Compromised extinction and signal-to-noise ratios of weak-resonant-cavity laser diode transmitter...

Compromised extinction and signal-to-noise

ratios of weak-resonant-cavity laser diode

transmitter injected by channelized and

amplitude squeezed spontaneous-emission

Yi-Hung Lin1, Gong-Cheng Lin

2, Hai-Lin Wang

2, Yu-Chieh Chi

1, and Gong-Ru Lin

1

1Institute of Photonics and Optoelectronics, Department of Electrical Engineering, National Taiwan University

No.1 Roosevelt Rd. Sec. 4, Taipei 106, Taiwan R.O.C. 2 Telecommunication Laboratories Advanced Technology, Chunghwa Telecom Co., Ltd., Taoyuan, Taiwan R.O.C.

*[email protected]

Abstract: By using a 200GHz AWG channelized ASE source in connection

with a saturable semiconductor optical amplifier (SOA) based noise blocker

as the injecting source at the remote node in front of the local optical network

units (ONUs), we demonstrate the spectrum-sliced ASE transmitter with

greatly suppressed intensity noise performance in WDM-PON network. Such

channelized SOA filtering technique effectively reduces the relative intensity

noise of the ASE source by at least 4.5 dB. The low-noise WRC-FPLD

transmitter improves its extinction-ratio (ER) from 8.9 to 9.6 dB and

signal-to-noise ratio (SNR) from 5.9 to 6.3 dB. In comparison with

broad-band ASE injection-locked WRC-FPLD transmitter at same power,

there is an improvement on receiving power penalty (∆PReceiver) by 2 dB at

BER 10−9

in back-to-back case, and the receiving power of BER 10−9

can

achieve −24 dBm even after 25km fiber transmission. With additional AWG

filtering, the intraband crosstalk effect between the upstream transmitted data

and the reflected ASE signal is significantly reduced by 6.3dB. The

compromised effects of ER and SNR on BER performance are also

elucidated via the modified SNR model for the WRC-FPLD under ASE

injection induced gain-saturation condition. The ∆PReceiver/∆SNR of 8.89 at

same ER condition is more pronounced than the ∆PReceiver/∆ER of 3.17

obtained under same SNR condition, indicating that the SNR plays a more

important role than the ER on enhancing the BER performance.

©2009 Optical Society of America

OCIS codes: (060.2330) Fiber optics communications; (250.5980) Semiconductor lasers;

(140.3520) Lasers, injection -locked.

References and links

1. H. D. Kim, S. Kang, and C. Lee, “A low-cost WDM source with an ASE injected Fabry–Pérot semiconductor

laser,” IEEE Photon. Technol. Lett. 12(8), 1067–1069 (2000).

2. S. L. Woodward, “P. P. lannone, K. C. Reichmann, and N. J. Frigo, “A spectrally sliced PON employing

Fabry–Pérot lasers,” J. Lightwave Technol. 10, 1337–1339 (1998).

3. K.-Y. Park, and C.-H. Lee, “Intensity Noise in a Wavelength-Locked Fabry–Perot Laser Diode to a Spectrum

Sliced ASE,” IEEE J. Quantum Electron. 44(3), 209–215 (2008).

4. A. McCoy, P. Horak, B. C. Thomsen, M. Isben, and D. J. Richardson, “Noise suppression of incoherent light using

a gain-saturated SOA: Implications for spectrum-sliced WDM systems,” J. Lightwave Technol. 23(8), 2399–2409

(2005).

5. S. Kim, J. Han, and J. Lee, and C. S. Park, “Intensity noise suppression in spectrum-sliced incoherent light

communication systems using a gain-saturated semiconductor optical amplifier,” IEEE Photon. Technol. Lett.

11(8), 1042–1044 (1999).

6. M. Zhao, G. Morthier, and R. Baets, “Analysis and optimization of intensity noise reduction in spectrum-sliced

WDM systems using a saturated semiconductor optical amplifier,” IEEE Photon. Technol. Lett. 14(3), 390–392

(2002).

(C) 2010 OSA 1 March 2010 / Vol. 18, No. 5 / OPTICS EXPRESS 4457#112911 - $15.00 USD Received 18 Jun 2009; revised 3 Sep 2009; accepted 18 Sep 2009; published 19 Feb 2010

7. X. Cheng, Y. J. Wen, Y. Dong, Z. Xu, X. Shao, Y. Wang, and C. Lu, “Optimization of Spectrum-Sliced ASE

Source for Injection-Locking a Fabry–Pérot Laser Diode,” IEEE Photon. Technol. Lett. 18(18), 1961–1963 (2006).

8. D. McCoy, B. C. Thomsen, M. Ibsen, and D. J. Richardson, “Filtering effects in a spectrum-sliced WDM system

using SOA-based noise reduction,” IEEE Photon. Technol. Lett. 16(2), 680–682 (2004).

9. G. P. Agrawal, and N. A. Olsson, “Self-phase modulation and spectral broadening of optical pulses in

semiconductor laser amplifiers,” IEEE J. Quantum Electron. 25(11), 2297–2306 (1989).

10. G. P. Agrawal, Fiber-Optic Communication Systems, Third Ed., (Willy Inter-Science, 2002) Chaps. 4–6.

11. Y.-C. Chang, Y.-H. Lin, J. H. Chen, and G.-R. Lin, “All-optical NRZ-to-PRZ format transformer with an

injection-locked Fabry-Perot laser diode at unlasing condition,” Opt. Express 12(19), 4449–4456 (2004).

12. L. Li, “Static and dynamic properties of injection-locked semiconductor lasers,” IEEE J. Quantum Electron. 30(8),

1701–1708 (1994).

13. K. Petermann, Laser Diode Modulation and Noise, (Publishers Dordrecht, The Nitherlands: Kluwer Academic,

1998), Chap. 2.

14. S. Gee, F. Quinlan, S. Ozharar, and P. Delfyett, “Two-mode beat phase noise of actively modelocked lasers,” Opt.

Express 13(11), 3977–3982 (2005).

15. J. S. Lee, Y. C. Chung, and D. J. DiGiovanni, “Spectrum-sliced fiber amplifier light source for mutichannel WDM

application,” IEEE Photon. Technol. Lett. 5(12), 1458–1461 (1993).

1. Introduction

The high bit-rate transmission of the wavelength-division-multiplexing passive optical network

(WDM-PON) is demanded due to the growing population in future broadband optical access

networks. Versatile cost-effective light sources were emerged to concurrently approach the

colorless injection-locking capability. A typical solution is the FPLD injected by broadband

ASE light source, however, its strong mode-extinction feature under the externally broadband

ASE injection not only opposes the wavelength independent criterion but also constrain the

broadband gain spectrum requirement [1,2]. Recently, an AR-coated weak-resonant-cavity

Fabry–Pérot laser diode (WRC-FPLD) with moderate front-facet reflectance is proposed to

perform the wavelength independent operation [3]. On the other hand, the ASE source

inherently suffers from large intensity noise (IN) caused by spontaneous-spontaneous beat

noise [4,5], such that the spontaneous–spontaneous beating noise injects into the FPLD, which

degrades the signal-to-noise ratio (SNR) and causes the penalty in receiving power for

obtaining up-stream transmitted data with sufficiently low bit-error-rate (BER). There has been

an approach to reduce ASE intensity noise by using a gain-saturated semiconductor optical

amplifier (SOA) [4–6] to filter the spectral-sliced ASE source before injection into an FPLD

[7]. On the other hand, another major reason leading to the limitation on transmission bit-rate up

to 2.488 Gbit/s in such a DWDM-PON is the intra-band crosstalk, which originates from the

interfered effect between the transmitted data and the reflections from the facets of array

waveguide grating (AWG) and feeder fiber patch-cord [8]. The crosstalk of optical reflection

caused by Rayleigh scattering, bad splicing connections, and ASE intensity noise from

spontaneous-spontaneous beat noise strongly affects the signal performance and network

capacity. In this work, we propose a novel DWDM-PON architecture consisting of an AWG

channelized and SOA filtered ASE source at the remote node in front of the local optical

network units (ONUs), which is employed to inject the WRC-FPLD transmitter with front-face

reflectance of 1% for increasing the direct modulation bandwidth and enabling the OC-48

transmission. Adding the AWG and SOA significantly reduces the optical reflection and

intensity noise of the injected ASE source when comparing with the conventional architecture.

When directly modulating the WRC-FPLD at 2.488 Gbit/s in such a DWDM-PON with AWG

bandwidth of 200 GHz, the proposed system benefits from a penalty of −1.8 dB on receiving

power at BER of 10−9

. Moreover, the correlation between SOA operation condition and the

SNR or extinction ratio (ER) are analyzed. The effects of ER and SNR on the BER performance

are also elucidated via the modified SNR model for the WRC-FPLD under ASE injection

induced gain-saturation condition.


2. Experimental setup

Most of previous works established the DWDM-PON system with broadband injecting source

at central office, as shown in Fig. 1(a). In contrast, the Fig. 1(b) schematically illustrates a

modified DWDM-PON system constructed by allocating the spectral-sliced and SOA-filtered

ASE source at remote node for injection-locked WRC-FPLDs at all ONUs. The external

injection-locking source is an Erbium-doped fiber amplifier (EDFA) based broadband source,

which filtered by 200 GHz AWG and then pass through SOA to inject the WRC-FPLD. The

WRC-FPLD exhibits a threshold current of about 25 mA, a longitudinal mode spacing of 0.6

nm, the back and front facet reflectivity of 100% and 1%. The maximum injection power of

WRC-FPLD is limited as −3 dBm due to the injection power too large would damage the

end-face AR coating of WRC-FPLD.

EDFA

Fig. 1. (a). The configuration a conventional ASE injecting transmitter wavelength independent

operation WDM-PON. (b)Configuration of the DWDM-PON with WRC-FPLD injection-locked

by the source of ASE through SOA at the end of remote node.

In our experiment, the operating current of the WRC-FPLD is detuned between 28 and 40

mA, correspond to a change from 1.1 Ith to 1.5 Ith. Latter on, the WRC-FPLD transmitter is

directly modulated at 2.488 Gbit/s with a NRZ pseudorandom bit sequence (PRBS) pattern

length of 223

-1. To suppress the intensity noise of ASE, the SOA has to be operated at high bias

and a large input power is necessary to ensure gain saturation. The maximum input power and

biased current of SOA are limited at −3 dBm and 350 mA, respectively, to prevent of the SOA

from damage.

3. Results and discussions

3.1 Effects of SOA and WRC-FPLD operating conditions on the up-stream transmitted data

performances (BER, SNR, and ER)

The power-current characteristics of the WRC-FPLD under different injection levels are also

shown in Fig. 2, which illustrates a reduction of threshold current by 6 mA with the external

injection power increasing by 9 dB. Note that when using the SOA filtered and spectrum-sliced

ASE source to injection-lock the WRC-FPLD, the threshold current is slight increased as

compared to that of the WRC-FPLD injected by spectrum-sliced ASE without SOA filter. The

SOA introduces additional phase modulation-induced chirp [9] and four-wave mixing (FWM)

effect, which inevibly broadens the spectra output of the AWG channelized ASE. The

broadened ASE spectrum remains the output power constant but attenuates the photon density


at specific wavelength, such that the equivalent optical intensity induced threshold current of

WRC-FPLD will be less affected accordingly. The WRC-FPLD output power can be increased

by at least 3 dB with its threshold current decreasing from 26 to 17 mA under enlarged ASE

injection. It is also important to improve the SNR and ER of the WRC-FPLD transmitted data

under the SOA filtered and AWG-sliced ASE injection with increasing power level.

0 10 20 30 40-0.5

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

Po

wer

(mW

)

Current (mA)

ASE injection -3dBm

ASE through SOA injection -3dBm

ASE injection -12dBm

ASE through SOA injection -12dBm

No injection

Fig. 2. Power-current curves of WRC-FPLD operated without and with injection power of −12

and −3 dBm.

-14 -12 -10 -8 -6 -4 -22.0

2.5

3.0

3.5

4.0

4.5

5.0

5.5

6.0

6.5

7.0

SN

R o

f S

OA

Ou

tpu

t S

ign

al

(dB

)

Input SOA Power (dBm)

Ibias

= 350mA Ibias

= 300mA

Ibias

= 250mA Ibias

= 200mA

Fig. 3. SNR of AWG-sliced ASE without (pink-dotted) and with SOA based filter at different

biased currents.

By externally modulating AWG spectrum-sliced and SOA filtered ASE, the SNR of the data

stream at 2.488 Gbit/s reveals an increasing trend with enlarging bias of SOA and increasing

power of ASE. Without the SOA based noise blocker, the transmitted data exhibits a SNR as

low as 2.3 dB. The SNR of the transmitted data can be improved by 0.8 dB if the SOA bias

increases from 200 to 350 mA. A further increment up to 1.2 dB can be done by increasing ASE

power up to −3 dB. When SOA is operated at 350 mA and the input power is −3 dBm, the SNR

of the transmitted data greatly improves from 2.3 to 6.8 dB, as shown in Fig. 3. Adding the SOA

based noise blocker essentially improves the SNR of upstream data from 6.0 to 6.3 dB. In

general, the SNR of the WRC-FPLD is correlated with the integral of its relative-intensity-noise

(RIN) spectrum by [10]

1

21( ) (1)

2SNR RIN dω ω

π

−+∞

−∞

≅

∫

Technically, the RIN suppression can be done by adding a SOA after the spectrum-sliced

ASE channel prior to injection-lock the WRC-FPLD. The RIN suppression ratio in decibel unit

is theoretically given by [6]


0 0 00 0

20 0

' ( ) ' ( )1 ' ( )( ) ( ) ( )

(2)' ( ) ' ( )1 1

( ) ( )

c

c c

g P z g P z g P zg N g N

h A h A h A

g P z g P z

h A h A

ν τ ν νη

τ ν τ ν

Γ Γ ΓΓ + Γ∝ =

Γ Γ+ +

where Γ is the mode confinement, g(N) is the gain coefficient, αint is the internal loss, J is the

injected current density, d is the active layer thickness of the SOA, A is the active region area, q

is the electronic charge and τc is the carrier lifetime. ω is the angular frequency at which the

perturbing signal varies, g' is the differential gain coefficient, P0, and N0 are the time-averages

of the output power and equivalent carrier density, and ∆P(ω) and ∆N(ω) stand for the noise.

Equation (2) shows that P0, Γg(N0) and Γg’ must be extremely large for obtaining high SNR.

For a given SOA, P0 and Γg(N0) can be increased by enlarging the input power Pin and the

biased current Ib. In experiment, we have measured the RIN spectra of the spectrally sliced ASE

source without and with additional SOA based noise blocker, and the WRC-FPLD under

free-running and injection-locking conditions, as shown in Fig. 4.

0 1 2 3 4 5 6

-138

-136

-134

-132

RIN

(d

B/H

z)

Frequency (GHz)

Free-running WRC-FPLD

ASE

SOA+ASE

SOA+ASE+WRC-FPLD

Fig. 4. Measured the RIN of the WRC-FPLD injection locked by different light sources.

-33 -30 -27 -24 -21 -18 -1512

11

10

9

8

7

6

5

-Lo

g (

BE

R)

Receving Power (dBm)

SOA Ibias

= 350mA

SOA Ibias

= 250mA

SOA Ibias

= 81mA

ISOA

Fig. 5. WRC-FPLD up-stream BER under the injection of AWG-sliced ASE with SOA filter at

different biases.

The original RIN of the spectrally sliced ASE source exhibits a RIN of −135 dB/Hz below 6

GHz. By adding the SOA based noise blocker, it is clearly seen that the RIN of the ASE source

decreases by 2 dB at f<3 GHz. With such an SOA filtered ASE injection, the WRC-FPLD

greatly reduces its RIN from −133 dB/Hz (free-running case) to −138 dB/Hz. These results have

elucidated that the WRC-FPLS essentially reduces its RIN by at least 4 dB after

injection-locking with the SOA filtered ASE source, such that the transmission performance of

the WRC-FPLS can further be improved. The BER of the upstream transmitted data from

WRC-FPLD injection-locked by the ASE with SOA filter at different biased currents are

analyzed in Fig. 5. When the SOA biased current is enlarged from 81 to 350 mA, the BER error


floor is decreased from 10−9

to 10−11

, reflecting that the RIN of the WRC-FPLD is significantly

diminished under SOA filtered ASE injection.

-32 -30 -28 -26 -24 -22 -20 -18 -16

12

11

10

9

8

7

6

5

4

3

-lo

g(B

ER

)

Receiving Power (dBm)

(a)

(b)(c) (d)

Fig. 6. BER and corresponding eye-diagrams of the SOA filtered ASE injection-locked

WRC-FPLD at different biases of (a) 28 mA, (b) 32 mA, (c) 36 mA, (d) 42 mA.

The WRC-FPLD biased current dependent BER analysis at bit rate of 2.488 Gbit/s is shown

as Fig. 6. The BER performance is greatly improved as the WRC-FPLD biased current

increases from 28 to 36 mA (i.e., from Ith to 1.4 Ith), however, which turns to be degraded when

the WRC-FPLD bias further increases to 40 mA or larger. At optimized bias of 36 mA, the

receiving power required for the up-stream transmitted data is −26 dBm at BER of 10−9

. At

lower biased condition, the SNR of WRC-FPLD transmitted data predominates the BER

performance, and there is a positive contribution of biased current for enhancing the SNR as

well as BER. Nonetheless, the ER becomes more pronounced than SNR to degrade the BER

performance as the WRC-FPLD biased current increases beyond 36 mA, while the ER

significantly degrades at higher biased current to introduce a positive power penalty of 4 dB at

BER of 10−9

.

28 30 32 34 36 38 408

9

10

11

12

13

4.0

4.5

5.0

5.5

6.0

6.5

ER

(d

B)

WRC-FPLD Biased Current (mA)

SN

R (

dB

)

Fig. 7. SNR and ER of the ASE injection-locked WRC-FPLD at different currents.


28 30 32 34 36 38 405.4

5.6

5.8

6.0

6.2

6.4

Q F

acto

r

WRC-FPLD Biased Current

BER of 10-9

Fig. 8. The calculated Q factor of WRC-FPLD at different currents

Figure 7 illustrates the SNR and ER as a function of the WRC-FPLD biased current.

Apparently, the SNR increases 5.1 to 6.4 dB as the WRC-FPLD gain enlarges when increasing

its bised current from 28 to 40 mA. However, the lowest biased current (28 mA) can provide the

highest ER of 13.5 dB for the WRC-FPLD transmitted data, whereas the ER oppositely

degrades to 8.2 dB at biased current of 40 mA. In experiment, we observe that the effect of the

ER on BER performance can only be neglected if the ER of the WRC-FPLD tansmitted data is

larger than 9 dB, and SNR becomes dominated at highly biased current. Theoretically, the Q

parameter which decides the lowest BER achieved in a communication system can be described

in terms of Q = [(ER-1)/(ER + 1)](M⋅SNR)0.5

with M denoting the gain of the optical receiver.

With increasing WRC-FPLD bias, the factor of (ER-1)/(ER + 1) is decreased from 0.89 to 0.74,

where the (SNR)1/2

is only improved from 1.6 to 2.1. That is, the Q parameter behaves like a

nonlinear function of WRC-FPLD bias, as shown in Fig. 8.

3. 2 Theoretical and experimental analyses on the ASE injection power dependent SNR, ER, and

BER of WRC-FPLD up-stream data

To investigate the effects of ASE injecting power and reflected ASE signal on the WRC-FPLD

transmitted upstream data, we further consider the SNR and ER of the injection-locked

WRC-FPLD under gain-saturation condition. Originally, the SNR of WRC-FPLD under

external injection is given by [10]

22

2 2

( )SNR = = , (3)

4

inj inj

out

sp

RGP GPI

S fσ σ< >

≈∆

where Pinj is the injection power, G is the optical gain, ∆f is the detector bandwidth, Ssp is the

spectral density of spontaneous-emission induced noise. Due to the gain saturation of the

WRC-FPLD under ASE injection (occurred at Psat > −7 dBm), the power gain of WRC-FPLD is

rewritten as

0

( ) 1exp[ ] (4)

( )

out inj out

out inj sat

P P PG G

P P P

−=

However, the power gain of WRC-FPLD is greatly modified under the ASE external

injection-locking condition as [11–13]

( )

2 2

0 2

41(5)

1

sp injc c

p m p mc

R h PkG G G

P P

ν βτ τ β

= −∆ = − −+


where G0 is the WRC-FPLD gain at free-running case, ∆G is the gain variation due to the SOA

filtered and AWG-sliced ASE injection, τp is the photon lifetime, Rsp denotes the spontaneous

emission rate, Pm denotes the total power at the injection-locked mode, βc denotes the linewidth

broadening factor, kc denotes the coupling coefficient of the WRC-FPLD for external injection,

and Pinj is the externally ASE injected power.

( )

( )

( )

, ,

2 2

2

,[ ] (6)

g'

1

'

4

' 1

out d th inj d th inj

i c

inj inj

d tr

i c

tr

i c p

d

sp injc c

i c m p mc

qh hP I I I N

q q

G N PqhI N

q

qI N

gh

q R h Pkq

g P P

ν νη η

η τ

νη

η τ

η τ τν

ην β

η τ τ β

= − = −

= − +

− + +

= + +

where I and Ith are the bias and threshold currents of WRC-FPLD, Ntr is the transparent carrier

number, g’ is the differential gain coefficient for ∆N denoting as the variation of carrier

numbers, ηd is the differential quantum efficiency, hν is the energy per photon, ηi denotes the

internal quantum efficiency, and τc denotes the carrier lifetime. When the AWG-sliced ASE

injection-locked WRC-FPLD is operated in linear-gain region, the Eqs. (3) and (5) are

combined to give the SNR as

( )

2 2

2

41SNR (7)

4 1

inj sp injc c

out

sp p m p mc

P R h Pk

S f P P

ν βτ τ β

≈ − −

∆ +

If the WRC-FPLD is operated at gain-saturated condition, the SNR is modified using Eqs.

(3)-(6), as described by

( )

( )

2 2

2

2 2

2

41 1SNR

4 ' ' 1

1

'1

exp4

' 1

sp injc c

out d tr

sp i c p i c m p mc

tr

i c p

d

sat sp injc c

i c m p mc

R h Pkq qhI N

S f q g g P P

qI N

gh

P q R h Pkq

g P P

ν βνη

η τ τ η τ τ β

η τ τνη

ν βη τ τ β

≈ − + + + ∆ +

− + +

+ +

(8)inj

P

−

As a result, the measured SNR and ER of the back-to-back transmitted WRC-FPLD up-stream

data are shown in Fig. 9. The SNR linearly increase with injection power, but saturate at high

injection level. At the same time, the P-I curve of WRC-FPLD illustrates a decreasing trend on

its threshold current with increasing ASE injection power. Note that the power-current slope of

the ASE injection-locked WRC-FPLD almost remains as constant with varying ASE injection

power, indicating that the reshaping on rising and falling edge by changing ASE injection

power is negligible. If we fix the WRC-FPLD bias and the PRBS NRZ amplitude, the

“off-state” level of upstream transmitted data will increase with larger ASE injection power,

whereas the “on-state” level remains almost unchanged. Thus, the ER monotonically decreases

with increasing ASE power when the WRC-FPLD enters into gain-saturation condition.


-12 -10 -8 -6 -4 -28.5

9.0

9.5

10.0

10.5

11.0

11.5

12.0

12.5

13.0

4.0

4.5

5.0

5.5

6.0

6.5

ER

(d

B)

ASE Injection Power (dBm)

New system

New system without SOA

Conventional system

SN

R (

dB

)

Fig. 9. SNR and ER of WRC-FPLD injection-locked by ASE with changing power levels in

different systems.

-32 -30 -28 -26 -24 -22 -20

12

11

10

9

8

7

6

5

Conventional system after 25km

New system without SOA after 25km

New system after 25km

Conventional system BTB

New system without SOA BTB

New system BTB

-lo

g(B

ER

)

Receiving Power (dBm)

Fig. 10. BER of filtered ASE with SOA (new system), filtered ASE (new system without SOA)

and broadband ASE injection-locked WRC-FPLD based WDM-PON (inset: back-to-back eye

diagram of (a) old system, (b) new system without SOA, (c) new system with SOA).

In comparison with conventional WRC-FPLD based DWDM-PON, the use of additional

200 GHz AWG placed after the ASE source effectively suppress the crosstalk and interfered

reflection between the upstream transmitted data and reflected ASE signal. Such a broadband

reflection from between the based DWDM Mux and DeMux is greatly reduced from −14.7 to

−21 dBm with a decreasing of 6.3 dB. We also find that the Schawlow-Towne’s linewidth

(∆ωst) becomes narrower in the new system even without SOA. Lower mode beating noise is

obtained as the mode extinction becomes significant by reducing the ASE reflection level when

comparing with conventional injection-locking scheme (with the ASE located in OLT) [14].

Due to the reduction of noise floor and mode beating noise, both the SNR and ER are improved

from 5.9 to 6.1 dB and from 8.9 to 9.3 dB, respectively. It is clear seen that adding the SOA

based noise blocker can further improve the SNR of upstream data by 0.3 dB. Moreover, the

“off-state” level of upstream data will decrease by slightly increasing the threshold current of

WRC-FPLD, such that the ER can essentially be ehanced from 9.3 to 9.6 dB. Apparently, the

combining effect of the 200 GHz-AWG based spectral slicer and the SOA based noise blocker

located after the ASE source essentially improves the ER from 8.9 to 9.6 dB and SNR from 5.9

to 6.3 dB. At ASE injection power of −3 dBm, the BER performances of the ASE

injection-locked WRC-FPLD transmitter with three kinds of ASE sources located at different


nodes are compared in Fig. 10. At BER of 10−9

, the newly proposed system shows a smallest

receiving power sensitivity of −26 dBm, and there is a negative power penalty of 1.8 dB as

compared to the conventional system in back-to-back case. After 25-km transmission in

single-mode fiber, the receiving power of BER 10−9

degrades to −23.8 dBm, whereas the system

withput SOA or the conventional system fails to achieve same BER level. The additional SOA

filtering leads to a negative power penalty by 1 dB due to the improved SNR and ER under the

reductions of crosstalk and mode beating noise.

3. 3 Distinguished influence of SNR and ER to the BER performance of the WRC-FPLD

transmitted up-stream data

To discriminate the individual contribution of ER and SNR to the BER, we adjust the ASE

injection power and the WRC-FPLD biased current to detune ER and SNR of upstream

transmitted data. By keeping the ER or SNR as constant, we investigate the effect of another

parameter on the BER, as shown in Fig. 11. Under the same ER of 10 dB, the receiving powers

of BER at 10−9

are required to exceed −26 and −18 dBm at SNR of 6.3 and 5.3 dB, respectively.

With constant SNR of 6.3 dB, the requested receiving powers of BER at 10−9

are enlarged from

−26 to −22 dBm when ER is degraded from 10 to 8.6 dB. As a result, the sensitivity slopes for

SNR and ER are determined as ∆PReceiver/ ∆SNR = 8.9 and ∆PReceiver / ∆ER = 3.2, which clearly

elucidate that the SNR plays a more important role than ER on BER performance.

-33 -30 -27 -24 -21 -18 -1512

11

10

9

8

7

6

5

4

-Lo

g (

BE

R)


SNR=6.3 dB, ER=10 dB

SNR=5.3 dB, ER=10 dB

-33 -30 -27 -24 -21 -18 -1512

11

10

9

8

7

6

5

4

-Lo

g (

BE

R)


SNR = 6.3 dB, ER = 10 dB

SNR = 6.3 dB, ER = 8.6 dB

Fig. 11. BER analysis of AWG-sliced and SOA-bleached ASE injection-locked WRC-FPLD

transmitter with changing SNR (left) and changing ER (right).

-33 -30 -27 -24 -21 -18 -1512

11

10

9

8

7

6

5

(b)

(c)

-Lo

g (

BE

R)


(a)

Fig. 12. BER of the WRC-FPLD transmitted data in the DWDM-PON systems with changing

ASE-Demux-Mux linewidths of (a) 1.0-1.0-1.0 nm, (b) 0.5-1.0-1.0 nm, and (c) 0.5-0.5-0.5 nm.

Moreover, we also observe the biased current of WRC-FPLD is more pronounced than the

injection power for ER and SNR of the upstream transmitted data. Apparently, the currents of

SOA and WRC-FPLD are the most important parameters to achieve better transmission

performance of the upstream data. In addition, it is worthy noting that the ASE channel

linewidth filtered by AWG also affects the WRC-FPLD signal quality. In the experiment, the

channel linewidth of 100 and 200 GHz AWG are 0.51 nm and 1.03 nm individually. In back-to


back case, the BER analysis of upstream transmitted data through the AWG with different

channel spacing is shown in Fig. 12. When the 200GHz AWG is replaced by a 100 GHz one to

filter ASE, the error floor of –log (BER) is degraded from 10.8 to 8.5. In the mean time, the

SNR is degraded from 6.3 to 5.4 dB and ER is increased from 9.6 to 10 dB concurrently. In

general, the EDFA or SOA based ASE light source exhibits two different noises, including the

signal-spontaneous beat noise, and the spontaneous-spontaneous beat noise, which

concurrently cause the degradation on the SNR of signals. In our case of the injection-locked

WRC-FPLD with external ASE injection, the signal-spontaneous beat noise no longer sustains

but the ASE-ASE beat noise injects into the WRC-FPLD and affects the noise performance of

WRC-FPLD. The equation of SNRASE is defined as SNR = P2ASE/(P

2sp-sp + P

2shot + P

2thermal) [15],

where PASE is the ASE average power, Psp-sp is the spontaneous to spontaneous beat noise

average power, Pshot is the shot noise average power, and Pthermal is the thermal noise average

power. The shot noise and thermal noise are ignored as they are usually not in optical path. Due

to the relationship of P2sp-sp = 2P

2ASEBe/m∆λ, the SNR can be rewritten as SNR = m∆λ/2Be,

where Be is the electrical bandwidth of the optical receiver, m is the number of polarization, and

∆λ is the spectral linewidth of the ASE injection-locked WRC-FPLD transmitter. It is straight

forward that the SNR as well as BER can be degraded by decreasing the spectral linewidth of

the injected ASE source. Therefore, if we employ the AWGs with channel spacing of 100 GHz,

it is clearly seen that the BER will be seriously degraded with a BER error floor as high as

4x10−7

. If we further release the transmission channel linewidth in the DWDM-PON by using a

200 GHz AWG based Mux and Demux with spectral window of 1.0 nm (twice larger than that

of a 100 GHz AWG based ones). The BER error floor of –log(BER) is greatly reduced from 6.5

to 8.4, while the SNR is enhanced from 4.2 to 5.5 dB and the ER is increased from 9.2 to 10 dB

concurrently. In comparison, the broadening in ASE filter linewidth further improves the BER

error floor of the AWG channelized and SOA filtered the ASE injection-locked WRC-FPLD

upstream transmitted data by four orders of magnitude down to 10−11

. That is, there is an

extremely large power penalty up to 12 dB when reducing the ASE spectral linewidth from 1.0

to 0.5 nm before injection-locking the WRC-FPLD. This again corroborates that the SNR

becomes the more pronounced effect than ER for promoting the BER performance. In addition,

it is also concluded that the AWG channelized ASE injection-locked WRC-FPLD transmitter

fails to promote the channeling capacity of the DWDM-PON system from 200 GHz to 100

GHz. Even the origined ASE spectral linewidth is broadened to 1.0 nm, the upstream

transmission linewidth is still constrained by the AWG based Mux/Deux channel window. This

eventually leads to the degradation of the receiving power sensitivity increasing from −27 to

−15 dBm at BER if 10−9

. From the ∆SNR, ∆ER and BER of this experiment, we conclude the

linewidth of AWG before WRC-FPLD has dominated effects on upstream signal quality, and

the SNR plays a more important role than ER on the WRC-FPLD upstream data quality as well.

4. Conclusion

By using a 200GHz AWG channelized ASE source in connection with a saturable

semiconductor optical amplifier (SOA) based noise blocker as the injecting source at local

ONU part, we demonstrate the spectrum-sliced ASE injection-locked WRC-FPLD transmitter

directly modulated at 2.488 Gbit/s with greatly suppressed intensity noise performance in a

200GHz channelized DWDM-PON network. The experiments show that the low noise

WDM-PON improves ER from 8.9 to 9.6 dB and SNR from 5.9 to 6.3 dB. In comparison with

conventional (broad-band) ASE injection-locked WRC-FPLD transmitter at same power, there

is an improvement on receiving power penalty (∆PReceiver) by 2 dB at BER 10−9

in back-to-back

case, and the receiving power of BER 10−9

can achieve −24 dBm even after 25km fiber

transmission. Besides, with the additional AWG filtering, the crosstalk effect between the

upstream transmitted data and the reflected ASE signal can be great reduced by 6.3 dB. The

compromised effects of ER and SNR on BER performance are also elucidated via the modified

SNR model for the WRC-FPLD under ASE injection induced gain-saturation condition. The


∆PReceiver/∆SNR of 8.89 at same ER condition is more pronounced than the ∆PReceiver/∆ER of

3.17 obtained under same SNR condition, indicating that the SNR plays a more important role

than the ER on enhancing the BER performance. In addtion, we research the impacts of the

AWGs channel linewidth for the WDM-PON. From the results, we successfully find the

currents of SOA and WRC-FPLD are the most important parameters for upstream signal and

the linewidth of AWG before WRC-FPLD has larger effects than the AWGs after WRC-FPLD

on upstream signal quality in this WDM-PON.

Acknowledgment

Financial support by the National Science Council of Taiwan Republic of China under grants

NSC98-2221-E-002-023-MY3 and NSC 98-2623-E-002-002-ET are acknowledged.


Compromised extinction and signal-to-noise ratios of weak-resonant-cavity laser diode transmitter...

Documents

Transcript of Compromised extinction and signal-to-noise ratios of weak-resonant-cavity laser diode transmitter...