the same performance and diversity order as LLL [4].