C-RAN中LTE回传压缩技术的研究与实现

发布时间：2018-04-09 21:03

本文选题：C-RAN　切入点：LTE　出处：《电子科技大学》2014年硕士论文

【摘要】：随着移动通信技术快速地发展,运营商要投入巨大的人力、财力、物力进行建设、运营、维护和升级等管理;网络覆盖越来越紧密,运营商还需投入大量基站,这势必导致高额能耗。C-RAN采用集中式基带处理池和分布式射频拉远单元的部署方式,有效地降低了无线接入网的建设与运维成本。然而基带处理池和射频拉远单元之间的回传网络必须能够承受巨量的实时基带信号,在20MHz带宽的LTE-A系统中,采用8天线配置,基带数字信号速率高达9.8304Gbps。本文针对上述背景,分析和研究了基于I/Q基带数据的压缩方法,利用MATLAB仿真了不同频域压缩器和量化器的配置对压缩模块性能的影响。还针对重采样后基带信号的统计特性,改进了压缩模块。经仿真,改进后的压缩模块的性能满足3GPP对EVM(dB)(Error Vector Magnitude,误差向量幅度)上限-21.9382dB的要求,其EVM(dB)达到-35.3308dB,压缩比率为2.1818。本文将I/Q压缩模块和改进后的压缩模块嵌入LTE系统的PDSCH链路进行仿真,除64QAM调制方式下有1dB内的损失,QPSK和16QAM调制方式下的误码率性能稳定。本文还设计了基于联合Turbo译码的上行协作回传压缩方法,利用两个射频拉远单元分别回传解速率匹配后的其中两流软比特信息,在基站端利用这四流软比特信息进行联合Turbo译码。经MATLAB仿真,QPSK调制方式下,误码率为610?时,两个协作的射频拉远单元需要的最小信噪比分别是4dB和4dB,而不协作的射频拉远单元需要的最小信噪比为7dB。协作的两个射频单元分别回传2流软比特信息相对于单个射频单元回传I/Q数据,QPSK调制方式下压缩比率为3.2,16QAM调制方式下的压缩比率为1.6,64QAM调制方式下的压缩比率为1.0667。最后,本文还将改进后的I/Q基带数据压缩算法在Freescale MSC8156 DSP上实现,模块的测试结果表明,DSP上压缩模块较MATLAB压缩模块的EVM(dB)有6.4445dB的损失,单个压缩模块占用存储器空间较少,处理周期为0.243854ms。此外,本文还将压缩模块和解压缩划分在AMC(Advanced Mezzanine Cards)板卡的两个核上,嵌入多核DSP上的PDSCH链路,测试结果表明,DSP链路较MATLAB链路达到610?误码率时有4dB左右的信噪比损失,QPSK调制方式下损失较小,16QAM和64QAM调制方式下损失稍大。
[Abstract]:With the rapid development of mobile communication technology, operators have to invest huge manpower, financial and material resources to build, operate, maintain and upgrade management.This will lead to high energy consumption. C-RAN will adopt centralized baseband processing cell and distributed RF remote unit deployment mode, effectively reduce the construction and operation cost of wireless access network.However, the backtransmission network between the baseband processing cell and the RF remote unit must be able to withstand a large amount of real time baseband signals. In the LTE-A system with 20MHz bandwidth, the baseband digital signal rate is up to 9.8304 Gbpswith 8 antennas.Based on the above background, the compression method based on I / Q baseband data is analyzed and studied. The effects of different frequency domain compressors and quantizer configurations on the performance of compression modules are simulated by MATLAB.According to the statistical characteristics of resampling baseband signal, the compression module is improved.The simulation results show that the performance of the improved compression module meets the 3GPP requirement of the upper limit of -21.9382dB for the EVM(dB)(Error Vector amplitude, and its EVMU dB) reaches -35.3308dB and the compression ratio is 2.18188dB.The simulation results show that the performance of the improved compression module can meet the requirements of the 3GPP (error vector amplitude) upper limit of -21.9382dB.In this paper, the I / Q compression module and the improved compression module are embedded into the PDSCH link of the LTE system for simulation. Except for the loss of 1dB in the 64QAM modulation mode and the stable bit error rate performance under the 16QAM modulation mode, the IQ compression module and the improved compression module are embedded in the PDSCH link of the LTE system.This paper also designs an uplink cooperative backpropagation compression method based on joint Turbo decoding, which uses two radio-frequency pullback units to send back two streams of soft bit information after the rate matching, respectively.In the base station, the four-stream soft bit information is used for joint Turbo decoding.The bit error rate (BER) of QPSK modulated by MATLAB simulation is 610?The minimum signal-to-noise ratio (SNR) required by the two cooperative radio-frequency remote cells is 4dB and 4dB, respectively, while the minimum signal-to-noise ratio required by the uncooperative RF-distancing unit is 7dB.The two cooperative radio frequency units respectively return 2 streams of soft bit information to the I / Q data / QPSK modulation mode of a single radio frequency unit. The compression ratio of the two radio frequency units is 1.66 67 in the case of a compression ratio of 3.2G / 16 QAM modulation mode and a compression ratio of 1.66 67 under the single radio frequency unit (RF-Q) mode respectively.Finally, the improved I / Q baseband data compression algorithm is implemented on Freescale MSC8156 DSP. The test results show that the compression module has the loss of 6.4445dB compared with the MATLAB compression module, and the single compression module occupies less memory space.The treatment period was 0.243854ms.In addition, the compression module and decompression module are divided into two cores of the AMC(Advanced Mezzanine cards, which are embedded in the PDSCH link on the multi-core DSP. The test results show that the PDSCH link is 610% higher than the MATLAB link.The loss of signal to noise ratio (SNR) about 4dB is smaller in QPSK modulation mode than in QPSK modulation mode and the loss is slightly larger in 64QAM modulation mode.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN929.5

【相似文献】