Research And DSP Implementation Of TD-LTE-A Mimo Detection Algorithm

LTE (Long Term Evolution) is being standardized by the3GPP, which improvesthe air interface technologies using the two key technologies, MIMO and OFDM.Depending on the LTE duplexer in a manner, it can be divided into the TDD and FDDtwo modes. TD-LTE-A is the subsequent evolution of TD-LTE. The overall technicalindicators of TD-LTE-A have been more than technical indicators of4G, which uplinkpeak rate is up to500Mbps, the downlink peak rate is up to1Gbps. It is the MIMOtechnology that plays an important role.On the premise that is without increasing the system bandwidth and power,Multiple Input Multiple Output (MIMO) technology, respectively, using of spatialmultiplexing and transmit diversity mode can increase the system channel capacity andenhance the system’s anti-interference ability. MIMO has become the key componentof the4G. However, with the increase of the number of antennas, and modulation order,the MIMO detection algorithm complexity will be rapid growth in the system receiver.So it is imperative to find the performance and low complexity detection algorithm.Firstly, this paper briefly describes the basic knowledge of the MIMO-OFDMsystem and TD-LTE-A physical layer and downlink MIMO technology. It introducesthe MIMO transmit diversity, spatial multiplexing and chiefly introduces thebeamforming technology. This paper researches the MIMO detection algorithms andbuilds a MIMO detection simulation platform. This paper does the performance testingand validation of the MIMO detection algorithms.Then this paper introduces the hardware platform and builds the MIMO detectionlink. This paper introduces particularly the solutions of the MIMO detection link. Andthis paper solves the inter-board communication, inter-core communication andmulti-core water and other key issues. Through the resources verification of the DSPplatform, it meets the demand of testing MIMO detection algorithm.Finally, this paper chooses the MMSE detection algorithm to realize. The MMSEdetection algorithm is divided into five parts, namely, matrix operation, matrixmultiplication, matrix inversion, MMSE filtering and multi-core parallel processing. Based on system optimization standard, the matrix multiplication and inversionmodule have been optimized in this paper. This paper has done detailed testing andanalysis, on the MMSE detection algorithm performance and hardware resources cost.It reaches a conclusion that the MMSE detection algorithm is reasonable and feasible,and can meet the demand of the project.