ST5D.3.pdf Advanced Photonics for Communications © 2014 OSA
Improved Laser Phase Noise Tolerance for
2.5 Gbaud DPSK Transmission Using
Trellis-Coded Modulation
James Mountjoy
1
, Anthony Walsh
2,3
, Liam P. Barry
2
, and Anthony Fagan
1
1
DSP Research Group, University College Dublin, Ireland.
2
Rince Institute, Dublin City University, Ireland.
3
Tyndall National Institute, University College Cork, Lee Maltings, Cork, Ireland.
james.mountjoy@ieee.org
Abstract: We demonstrate improved performance for differentially encoded phase mod-
ulated optical transmission systems, in the presence of phase noise, by using trellis-coded
modulation.
OCIS codes: (060.5060) Phase Modulation; (060.1660) Coherent communications.
1. Introduction
As next generation optical communications networks progress toward higher data rates and increased spectral effi-
ciency, advanced modulation formats such as quadrature phase shift keying (QPSK) and quadrature amplitude mod-
ulation (QAM) are being employed [1]. In coherent optical communications, phase noise (PN) has been cited as a
limiting factor in performance [2], especially as systems expand to higher-order modulation formats. Trellis-Coded
Modulation (TCM) is a well-known error control technique.
In this paper, we present a TCM scheme tailored to be robust against PN in contrast to conventional TCM schemes
that are designed specifically for additive white Gaussian noise (AWGN). The results in this paper compare TCM
encoded differential QPSK (DQPSK) with an uncoded differential binary PSK (DPSK) signal but the same principle
can be used on higher order PSK constellations and QAM signals.
The simulation results, supported by experimental results, show that TCM gives an enhanced tolerance to large
laser linewidths for phase shift keyed (PSK) signals, especially as the AWGN decreases. The scheme can be used to
improve the overall performance of future coherent systems. A gain of approximately 3 dB is shown at a bit error rate
(BER) of 10
-4
.
2. Trellis-Coded Modulation
In TCM, the redundancy from coding is in the signal space rather than in time [3]. Many TCM schemes have been
designed to deal with AWGN only and use Euclidean distance as a performance metric to inform decoding. We have
devised the following scheme for a system enduring strong PN. Three aspects of the TCM encoding and decoding
process are described in this section.
2.1. Convolutional Encoder
The proposed scheme in this paper differs from [4] (where a Reed-Solomon encoder was used to mitigate the effects of
non-linear PN) primarily in the encoding and decoding schemes. In this work, a convolutional encoder is implemented
T T T T T T
+
+
Input Bit x
i
Output Coded Bit A
i
Output Coded Bit B
i
(a) Convolutional Encoder
00 01
10 11
In-phase
Quadrature
(b) Set Partitioned Constellation
-π -π/2 0 π/2 π
-1
0
1
MSB
LSB
Phase (radians)
(c) Pseudo-Soft Decision Metric
Fig. 1: Encoding and Decoding of Mapping