Research Of Detection And Precoding Technology Based On Lattice Reduction

Lattice Reduction is a powerful mathematics concept that based on lattice theory. Lattice reduction has been successfully used in many fields, particularly data detection and precoding in wirless communication systems. Lattice reduction is aimed to transform the channel matrix into a better conditioned one in order to improve the system performance. In this article, we focus on the lattice reduction application in wireless application from two aspects:data detection and precoding.For the receivers, we innovatively propose an improved joint detection approach based on lattice reduction for TD-SCDMA downlink system. Through lattice reduction, the system matrix of TD-SCDMA is transformed to a new matrix consisting of roughly orthogonal basis vectors so as to improve the performance of TD-SCDMA multicell joint detection. Due to the large dimensions of system matrix, the complexity of lattice reduction is intolerant. Therefore, according to the composition characteristics, we propose a preprocessing method for the system matrix in order to reduce the complelxity of lattice reduction.For the transmitters, we study the lattice reduction application for linear precoding, including codebook based precoding and non-codebook based precoding. For the non-codebook based precoding, we study the lattice reduction aided (LRA) precoding based on ZF (Zero Forcing) and MMSE (Minimum Mean Square Error) criteria. Simulation results show that the LRA non-codebook based precoding method can dramatically improve the BER (Bit Error Rate) performance. For the codebook based precoding, a LRA codebook based precoding mechanism with quantization codebook is proposed. In this LRA codebook based precoding mechanism, the transmitters and receivers share the same codebook quantization information, which is comprised with quantization bits. Transmitters restore the precoding matrixs from the codebook quantization information by the feedback index numbers that receive from the receivers. Based on the Seysen algorithm of lattice reduction, we propose a mechanism for designing codebook quantization information by researching the features of unimodular matrix. In the forth chapter, the4x4and4x8MIMO (Multiple Input Multiple Output) systems are used as the simulation models. The simulation results show that different quantization degrees lead different BER performance. We can design the codebook quantization information according to the performance demand.