Signal Processing Technology For Digitized Radio Over Fiber System

Radio over fiber?RoF?technology has become one research hotspot in the area of mobile communication.It combines the characteristics of broad bandwidth and low loss of optical fiber communication and flexible access of wireless communication.The conventional analog RoF?ARoF?system transmits analog RF signals,which is essentially an analog system.ARoF system has the advantages of simple structure and low-cost.However,the analog signals suffer from the dispersion and nonlinearity in the optical link,which results in limited transmission performance.The digitized RoF?DRoF?system has received much attention in recent years,which overcomes the inherent defects of the ARoF system and greatly improves the transmission performance.In this thesis,we briefly introduce the system structure and development direction of RoF,then we compare the advantages,disadvantages and the applicable environment of ARoF and DRoF systems.Finally,an in-depth research on DRoF technology is carried out.In order to solve the problem of low spectral efficiency of the digital baseband signal in the DRoF system,pulse code modulation?PCM?based on non-uniform quantization and differential pulse code modulation?DPCM?schemes are introduced.Moreover,the vector quantization with vector linear prediction?VLP-VQ?and self-organizing feature map?SOFM?neural network clustering for vector quantization schemes are researched to improve the spectral efficiency of the DRoF system.This thesis for the design and implementation of DRoF system is organized as following:?1?The basic model of DRoF system is studied and verified.By employing the VPItransmissionMarker^TM simulation platform,the DRoF transmission link based on PCM and DPCM schemes are built and 20 km optical link transmission of DRoF system is achieved.The simulation results show that compared with the conventional uniform quantization based PCM scheme,the PCM based on non-uniform quantization and DPCM schemes can bring higher performance improvement to the DRoF system.?2?A VLP-VQ scheme for the DRoF transmitter is employed and experimentally demonstrated.The VLP technique is used to preprocess the OFDM sampling signal,which reduces the mean square error of the signal by 5%⁵0%,thus greatly reducing the quantization noise of vector quantization.The experiment verified the 25 km 10 GBaud/?PAM-4 IM/DD system transmission based on digital mobile fronthaul?MFH?.The experimental results show that the performance of error vector magnitude?EVM?at the receiver of DRoF system based on VLP-VQ technology is greatly improved compared to the conventional k-means clustering.?3?A vector quantization scheme based on SOFM neural network clustering is researched and a DRoF transmission experiment is carried out.The optimal codebook required for vector quantization is obtained through offline training to the input vector signals by using SOFM neural network.The online quantization is carried out through the testing of SOFM neural network.A 25 km 5 GBaud/?PAM-4 IM/DD optical transmission link was built to investigate the employed vector quantization scheme.The experimental results show that compared with the traditional PCM scheme,the vector quantization scheme based on SOFM neural network clustering can reduce the EVM value of the receiving signal by about 11%,6%,3%and 1%when the quantization bits per sample are 3,4,5,and 6,respectively.