Studies On Optimizations And Channel Equalizations In Millimeter-wave Radio Over Fiber System

In order to meet Enhance Mobile Broadband(eMBB)requirements in 5G communications,analog radio over fiber(RoF)system,as a combination technology of optical and wireless communications,is proposed to be utilized in mobile fronthaul network.According to different functional split strategies,RoF technique can be separated into two categories,one is analog RoF and the other is digital RoF.Analog RoF technique directly adds analog wireless signals onto the fiber for transmission.The remote radio unit doesn't have capability on digital processing.Analog RoF has the advantages of low latency,high frequency efficiency which can meet the requirement on increasing transmission bandwidth,thus satisfying eMBB in 5G.However,analog RoF system suffers from the problems on limited dynamic range,linear and nonlinear impairments.In order to solve these one-user point-to-point transmission issues,this paper proposes a new architecture of analog RoF system and studies on the compensation algorithms for linear and nonlinear impairments.In order to cope with multi-point-to-single-point transmission,we employ artificial neural network(ANN)algorithm to solve the problems.Digital RoF still plays an indispensable role in the future 5G mobile fronthaul.It can realize ultra-reliable transmission at cost of bandwidth efficiency by digitizing the analog wireless signals through some interfaces.This paper explores the solutions on the bandwidth-limit and low receiver-sensitivity problem in digital RoF system..The innovations and achievements in this paper are as following:1.In single-user point-to-point transmission,in order to extend the transmission distance of analog RoF,this paper proposes a two-stage single sideband modulation analog RoF system based on directly modulated laser(DML).We employ a special dual-stage single sideband modulation in both optical and electrical field in order to overcome the power fading caused by chromatic dispersion in fiber,thus the transmission distance is extended.In the system,we successfully transmit 14Gbps QPSK signals over 200-km standard single mode fiber(SSMF)and 1-m 60-GHz millimeter-wave channel under bit error rate(BER)3.8×10-3.2.In single-user point-to-point transmission,in order to overcome the linear impairments in analog RoF system,according to the optical-wireless channel characteristic,this paper introduces two kinds of special linear equalizers.First,we propose a variable stepsize decision-directed least mean square(LMS)equalizer,which has a faster convergence speed,tracking the time-variant wireless channel in the RoF system.Second,we separate the RoF link into two sections,one fiber channel and the other one is wireless channel.According to the separated channel,we design a LMS with two sub-equalizers,one is fiber channel equalization and the other is for the wireless.Compared to the traditional LMS equalizer,it has much better equalization effect and the power sensitivities increase by 4dB.3.In order to solve the nonlinearity in point-to-point analog RoF system,this paper employs a multi-level complex-valued ANN nonlinear equalizer to equalize the signals.At the beginning,we analyze the nonlinearity that may happen in RoF system in detail and show that the nonlinearity in analog RoF is an inevitable factor.Then,we design a complex-valued multi-level activation function for ANN to make it better adapt to high order modulation formats and overcome the nonlinear impairments.The BER can be decreased from 10-2 to 10-4.4.In order to solve the problems of muti-user and multi-cell transmission in analog RoF system,this paper proposes two application scenarios of ANN in the millimeter-wave RoF system.In the first case on multiple user uplink transmissions,multiple users will be aggregated in a common fiber channel,increasing the peak-to-average ratio(PAPR)of the signals leads to nonlinear inter-user cross modulations.We introduce a joint-user equalization scheme to mitigate the inter-user modulations.The second case is the coordinated small-cell transmission.This paper designs an ANN-based MIMO de-multiplexer,which has a memory effect so that it can tolerate a time discrepancy between different channels.When the transmission data rate is 2.344 Gbps,the ANN has a time discrepancy tolerance of 4 ns.5.In order to solve the problem of bandwidth-limited and receiver sensitivity limited issues in the current digital RoF system,this paper proposes two solutions.First,in order to increase the available bandwidth in fiber channel,we propose a polarization-modulation and direct-detection system.It can alleviate the power fading caused by chromatic dispersion in fiber,and increase the available baseband bandwidth by 41.4%.Second,to increase the sensitivity of the receiver,making one wavelength shared by more RRUs in mobile fronthaul and increase the signal quality,this paper proposes a maximal ratio combine algorithm to combine coherent receiver and direct-detection receiver.It takes advantages of the two kinds of receivers to realize uRLLC and obtains high sensitivity.Compared with the traditional direct-detection receiver,the sensitivity of the proposed receiver increases by 18.5 dB.