Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 13

Improvement of Performance of a Semiconductor Optical Amplifier based Delayed

Interference Signal-wavelength Converter with Phase Offset of a Mach-Zehnder

Delayed Interferometer

Masatoshi Namiki, Takayoshi Mori, Hiroyuki Uenohara and Kohroh Kobayashi

Photonics Integration System Research Center, Precision and Intelligence Laboratory,

Tokyo Institute of Technology, Yokohama, Kanagawa 226-8503, Japan

We investigated improvement of wavelength conversion performance of a SOA based

Delayed Interference Signal-wavelength Converter (DISC) with phase offset of a

Mach-Zehnder Delayed Interferometer(MZDI). In experiments, power penalty of

wavelength conversion output and transmission characteristics were improved with

negative phase offset of -0.12rad at 10Gbps. We also investigated its effect in

simulations and studied the principle of the effect.

1
Nowadays, the Internet traffic is increasing rapidly and larger transmission

capacity network and higher speed, larger throughput routers are required. However, an

electrical router needs higher consumption power at higher speed. This issue has been

one of the most serious problems recently. So much attention has been paid to optical

packet switches1) due to their high-speed performance and low power consumption. And

a buffering system2) is indispensable for optical packet switches since packet loss

probability of optical packet switches is increasing in case of several packets going to

the same output port. We investigate an optical buffer consisting of a scheduler, optical

switches and delay lines. As an optical switch, wavelength switching scheme using a

delayed interference signal-wavelength converter (DISC)3) is introduced because low

packet loss can be expected due to WDM and number of devices can be reduced due to

the usage of single semiconductor optical amplifier (SOA). For using the DISC as an

optical switch, high wavelength conversion performance is required such as small signal

degradation. As for one of the methods to improve wavelength conversion performance

of the DISC, improvement of extinction ratio by using the phase bias adjustment has

been reported4). However, improvement of bit error rate (BER) with phase offset has not

been evaluated yet. And improvement of transmission characteristics is also important

for the DISC for optical switches, but the detailed optimized conditions have not been

reported so far.

In this letter, we demonstrate improvement of wavelength conversion performance and

transmission characteristics of the DISC with phase offset in experiment. And we also

report the investigation of the effect of phase offset by simulation.

An experimental setup of a DISC is shown in Fig.1. Input signal was 2 7-1

PRBS with RZ format at a bit rate of 10Gbps.Its wavelength and average power were

2
s=1550nm and Ps=-6dBm, and it was generated with cascaded 2-stage electro-

absorption modulators (EAMs). It was input into the output port of the SOA through an

optical circulator. The input signal waveform is shown in Fig.2. The SOA we used was a

commercial product (SOA-NL-OEC-1550, CIP product). Injection current and 10% -

90% gain recovery time were 300mA and ~60ps, respectively.

CW probe light at a wavelength of p=1540nm with power of Pp=3dBm was also input

into an input port of the SOA. Carrier density of the SOA is changed by input signal and

phase of probe light is modulated due to the change of refractive index, so called cross

phase modulation. Output probe light from the SOA was injected to the Mach-Zehnder

Delayed Interferometer(MZDI) which has a voltage controlled phase bias in one arm

with FSR=40.14GHz. In the MZDI, phase-modulated probe light by input signal is

divided into two, and one is delayed. The optical phase bias between divided probe

lights is set to ‘’ and divided light is interfered with the other destructively to generate

wavelength converted output.

Then, we measured waveforms and BER of wavelength conversion outputs with an

oscilloscope and an error detector, respectively. We investigated improvement of

wavelength conversion performance in terms of power penalty of the receiver

sensitivity at a BER = 10-9 compared with input signal.

Output signal waveforms are indicated in Fig.3. Output signal waveform with

phase offset of -0.12rad was improved from ones without phase offset. Power penalty

(triangle in Fig.4) was 0.8dB with phase offset of -0.12rad, which was the lowest and

improved by 1.6dB from one without phase offset.

Here, phase offset is defined as the shift from’’ phase bias of the MZDI in one

arm and calculated from difference between center wavelength of DISC output and

3
wavelength of null trasmissivity of the MZDI. We should notice that the resolution of

optical spectrum analyzer for measurements of their wavelengths was not enough for

measurement of -0.12rad offset. So phase offset may have some minor deviations of

±0.1rad.

Improvement of transmission characteristics of the DISC is also important

since the DISC is used in optical packet switch and the output of one optical packet

switch is transmitted to another.   We investigated improvement of transmission

characteristics with -0.12rad offset, which was suitable for back to back condition. The

squares in Fig.4 show power penalty of wavelength conversion output after 25km

transmission. Power penalty with phase offset of -0.12rad was improved by 1.7dB from

one without phase offset. Consequently, we confirmed that phase offset is effective for

improving DISC waveform both before and after transmission and we can use the same

condition.

In this section, we show numerical simulation results investigating the effect of

phase offset and compare simulation waveforms to experimental ones. In simulation, we

used Transfer Matrix Method and Rate Equation expressed by Eqn. (1).

∂Ni I
= −N i ( c1 + c2 N i+ c 3 N i2 )− ∑ v g Γgm S w ,i ............................ (1)
∂ t qV w=1,2,3
w, i

where I is injection current(300mA in this simulation), N is carrier density ,S is photon

density and gm is material gain. Index i corresponds to different small section of SOA

into which we divide for Transfer Matrix Method and w refers to different input lights,

which are input signal, CW probe and ASE light. SOA parameters used for simulation

are summarized in Table 1.We used a model of gain profile of the SOA that took the

variation of gain spectrum with carrier density into account5).

4
In simulation and experiment in this section, wavelength, FSR of a MZDI,

SOA injection current, bit rate and data stream of signal light were set at the same

parameters as mentioned in Section 3 though pulse width of input signal was narrower

than that used in experiment. And we changed input optical power to optimize DISC

waveforms.

Figs.5 (a) and (b) show extinction ratio (ER) and Q factor against phase offset

in experiment and simulation, respectively. They are expressed by Eqns. (2) and (3) as

P1
ER= .................................................... (2)
P0

P1−P0
Q= .................................................... (3)
σ 1−σ 0

where P1 and P0 are average power of “1” and “0” level of the signal, and 1 and 0 are

distribution of “1” and “0” level.

ER and Q factor were improved with negative offset (around -0.1rad) and we

confirmed the effect of phase offset also in simulation (Fig. 5). And ER and Q factor

were also improved with negative offset (around -0.1rad) after 25km transmission as

shown in Fig.6 so the same phase offset can be suitable for both before and after

transmission. Here, the way of measurements of phase offset in the experiment was the

same as Section 3.

We consider the reason why DISC waveform was improved with negative

phase offset. Fig. 7 shows DISC output and phase shift of probe light which are divided

into two parts and one of them is delayed in a MZDI. The delayed one has  rad phase

bias from the other so phase difference between two lights is  rad when phase shift of

two lights are the same value in Fig. 7. In the case of 0rad offset, phase shift of delayed

light decrease across phase shift of the other, that is, phase difference between two lights

5
exceeds rad. So extra subpulse is generated behind a main pulse (Fig. 7 (a)). Subpulses

cause degradation of DISC waveform. Phase offset can manage exceeding phase shift of

probe light and suppress subpulses as shown in Fig.7 (b). However, too large phase

offset makes “0” level of DISC output larger than the desired level. So, investigation of

appropriate phase offset is important to improve DISC performance.

In summary, we investigated improvement of wavelength conversion

performance of the DISC with phase offset from BER measurements. Measurement of

power penalty of wavelength converted output signal was examined at 10Gbps. Under

back to back condition, it was improved by 1.6dB with phase offset of -0.12rad in

comparison to one without offset. And it was also improved by 1.7dB after 25km

transmission with the same offset. In simulation, the phase offset was also improved in

both before and after transmission case. And we also showed that suppression of

subpulses with negative phase offset was the main cause of improvement.

6
References

1) N.Wada, H.Harai, and F.Kubota: IEICE Trans. Electron. E87-C (2004) 1090.

3) W.D.Zheng, R.S.Tucker: J. Lightwave Technol. (1998) 1737

3)J.Leuthold, B.Mikkelesen, G.Raybon, C.H.Joyner, J.L.Pleumeekers, B.I.Miller,

K.Dreyer and R.Behringer: Opt. Quantum Electron. 33 (2001) 939.

4) B.Zhang, S.Kumar, L.S. Yan and A.E.Willner: Opt. Commun. 280 (2007) 202.

5) Satoshi Shimizu and Hiroyuki Uenohara: Proc. 14th Opto-Electronics and

Communications Conf. (2009) No.WO5.

7
Fig. 1 Experimental setup of operation performance of a SOA-based DISC.

Fig. 2 Input signal waveform.

Fig. 3 Waveforms of wavelength conversion output with phase offset of (a) 0rad and (b) -0.12rad.

Fig. 4 Power penalty of wavelength converted output signal with various phase offset.

(triangle : Back to Back, square:25km transmission)

Fig. 5 Extinction Ratio and Q factor of wavelength converted output signal under back to back

condition with various phase offset (a) in experiment(triangle: Extinction Ratio, square: Q factor)

(b) in simulation (solid line: Extinction Ratio, dashed line: Q factor).

Fig. 6 Extinction Ratio and Q factor of wavelength converted output signal after 25km transmission

with various phase offset (a) in experiment(triangle: Extinction Ratio, square: Q factor) (b) in

simulation (solid line: Extinction Ratio, dashed line: Q factor).

Fig. 7 Intensity and phase shift of DISC waveforms with (a)0rad phase offset and (b)-0.1rad phase

offset.

8
Table 1 Parameters of SOA for simulation
Symbol(unit
Description Value
)
Optical Confinement factor  0.14
Spontaneous emission coefficient  2.0010-5
Carrier density at transparency N0 (m-3) 1.921024
Group index ng 4.00
Differential refractive index dn/dN (m3) 1.9310-26
SOA length L (m) 2.0010-3
Active layer volume V (m3) 1.8210-16
c1 (s-1) 3.00108
Recombination rate c2 (m3/s) 7.5010-17
c3 (m6/s) 4.7010-40
Nonlinear gain compression coefficient m3) 1.30×10-23
Loss in active layer a (m-1) 1.40×104
Loss in cladding layer c (m-1) 2.00×103
Scattering loss scat (m-1) 1.00×102

10Gbps RZ Signal
Clock
4chPPG Data 0.35
0.3
Intensity[arb. unit]

1550nm 0.25
T-LD1 EAM EAM 0.2
0.15
0.1
ED
1 0.05
T-LD2 SOA 2 3 OATT O/E OSC 0
1540nm 300mA MZDI BPF 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Δt=25ps 0.3nm Time[ns]

Fig. 1 Fig. 2

9
0.35 0.35
(a) (b)

Intensity[arb. unit]

Intensity[arb. unit]
0.3 0.3
0.25 0.25
0.2 0.2
0.15 0.15
0.1 0.1
0.05 0.05
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time[ns] Time[ns]

Fig. 3

25 25
Extinction Ratio[dB]

20 20
1.7dB
15 15
10 10 Q factor
1.6dB
5 5
0 0
-0.4 -0.3 -0.2 -0.1 0
Phase Offset[rad]

Fig. 4

10
25 25

Extinction Ratio[dB]
20 (a) 20
15 15

Q factor
10 10
5 5
0 0
-0.4 -0.3 -0.2 -0.1 0
Phase Offset[rad]

(b)

Fig. 5

11
20 20

Extinction Ratio[dB]
(a)
15 15

Q factor
10 10

5 5

0 0
-0.4 -0.3 -0.2 -0.1 0
Phase Offset[rad]

(b)

Fig. 6

12
Fig. 7

13

You might also like