Pilot Design And Channel Estimation Of Massive MIMO System

With the continuous development of era, the mobile communication system has developed from the initial communication system of first generation (1G) to present mature one of the fourth generation (4G), and starts to develop towards 5G,while Massive MIMO receives widespread research as the key candidate technology of 5 G At present, most research about Massive MIMO system is based on TDD mode under this system.The reciprocity between uplink and downlink of TDD is used to avoid relatively complex downlink channel estimation. But a series of problems also derived from TDD mode, while FDD mode was widely applied and popularized in actual mobile communication system. Therefore, research over pilot design and channel estimation of downstream link of FDD mode in Massive MIMO system has great significance.The downlink of Massive MIMO could be considered as a MIMO-OFDM system. There are dozens and even hundreds of antenna on the base station, and the number of pilot. for channel estimation is quite huge, which causes great expenses of pilot. In response to this question,the paper does research on a kind of two types pilot structure with lower pilot density so as to reduce the corresponding expense.On the basis of pilot design, the pilot shall be used to receive good channel estimation on the receiving end. The widely used channel estimation methods like LS, DFT and ML algorithm basically do not use channel statistical parameters like time delay power spectrum and noise variance, so their performance is not superior enough. Although MMSE algorithm uses channel statistical parameters, it could hardly be used in actual system due to difficulties in obtaining channel statistical parameters. Therefore, the paper raised a channel parameters estimation method based on designed pilot mode, which mainly include multipath number, multipath location, multipath energy and channel noise variance.Two channel estimation methods, the time domain transformation method and frequency domain MMSE algorithm based on time delay power spectrum estimation, were improved on such basis, and the estimated channel statistical parameters was used to improve the performance of channel estimation.Finally, the paper simulates the accuracy and reliability of channel statistical parameters estimation methods on MATLAB platform. The performance of two channel estimation methods was simulated, and algorithm complexity analysis was made. The simulation result indicates that the channel statistical parameters estimation method have relatively good accuracy and reliability, and two channel estimation methods also have high performance.