Application Of Digital Signal Processing Algorithm In Real-time System

In recent years,with the development of communication technology,emerging industries such as the Internet of Things,ultra-high definition multimedia,broadband services and virtual reality technology have developed rapidly.Communication transmission speed,total transmission traffic and the number of users have continued to grow.The booming development of these new services is inseparable from high-speed optical communication as the basic physical layer in the field of communication.Coherent transmission is the preferred transmission mode in optical communication systems with high speed,large capacity and long distance.With the continuous development of the receiver ADC and the FPGA chip,high speed real-time coherent optical transmission system has become one of current research hotspots.This research focuses on the receiver digital signal processing algorithms in the application of real-time systems,including reducing complexity of algorithms and optimize the performances of algorithms in real time systems.This paper focuses on the application of digital signal processing algorithm in real-time system,studies and analyzes the algorithm from three aspects:parallelization,fixed-point and latency.The main contents and innovations of this paper are as follows:1.The differences between offline verification and real-time verification in coherent receiver systems are studied in detail,including the difference between serial and parallelization,floating point and fixed point,and adding latency or not.We introduce the validation process of DSP in real-time coherent optical communication,analyze the performances of typical DSP algorithms in real-time systems.Then,we focus on constant modulus algorithm,which is one of the highest complexity algorithms,summarize the problem of realizing DSP algorithms in real-time system.2.A low complexity polarization demultiplexing algorithm based on constant mode algorithm is proposed,and the algorithm realizes multiplierfree in tap coefficient update module.Off-line experimental verification was carried out in a double-biased QPSK back-to-back transmission system with a rate of 112Gbps.In terms of performance,compared with the traditional constant mode algorithm,the new algorithm has no significant error performance deterioration under different number of parallel paths and different bit widths of input data.3.A low-complexity two-stage joint algorithm based on CMA is proposed to compensate receiver IQ imbalance and equalization.The offline experiment is carried out in a 112Gbps DP-QPSK back-to-back transmission system with emulation module of receiver IQ amplitude and phase imbalance.In terms of performance,compared with the GSOP+2×2 MIMO,the new algorithm has no obvious error performance deterioration when the IQ amplitude imbalance is less than 3dB or the IQ phase imbalance is less than 30 degrees under different parallels.In terms of hardware resources,compared with the traditional algorithm,logic utilization and total register number of the two-stage joint algorithm are reduced by 37.9%and 49.6%,respectively.4.Singularity problem is to restrict the constant modulus algorithm performance is an important factor,in order to double the partial real-time transmission system performance optimization of constant modulus algorithm,this paper probes into the root cause of the singularity problem and form of expression,this paper introduces the impact probability of singularity problem rotation angle,polarization dependent loss and polarization of the differential group delay the three optical fiber channel parameters,The effects of parallelization,fixed-point and latency of algorithms on the receiver depolarization matrix are studied.Based on this point,this paper introduces the offline system singularity problems have been reported from the probability of the algorithm,and chose two kinds of simple structure,easy to be implemented in real-time system of the improved algorithm,and set up a single channel model is used to the traditional constant modulus algorithm and the two improved algorithm performance evaluation,from floating-point and fixed-point,serial and parallel,and relaxation of the three angle consider algorithm,The singularity probability of these three algorithms under different polarization rotation angles,polarization dependent loss and differential group delay is studied in detail.At the same time,Quartus 2 is used for synthesis to compare the hardware resource of algorithms.