Research On The Key Algorithm And Optimization Method Of LTE-A Uplink Receiver Based On The Platform Of General-Purpose Processors

Although application specific integrated circuit is widely used in the baseband signal processing of traditional base stations, deficiency exists in terms of the flexibility and programmability, which forces the mobile operators to upgrade their equipment when upgrading mobile networks. The advantages of software radio technology based on general purpose processors (GPP) are low-cost, flexible expansibility and compatibility, which make GPPs the most prospective technology of the implementation of LTE-A base stations. However, uplink multi-antenna technology has already been introduced into the LTE-A systems, and single-user MIMO supports up to four-layer space multiplexing scheme. As the computation complexity significantly increases for large dimensional MIMO systems, especially for channel estimation and MIMO detection, great challenges are posed to the implementation of real-time LTE-A uplink processing based on the GPP platform. Therefore, the research focus of this thesis is to implement real-time channel estimator and MIMO detector for LTE-A uplink based on the GPP platform.The thesis firstly reviews the principles of LTE-A uplink. It mainly covers the basic principle of OFDM, uplink multi-antenna technology, time-frequency structure and3GPP channel model. Then, it introduces the uplink physical processing flow and the algorithm of LTE-A uplink channel estimation and MIMO detection, and evaluates the performance and computational complexity of both the algorithms. And DFT channel estimation and MMSE MIMO detection are determined to be optimized based upon the GPP platform. Finally, the thesis analyzes the bottleneck of implementing the real-time channel estimator and MIMO detector for LTE-A uplink with multi-antenna based on the GPP platform. Then, it introduces the optimal strategy of algorithm level, and the detailed optimization methods by using lookup table, SIMD instructions, and multi-thread technology. By using the optimization method above, the processing delay of both channel estimation and MIMO detection are decreased significantly.Referring to the optimal design above, the optimization of LTE-A uplink channel estimation and MIMO detection based on the GPP platform are accomplished in this thesis. The following conclusions are drawn through the delay test and optimization analysis. For the most computational complex4x4MIMO, the processing delay of channel estimation after optimization is163.8μs, and that of MIMO detection after optimization is166.1μs.