光学相干突发模式接收器：并行与流水线设计的延迟通道均衡反馈

29 浏览量更新于2024-08-27 收藏 733KB PDF 举报

"来自并行和流水线设计的具有延迟通道均衡反馈的光学相干突发模式接收器" 这篇研究论文深入探讨了光学相干突发模式接收器（Burst-Mode Receiver, BMR）的设计，特别关注了并行数字信号处理和流水线架构在其中的应用。光学相干突发模式接收器是一种用于高速光通信系统的接收设备，它能够处理突发性的数据流，这种数据流在光通信中特别是在数据中心和未来光网络中是常见的。在本文中，研究人员提出了一个112-Gbps极化复用四相移键控（PDM-QPSK）调制的BMR设计方案。论文的主要贡献在于研究了数字信道均衡器的收敛率特性，这在考虑到非忽略不计的计算延迟时尤为重要。信道均衡器是用来补偿由于信道衰减和干扰引起的信号失真的关键组件。在实际系统中，由于处理延迟的存在，均衡器的性能可能会受到影响，因此需要深入理解这一现象。作者通过仿真研究了这个问题，并发现信道均衡器与频率偏移估计（FOE）之间存在交互作用。这种交互可能导致性能下降，因为两者之间的不协调可能会破坏均衡过程。为了解决这个问题，他们提出了一种新颖的均衡器自适应方法，该方法利用最新的可用均衡器输出和最新的FOE信息来优化调整过程。这种方法旨在改善系统的整体性能，提高接收机的误码率性能，同时减少由于延迟带来的负面影响。此外，论文还可能涵盖了并行处理和流水线设计的优势，这两种技术可以提高处理速度，降低延迟，并且在处理大量数据时能有效提升系统效率。并行处理通过同时处理多个数据块来加速计算，而流水线设计则通过将处理任务分解成多个阶段，使得每个阶段可以在不同时间进行，从而实现连续的数据处理，提高了吞吐量。这篇论文为高性能光学通信系统的接收端设计提供了新的思路，尤其是在处理突发性数据和应对信道延迟挑战方面。提出的延迟通道均衡反馈方法对于提高光学相干接收器的性能和稳定性具有重要的理论和实践价值。

Optical coherent burst-mode receivers with

delayed channel equalisation feedback from

parallel and pipelined design

ISSN 1751-8628

Received on 28th July 2014

Accepted on 17th December 2014

doi: 10.1049/iet-com.2014.0735

www.ietdl.org

Bo Xu

✉

, Xinyu Liu, Chenyu Wu, Kun Qiu

Key Laboratory of Optical Fiber Sensing and Communications (Education Ministry of China), School of Communication and Information

Engineering, University of Electronic Science and Technology of China, Chengdu 611731, People’s Republic of China

✉ E-mail: xubo@uestc.edu.cn

Abstract: A digital optical coherent burst-mode receiver (BMR) with 112-Gb/s polarisation-division multiplexing-quaternary

phase-shift keying (QPSK ) modulation is p roposed with parallel digital signal processin g and pipelined design. The

convergence rate property of the digital channel equaliser with non-negligible computational delay is studied extensively

using simulations. Owing to the interaction between the channel equalisation and frequency offset estimation (FOE), a

novel equaliser adaptation method using the latest available equaliser output and the latest available FOE result is

proposed for a faster convergence than the conventional method. Simulation results reveal that the computational delay

have a great impact on the convergence rate. A smaller step size for tap coefficient updating with slower convergence rate

should be used to avoid divergence. Compared with the FOE module, the computational delay from the channel equaliser

has the dominant effect on the convergence. The design of the equaliser with short computational delay is the key for a

fast convergence rate. The parallel and pipelined digital optical coherent BMR after successful initialisation has negligible

error rate performance loss when compared with the ideal serial signal processing within a wide frequency offset range.

1 Introduction

Optical burst transmission is an efﬁcient way to build elastic optical

network with sub-wavelength granularity [1–3]. In such a network,

the data is transmitted in the form of optical bursts from different

source nodes to different destination nodes and different optical

bursts share one wavelength channel with time multiplexing.

With the fast development of coherent detection and polarisation-

division multiplexing (PDM) to improve the channel efﬁciency,

optical burst-mode receiver (BMR) with coherent detection for

PDM-QPSK signals is expected to be one of the key technologies

for optical burst transmission [4, 5].

In an optical coherent BMR with digital signal processing (DSP),

adaptive equalisation is used to compensate ﬁbre dispersion,

polarisation-mode dispersion (PMD) etc. Different from

conventional optical coherent receivers working in a continuous

mode, an optical coherent BMR should be able to make fast

re-initialisation for receiving the next optical burst which may

come from a different source node in the network and encounter

different channel condition. The fast initialisation of the adaptive

equaliser in an optical BMR is crucial for improving the

transmission efﬁciency. A training sequence-based initialisation

method with least-mean square (LMS) algorithm for coefﬁcients’

adaptation is thus used in this work [6, 7].

Real-time implementation of DSP functions is the bottleneck for

an optical coherent BMR at a transmission rate of 100-Gb/s or

above [8]. Parallel processing with pipelined structure must be

used [9, 10]. Some early works have proposed ways for block

implementation of adaptive digital ﬁlters which can process a

block of signals simultaneously [11–13]. The effect of delays on

the block adaptive LMS digital ﬁlters were then found to slow

down signiﬁcantly the convergence speed of the equaliser through

simulations [6, 7, 14]. However, these early works did not

consider the interaction of the adaptive LMS ﬁlters with the

frequency offset which is another limiting factor for optical

coherent receivers. A detailed simulation study on the effect of

parallelism and pipelined structure is included in this paper.

This paper is organised as follows. The pipelined structure of the

proposed optical BMR with parallel signal processing is given in

Section 2. Section 3 explains the key signal processing functions

used in the BMR. Section 4 studies the convergence speed of the

equaliser taking into account the parallelism and computational

delay in the pipelined structure. Simulation results on the error rate

performance of the optical coherent BMR are also given in

Section 4. Section 5 concludes the paper.

2 Structure of the optical coherent BMRs

Fig. 1 gives the schematic of the optical coherent BMR with parallel

DSP. Under the assumption that a 28 G baud rate PDM-QPSK signal

is transmitted over the channel and sampled at a rate of two samples

per symbol at the receiver side, the DSP modules should be able to

process the discrete-time signals at a rate of 56 Gsamples/s. Current

electronic processing digital circuits cannot deal with such a high

speed in a serial way and parallel signal processing is imperative.

For instance, the optical coherent BMR in Fig. 1 has 128 parallel

branches and is designed to process 128 symbols simultaneously.

The clock rate for the DSP modules can then be effectively

reduced from the 28 G baud rate for the original signals to a value

of 218.75 MHz. This paper focuses on two DSP modules, the

channel equaliser (EQ) and the frequency offset compensation and

estimation (FOC and FOE) module as shown in Fig. 1.EachEQ

is build with a butterﬂy structure for simultaneous channel

equalisation and polarisation division demultiplexing with a

schematic shown in Fig. 1b. The four sub-equalisers, H

and H

, are assumed to have equal number of taps in this

study. The structure of the sub-equaliser H

is given in Fig. 1c

with 11 taps as an example. A data-aided LMS-based adaptation

algorithm is used on the EQ’s tap coefﬁcients in this study for a

faster convergence speed.

Carrier phase estimation (CPE) is also a commonly-used module

in optical coherent receivers because of non-negligible laser

linewidths. For small laser linewidth <1 MHz, the laser linewidth

induced phase variation from one block of 128 symbols to the

next block of 128 symbols has a variance <0.03 for the studied 28

G baud rate PDM-QPSK signals. Such a small phase variation can

be tracked by the data-aided LMS-based coefﬁcient updating of

IET Communications

Research Article

IET Commun., 2015, Vol. 9, Iss. 7, pp. 975–981

975

The Institution of Engineering and Technology 2015

下载后可阅读完整内容，剩余6页未读，立即下载

weixin_38614952

粉丝: 7
资源: 887

光学相干突发模式接收器：并行与流水线设计的延迟通道均衡反馈

fpga的流水线设计（PipelineDesign）、并行处理 (1) fpga开发.pdf

FIR_滤波器；Verilog_并行流水线_fir_源码

Verilog HDL语言32位MIPS流水线CPU的设计

实战MATLAB之并行程序设计

FPGA全流水线：低延迟高质量图像旋转设计与比较

优化DSP设计：流水线与并行处理提升性能与功耗控制

并行模拟退火算法：加速衍射光学元件优化设计

光学相干断层成像(OCT)原理与技术发展

FPGA流水线设计详解：优化开发效率的实用策略

优化H.264 CABAC硬件实现：五级流水线设计与性能提升

最新资源