Research And Implementation Of Backhaul Copmression Technology Of Lte In C-RAN

Mobile operators are under pressure as wireless communication develops rapidly: huge amounts of labors, money and resources are thrown into construction, a mass of base-stations produce more and more energy consumption because of the denseness of RAN.C-RAN is an evolution of deployment composes of centralized BBUs and distributed RRHs, which decreases the costs of operations and maintenance. However, a huge amount of baseband data-rate transported in the backhaul between BBU and RRH should be considered, take 20 MHz bandwidth LTE-A system with 8 antennas as example, the bit-rate is 9.8304 Gbps, which impacts the backhaul of C-RAN seriously.An I/Q compress method is studied in this thesis, performances are measured by simulating different spectrum compressor and quantizer with MATLAB. A modified compress method is introduced according to the statistics of baseband signals, simulation shows the performances of modified compression EVM(dB)(Error Vector Magnitude) equals-35.3308 dB, which meets the demanding of 3GPP with the largest EVM(dB) ?21.9382 dB, and compress ratio reaches 2.1818. In addition, I/Q compress module and modified compress module are embed in PDSCH link-level simulation, validities are confirmed by the link-level simulation, which shows there is at most 1dB loss of SNR with 64 QAM modulation, and the BER stays stable during QPSK and 16 QAM modulation.This thesis also designs a compress method with joint Turbo decoding based on cooperated uplink backhaul. In this method, two RRHs back hual sectional baseband signal under cooperation, two streams of soft bits after rate-dematching are back hualed by each RRH, then joint Turbo decoding performs in BBU. Simulation shows each minimum SNR of the two coorperated RRHs is 4dB when BER reaches 610?, the minimum SNR is 7dB when non-coorperated RRH back haul 3 streams of soft bits. The compress ratio is 3.2 when modulated by QPSK, 1.6 by 16 QAM and 1.0667 by 64 QAM, compared with single RRH back haul I/Q data.This thesis implement the modified I/Q compression of C-RAN on Freescale MSC8156 AMC, and also measures memory utilization and process cycles with Code Warrior. Test result shows the compression is valid and real-time processed, there is 6.4445 dB loss of EVM(dB) compared with simulation, memery usage of compress and decompress module is low, process cycle is 0.243854 ms. Besides, compress module and decompress module are embed in PDSCH link-level multi-core DSP project, compression and decompression are assigned to two cores on AMC(Advanced Mezzanine Cards), the performance of compression is further validated, there is 4dB loss of SNR when reaches BER 610? compared with link-level simulation, less loss is presented during QPSK, and much more loss is presented during 16 QAM and 64 QAM modulation.