Research And Implementation Of Coordinated Multiple Points Precoding In C-RAN Network

With the rapid development of the mobile internet and rapid growth in mobile data, the fourth-generation mobile communication LTE-Advanced began to rise.In order to solve maintenance costs and infrastructure investment issues faced by Mobile operators, the C-RAN network architecture is proposed which contains centralized processing, collaborative radio, and real-time cloud infrastructures a. This brings a number of technical challenges, including advanced collaboration transmitter/receiver technology which is the key to achieve higher spectral efficiency. In LTE, adjacent cells use the same frequency resources and inter-cell interference is the main factor influencing spectral efficiency. 3GPP proposed multi-point collaboration technology(Co MP) to effectively suppress such interference. In this paper, several Co MP precoding algorithms are studied and simulated.Because there is a large number of matrix operations in algorithms, analysis of the matrix calculation implementation has been done in this paper. Finally, the pre-coding matrix calculation module has been implemented on DSP, testing the latency and throughput, chosing the best method to ensure a real-time processing system. The main research achievements are as follows:First, this paper introduces the structure and characteristics of C-RAN network, proposing the superiority of Co MP technology used in this architecture, affirming its research value. A brief introduction of relevant principles of Co MP technology is given, mainly containing collaboration scenarios and Co MP transmission mode, Co MP user’s judgment, coordination and other selected clusters. Then two core technology of LTE physical layer OFDM and MIMO are introduced.Second, the common precoding techniques in CS/CB and JP transmission modle are summarized in this thesis. We introducethe zero forcing algorithm(ZF), the block diagonalization algorithm(BD-SVD) and signal-leakage noise ratio(SLNR) algorithm. Because there are a large number of matrix operationsin the hardware implementations of precoding technology, the rate and accuracy of matrix inversionand SVD are practical problems. For this, research of matrix calculation is done to ensure the rate and accuracy.Finally, to further evaluate the complexity of precoding algorithm for hardware implementation and the feasibility of Network deployment, the Freescale MSC8156 ADS was used for building the hardware simulation platform. What’s more, this platform ensures real-time communication by considering DSP software optimization and pipeline processing.The software simulation results show that in the same pre-coding algorithm JP mode has better error rate performance than CS/CB mode. Comparing the performance of each algorithm, SLNR algorithm is the best and BD_SVD algorithm performance followed. The hardware implementation results show that the throughout of matrix inversion and matrix eigenvector decomposition meet the needs of LTE protocols, but SVD matrix decomposition requires two DSP core to meet throughput requirements.