The Application Of Matrix Inversion In Massive MIMO

Thanks to the rapid development of the fourth generation of mobile wireless communication technology, it has brought great convenience to our lives. With the increasing demand for data services, the fifth generation of wireless communication systems(5G) also entered the people’s vision. Wherein, the number of antenna in Massive MIMO is much larger than that in the conventional MIMO systems, and the Massive MIMO is the key technology in 5G. It has also greatly enhanced the system performance, and leads to broad application prospect. At the same time, it also enhances spectrum efficiency greatly with the limited bandwidth and transmitting power. In short, the study of the Massive MIMO is of great significance, and will greatly promote the application of the fifth generation of mobile communication technology.In the second chapter of this thesis, we have done some research on the popular algorithms of the Massive MIMO system, such as the channel detection and pre-coding algorithm. For the channel detection algorithms in the Massive MIMO, it is divided into three areas, containing optimal detection algorithm, linear and non-linear detection algorithm. And they are applied to variable communicate models with various performances. We also analyze the advantages and disadvantages of each method and the application of matrix inversion in channel detection. What is more, we do research on the pre-coding algorithms in the Massive MIMO system, and study the differences between linear and non-linear pre-coding algorithms. In the end, this thesis analyzes the advantages and disadvantages of various pre-coding algorithms in different Massive MIMO scenarios and the application of inverse matrix in the precoding aspect.The calculation of matrix inversion in the Massive MIMO occupies an important position in channel detection and the pre-detection algorithms. The complexity of Matrix inversion calculation is relatively complex, the consumption of the hardware resource also occupies a large scale. Thus, it is essential to select a superior performance matrix inversion algorithm. The third chapter introduces the most frequently used methods of matrix inversion, and makes comparison of the advantages and disadvantages among various algorithms. For channel detection and pre-coding algorithms, the types of channel matrix involved is Hermite matrix. Based on the above characteristics, we choose the improved Cholesky decomposition-based matrix inversion algorithm whose complexity is relatively low and easy to implement.In this thesis, it also comes out the design of the Massive MIMO system model whose channel matrix order is 32?8. First, The matrix inversion hardware circuit is designed whose matrix order is 8. And the floating point and fixed-point simulation is done with Matlab as well as the statistical error in matrix inversion module is calculated and make sure that the resources are saved as much as possible within a certain error requirement Then the matrix inversion module is used in the MMSE detection module for fixed-point simulation, and ultimately determine the fixed-point solution. What is more, the verilog coding scheme is designed according to the modular design in the matrix inversion module. Then the simulation of the matrix inversion module hardware is designed by Modelsim, and a test platform is built to test the correctness of the circuit design. The thesis the matrix inversion module is used in the MMSE linear detection module to ensure that it meets the needs of the project.In addition, it is demonstrated that the design of the hardware circuit which is realized in the thesis can be designed to be more compatible with the communication model of the antenna.