Research And Implementation Of LTE PUSCH Receive Based On GPP

Mobile communications has maintained a rapid development for a long time.The latest systems are quite different from previous ones.However,in order to support all devices,it is still necessary to maintain each platform.At the same time,continuing to evolve will also lead to the development of new hardware.Those are leading to a heavier and heavier burden.Cloud Radio Access Network(C-RAN)may become one of the key solutions in the fifth generation mobile communication technology.One of its core concepts is to implement different communication protocols through Software Defined Radio(SDR).Based on the project,this article carefully studied the PUSCH channel of LTE,and set up a Linux-UBUNTU-GCC software platform based on GPP(General Purpose Processor).When implementing sending and receiving,this article finds that the delay at the receiving is too large,so those modules which take too long should be optimized.Concerning the selection of GPP,we use the most marketable product on the market,Intel CPU.In order to complete the optimization,the CPU should be deeply understood.This article simply describes its architecture features,briefly introduces the new technology in the CPU,and focuses on the Single Instruction Multiple Data(SIMD)used in the project.Technical instructions explain the principles of SIMD technology and the effects of instructions,and explain how to use them and how to optimize them: use SIMD instructions as much as possible and adjust data to a parallel structure.This paper optimizes five modules with large delays: Channel estimation,MultipleInput Multiple-Output(MIMO)detection,Quadrature Amplitude Modulation(QAM)demodulation,Turbo decoding,and Cyclic Cyclic Redundancy Check(CRC)is optimized.This article explains in detail the principles of these modules and then presents their optimization algorithms.When using SIMD optimization,this paper proposes a general-purpose complex multiplication using SIMD optimization.For channel estimation,MIMO detection,and QAM demodulation module,this paper uses the SIMD instruction to rewrite,and the performance is greatly improved after optimization.For example,the QAM demodulation performance is improved by 4 to 5 times compared with other documents.For the more difficult parallel Turbo decoding and CRC module,this paper deduces the parallelism from the algorithm,and finally implements parallelism based on the GPP and SIMD instructions.Finally,the optimized comparison is given.The CRC is improved by 7~8 times compared with other documents.Finally,compared with the literature [6],the optimized module has better performance and the performance is improved between 1.7 and 7.6 times.In summary,this paper implements a PUSCH channel receiver based on the GPP fixed point,and achieves a higher performance of the five modules with higher latency through parallel algorithms and SIMD instructions,which is superior to other literature performance and meets the requirements of the project.need.