IEEE PHOTONICS TECHNOLOGY LETTERS, VOL. 29, NO. 17, SEPTEMBER 1, 2017 1439
Subtraction-Clustering-Based Modulation
Format Identification in Stokes Space
Pengyu Chen, Jie Liu, Xiong Wu, Kangping Zhong, and Xiaofeng Mai
Abstract—A Stokes-space-based modulation format identifica-
tion (MFI) technique with low complexity is proposed for coher-
ent optical receivers, on the basis of the subtraction clustering
algorithm. Successful MFI with good optical signal-to-noise ratio
performance can be realized among polarization-multiplexed
BPSK, QPSK, 8PSK, 8QAM, and 16QAM signals. Experimental
verifications are also performed to prove the feasibility of the
proposed MFI in long-distance optical fiber transmission systems.
Index Terms—Coherent detection, modulation format identi-
fication (MFI), polarization multiplexing, optical fiber communi-
cation, Stokes space, subtraction clustering.
I. INTRODUCTION
C
OGNITIVE optical networks (CONs) combined with
the techniques of coherent detection and digital signal
processing (DSP), providing flexible bandwidth allocation and
high spectral efficiency, have been considered a promising
solution for next generation optical networks [1]. In CONs,
adaptive transmission parameters, such as modulation format,
data rate, etc., are required at the transmitters of CONs,
to accommodate the dynamic traffic requirements as well as
increase the spectral efficiency [2]. As a result, it is necessary
for the receivers in CONs to blindly retrieve these transmission
parameters from the received signals, so that proper activ-
ities can be adopted subsequently to achieve the optimum
performance. In particular, modulation format identifica-
tion (MFI) becomes essential for the receivers in CONs.
MFI techniques have entensively developed in wireless
communication domain [3]. By contrast, MFI techniques in
the field of optical fiber communication are still at the early
stage with great potential for research. Several MFI approaches
have been proposed in previous literatures for the optical fiber
communication systems [4]–[8]. Among these techniques, the
Stokes-space-based MFI techniques are quite attractive, as they
Manuscript received April 14, 2017; revised June 26, 2017; accepted
June 27, 2017. Date of publication July 5, 2017; date of current version
August 1, 2017. This work was supported in part by the Natural Science
Foundation of China under Grant 61505266 and in part by the Natural Science
Foundation of Guangdong Province, China, under Grant 2014A030310364
and Grant 2016A030313289. (Corresponding author: Jie Liu.)
P. Chen and X. Mai are with the State Key Laboratory of Optoelec-
tronic Materials and Technologies, School of Electronics and Information
Technology, Sun Yat-sen University, Guangzhou 510006, China, and also
with the School of Data and Computer Science, Sun Yat-sen University,
Guangzhou 510006, China.
J. Liu and X. Wu are with the State Key Laboratory of Optoelec-
tronic Materials and Technologies, School of Electronics and Information
Technology, Sun Yat-sen University, Guangzhou 510006, China (e-mail:
liujie47@mail.sysu.edu.cn).
K. Zhong is with the Photonics Research Centre, Department of Electron-
ics and Information Engineering, The Hong Kong Polytechnic University,
Hong Kong.
Color versions of one or more of the figures in this letter are available
online at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/LPT.2017.2723606
Fig. 1. DSP module with proposed MFI method for coherent optical
communication system.
are independent of the carrier phase noise, frequency offset
and polarization mixing [6]–[8]. However, most of Stokes-
space-based MFI schemes proposed by now require machine
learning algorithms with high complexity, such as expecta-
tion maximization algorithm [6], [7], maximum likelihood
algorithm [8], etc.
In this letter, a blind MFI technique with low complexity
is proposed for the coherent optical receivers, on the basis
of subtraction clustering algorithm combined with a hybrid
analysis method. Successful MFI among modulation formats
of BPSK, QPSK, 8PSK, 8QAM and 16QAM can be realized in
the dual-polarization simulation system within practical optical
signal-to-noise ratio (OSNR) range. Comparisons between the
proposed MFI technique and other clustering-algorithm-based
MFI schemes in Stokes space are also performed, showing that
good optical OSNR performance but lower complexity can
be achieved by utilizing our method. In addition, experiment
demonstrations of polarization multiplexed (PM)-QPSK and
PM-16QAM systems are also implemented to verify the pro-
posed MFI technique in long-distance fiber-optic transmission
systems.
II. O
PERATING PRINCIPLES AND SIMULATION RESULTS
The DSP module with MFI capability for coherent
optical communication system is shown in Fig. 1. The
proposed MFI algorithm is placed after the modulation format-
independent algorithms [timing phase recovery and chromatic
dispersion (CD) compensation] [9], [10] and before the modu-
lation format-dependent algorithms (adaptive equalization for
polarization demultiplexing, frequency offset compensation
and carrier phase estimation) [11]. The proposed MFI scheme
mainly includes three steps: Stokes space mapping and least
square plane (LSP) estimation, subtraction clustering as well as
feature extraction and classification. Details will be presented
in the following parts of this section.
A. Stokes Space Mapping and LSP Estimation
Assuming e
x
and e
y
are the two received signals with
polarization multiplexing, which are timing phase recovered,
1041-1135 © 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.