Research On Sampling Frequency Offset Estimation And Compensation Algorithm In Analog Fronthaul System

As the explosive growth of mobile traffic,the user demand of wireless access network capacity has been continuously increasing.At the same time,Cloud Radio Access Network(C-RAN),as the wireless access network closest to the UE,has drawn extensive attention in the industry due to its convenient resource management and low energy consumption,which is considered a reliable solution of future mobile communication.C-RAN architecture improves system operation efficiency by separating the baseband processing unit(BBU)from the remote radio unit(RRU).The data transmission interface between the BBU and the RRU is referred to as the mobile fronthaul.Mobile fronthaul is an important part of implementing C-RAN architecture.Among them,D-RoF and A-RoF are two effective candidates for current mobile fronthaul.And the Common Public Radio Interface(CPRI)is a typical D-RoF interface standard.The huge capacity of the future mobile communication needs more transmission bandwidth,which will set stringent requirements on the wavelength resource and the optoelectronic device bandwidth of the CPRI-based D-RoF system.However,A-RoF uses QAM modulated radio,so compared with the D-RoF scheme,its demand for wavelength resources is lower and the system overhead is also smaller,which makes A-RoF the research focus of mobile fronthaul.However,the OFDM-based A-RoF systems are sensitive to the clock synchronization between the receiving and the transmitting side,and synchronization mainly refers to the sampling frequency offset(SFO)between the DAC and the ADC.In the A-RoF system,the influence of SFO mainly has the following two aspects: 1)SFO of multi-intermediate frequency OFDM: the multi-intermediate frequency OFDM signal will introduce frequency residual in the down-converted time-domain signal,resulting in un-demodulation and serious impact on the system performance.2)SFO of baseband analog OFDM: The baseband OFDM signal will lead to the phase rotation of the frequency domain signal,inter-subcarrier interference and amplitude attenuation.The main research contents of this dissertation are as follows:1.SFO estimation algorithm in Multi-IFoF mobile Fronthaul systemMulti-IF analog mobile fronthaul is one of the schemes of current A-RoF.In order to solve the problem that SFO results in un-demodulation of system signal,this paper first establishes the mathematical model of signal containing SFO.Then,we propose a non-pilot aided and high accuracy SFO Estimation algorithm.More specifically,starting from the down-converted time-domain signals,a simple digital signal processing and auto-correlation algorithm are applied to extract the frequency offset.The autocorrelation operation allows the estimation algorithm to maintain high accuracy at a low received optical power and dispersion loss.The original IF signal undergoes the sampling frequency adjusted down-conversion again to eliminate conversion residual.The residual SFO in the baseband OFDM signal can be removed by channel estimation.The experimental results show that the estimation error remains within 12 ppb for the SFO range of-100 ppm to 100 ppm,SFO can be estimated as low as-27 dBm,and error vector magnitudes(EVM)after BTB and 20 km fiber transmission are below 9% and 11% respectively at-15 dBm and-21 dBm received optical power.2.SFO estimation algorithm in Time-domain Symbol-Cascade Mobile Fronthaul systemTime-domain Symbol-Cascade mobile fronthaul system is also one of the schemes of current A-RoF.The system analysis shows that it belongs to IMDD-OFDM system and transmits baseband OFDM signal.In this paper,SFO estimation and compensation algorithm is based on frame structure and channel estimation.Specifically,this paper starts with the phase rotation of the demodulated symbol after SFO.A training symbol is inserted at the head and tail of each OFDM frame,the adjacent frames share the same training symbol,and the tail training symbols are used to perform SFO estimation.An additive averaging algorithm is used to reduce the influence of Gaussian white noise to obtain the accurate SFO estimation value.Finally,the phase rotation is compensated for the phase rotation by multiplying the corresponding phase value in the frequency domain.The inserted training symbols are then used for channel estimation to eliminate the residual SFO.The algorithm has low complexity,maintaining the characteristics of the original transmission system.The experimental results show that in the SFO range of-100 ppm to 100 ppm,the compensated EVM is about 10% with an estimated error of 3ppm.