All-optical wavelength conversion and
five-channel multicasting for
20 Gbit/s QPSK signals in a silicon waveguide
Xiaoyan Wang, Lingchen Huang, Ke Yi, Xianglian Feng, and Shiming Gao*
Centre for Optical and Electromagnetic Research, State Key Laboratory of Modern Optical Instrumentation,
Zhejiang University, Hangzhou 310058, China
*Corresponding author: gaosm@zju.edu.cn
Received August 20, 2014; revised September 25, 2014; accepted September 25, 2014;
posted September 26, 2014 (Doc. ID 221394); published October 17, 2014
Wavelength conversion and five-channel multicasting for 20 Gbit∕s quadrature phase-shift keying signals have been
experimentally demonstrated based on four-wave mixing in a silicon waveguide with digital coherent detection.
The eye diagrams and constellation diagrams of the converted and multicasting idlers are successfully observed.
Moreover, the bit-error rates (BERs) of the generated idlers are measured and the power penalties are all less than
0.7 dB at a BER of 3 × 10
−3
. © 2014 Optical Society of America
OCIS codes: (130.3120) Integrated optics devices; (130.7405) Wavelength conversion devices; (190.4390) Nonlinear
optics, integrated optics; (190.4380) Nonlinear optics, four-wave mixing.
http://dx.doi.org/10.1364/OL.39.006122
All-optical signal processing is emerging as a range of key
technologies for next-generation optical communication
networks. In particular, all-optical wavelength conver-
sion, which optically converts one wavelength-division-
multipelxing (WDM) channel to another, and wavelength
multicasting, which copies one input channel onto
several output channels with different wavelengths to
support the data parallelly transmitting to different desti-
nations in wavelength-routing networks, are two poten-
tial operations to increase the bandwidth utilization and
ease the complexity of the transmitter and receiver [
1–3].
Four-wave mixing (FWM) in nonlinear media shows its
advantages for wavelength conversion and multicasting,
such as ultrafast response and strict transparency to the
signal bit rate and modulation format. These advantages
are due to the inherent principle of FWM: both of the
intensity and phase information can be completely pre-
served in the process of generating new wavelengths.
With the rapid development of optical communication
technologies, the system capacity becomes an urgent
issue. High-order and spectrally efficient modulation for-
mats such as quadrature phase-shift keying (QPSK) and
quadrature amplitude modulation (QAM) have been well
concerned to increase the channel capacity and satisfy
the increasing data traffic demand [
4]. Wavelength-
conversion schemes based on FWM (or cascaded
second-order nonlinear processes) for these high-order
modulation formats have been demonstrated in quantum-
dot semiconductor optical amplifiers (QPSK, 8-PSK,
and 16-QAM) [
5], highly nonlinear fibers (QPSK and
OFDM-QPSK) [
6,7], and periodically poled lithium nio-
bate waveguides (DQPSK) [
8].
Compared with the above conventional media, silicon
waveguides have been of great interest because of
the ultrahigh level of integration and high nonlinear co-
efficient. In silicon waveguides, FWM-based wavelength
conversion or multicasting has also been realized
for optical data including amplitude-shift keying (ASK)
[
9] and differential phase-shift keying (DPSK) modulated
signals [
10–12]. For QPSK signals, FWM-based
wavelength-conversion operation has been explored in
a silicon germa nium waveguide [
13] or a silicon-organic
hybrid strip waveguide [
14]. In this Letter, we propose
and experimentally demonstrate both wavelength con-
version and multicasting for 20 Gbit∕s QPSK signals
using FWM in a silicon waveguide. Wavelength conver-
sion and five-channel multicasting are realized by
adopting one or two incident pumps. Eye diagrams, con-
stellation diagrams, and bit-error rate (BER) results are
measured. The power penalties are less than 0.7 dB
at a BER of 3 × 10
−3
for both wavelength conversion
and multicasting.
An idler at ω
I
2ω
P1
− ω
S
will be generated from the
degenerate FWM (DFWM) process in a silicon wave-
guide, as shown in Fig.
1(a). For a QPSK signal, the
sequence of data is modulated on the phase of the signal
optical carrier. Supposing the phases of signal S and
pump P
1
are φ
S
and φ
P1
, respectively, the phase of the
idler will be φ
I
2φ
P1
− φ
S
[15]. If the pump is provided
by a continuous-wave (CW) light, whose phase is reason-
ably supposed to be φ
P 1
0 for QPSK demodulation, the
idler phase is derived as φ
I
−φ
S
. Therefore, the QPSK
sequence on the signal can be converted to the idler with
a new wavelength.
When dual pumps P
1
and P
2
are used, more FWM proc-
esses occur and five useful idlers (I
1
–I
5
) are generated,
as shown in Fig.
1(b). In the five idlers, I
1
, I
3
, and I
4
are
generated by nondegenerate FWMs (NDFWMs) among
Fig. 1. Schematic descriptions of (a) wavelength conversion
based on single-pump FWM and (b) wavelength multicasting
based on dual-pump FWM for QPSK signals.
6122 OPTICS LETTERS / Vol. 39, No. 21 / November 1, 2014
0146-9592/14/216122-04$15.00/0 © 2014 Optical Society of America