Download as pdf or txt
Download as pdf or txt
You are on page 1of 3

OSA / OFC/NFOEC 2010

a889_1.pdf
OThI4.pdf

Burst-Mode Optical Amplifier


Yoshinari Awaji, Hideaki Furukawa, Benjamin J. Puttnam, Naoya Wada
National Institute of Information and Communications Technology (NICT), 4-2-1, Nukuikita, Koganei, 184-8795 Tokyo, Japan
{yossy, furukawa, wada}@nict.go.jp, bputtnam@ee.ucl.ac.uk

Peter Chan, Ray Man


Amonics ltd., Unit 1803-1804, 18/F, Perfect Industrial Building, 31 Tai Yau Street, San Po Kong, Kowloon, Hong Kong
{peter, rayman}@amonics.com

Abstract: A series of technologies burst-mode amplification are described. We optimize EDFA


design to establish transient-suppressed EDFA (TS-EDFA) and discuss two scenarios of
burst-mode amplification in WDM environment. We also investigate supplementary
transient-suppression with optical-feedback.
2010 Optical Society of America
OCIS codes: (060.2320) Fiber optics amplifiers and oscillators, (060.4259) Networks, packet-switched

1. Introduction

50ns/div

50ns/div

negative exponent of BER

A granular signal input into EDFA will be distorted by dynamic gain transience. It has been serious issue [1] in
-switched networks and, more recently, in optical packet-switched (OPS) networks. In these networks, impairment
arising from the transience appears as a power penalty (Fig. 1) and reduction of receiver tolerance.

Baseline ( = 0.81%)
Amplified ( = 0.81%)

5.6dB

6
7
8
9

(a)
(a)

Baseline ( = 34%)
Amplified ( = 34%)

-20

-18

-16

-14

-12

-10

Received power [dBm] (calibrated)

-8

(b)

Fig. 1 Gain transience of EDFA and its impact


Distortion of packet envelope by gain transience of EDFA, 500bytes@10Gbps, Traffic density ()=0.81%
(b) Power penalty raised by the transience, two cases of traffic density () were measured.

Two major categories of counter-measure techniques to this issue have been electronic gain control [2] and gain
clamping by optical feedback [3]. Both schemes need rather complex configuration in spite that EDFA is necessary
almost everywhere in optical network. Hence, a simpler and cheaper solution is needed.
Here, we describe the design and optimization of an EDFA specifically for burst-mode operation, to construct an
intrinsically transient-suppressed EDFA (TS-EDFA). The principle behind the TS-EDFA is not compensation but
mitigation. Therefore, there is no bandwidth limitation for fast response. Moreover, it was just the improvement of
gain-block, hence TS-EDFA can be used with any existing supplementary control techniques in order to enhance
their performance.
In this paper, we provide an overview of the prototype TS-EDFAs and burst-mode operation in a WDM
environment. We also review the supplemental combination of TS-EDFA with the optical feedback technique.
2. Design principle of TS-EDFA and prototypes
The gain transience of EDFA can be approximated by following formula [4].

G ' ( 0) =

[G () G (0)]

[1 + j

P out ( j )
P ( j )
IS

P IS ( j ) =

hS
[ a ( j ) + e ( j )] j 0

,
G(0) is the gain before the transience and G() is the steady gain after the transience, respectively. 0 is the
intrinsic lifetime of the upper level of the erbium ions (assuming, a two level systems of Er3+ even for three level
pumping). PIS is the intrinsic saturation power at the wavelength channels. S is the active erbium area of EDF, a and
e are the absorption and emission cross section at the wavelength channels, is the overlap factor between the Er3+
ions and the mode field of light, respectively. In this formula, signal and pump light can be treated identically [5].

978-1-55752-884-1/10/$26.00 2010 IEEE

OSA / OFC/NFOEC 2010

a889_1.pdf
OThI4.pdf

The G(0) is the initial slope of gain transience and it is clear that an increase of PIS acts to reduce G(0) while
the output power of EDFA (Pout) is maintained. That is to say, suppression of transience can be achieved by
alleviation of saturation of EDFA.
The most intuitive method to alleviate saturation is to increase S. Hence, we made prototypes of TS-EDFA
optimized for OPS adopting an EDF with enhanced active area whose diameter were 4.3m [6] and 7.8m [7],
respectively. Figure 2 shows the comparison of two types of commercially available EDFAs (Type A and B) with 1st
and 2nd prototypes of TS-EDFA (Type C and D).
=32%

=32%

=32%

(b1) 100ns/div
=0.037%

(c1) 100ns/div
=0.037%

Type D

Type C

Type B

Type A

(a1) 100ns/div
=0.037%

=32%

(d1) 100ns/div
=0.037%

(c2) 100ns/div
(b2) 100ns/div
Fig. 2 Comparison of EDFAs

(a2) 100ns/div

(d2) 100ns/div

Actually, the 1st prototype was almost same with type B, but reduced the transience by about 30%. On the other
hand, the 2nd prototype showed remarkable improvement. As a reference, 0.037% of traffic density corresponds to
1ms of interval time between packets.
3. Alternative scenario toward burst-mode amplification in WDM environment

tilt

Ga
in

tra
ns

Ga
in

(b)Mid
(b)MidpositionGFF+overpumping
positionGFF+overpumping

(a)ConventionalEDFA

Intensity

Intensity

ien
ce

Tr
aff
ic

Av
era
ge
p

de
ns

ow
e

ity

Previous work on burst-mode amplification mostly


Packet intensity constant
Average power constant
focused on measures to reduce the degree of
transience assuming prefixed burst traffic in the
optical domain. To the best of my knowledge, there is
scarce discussion about the change of traffic density
affecting on optical amplification [8].
As shown in Fig.3, if average power in fiber is
2. Gain-flat
1. Low-transient
kept constant, the intensity of each packet will
burst-mode
burst-mode
change drastically according to traffic density
EDFA
EDFA
Particularly in scarce traffic, large packet intensity
results in severe transience in EDFA [8]. On the other
t
t
hand, adjusting the average power according to traffic
Fig. 3 Two scenarios of burst-mode EDFA
density seems effective to reduce the transience. As a
result of such adjustment, the packet intensity is kept
almost constant. However, in such a scheme, the average power changes enormously and, as a result, it become quite
(c)LOA+EDFAhybrid

Mid-GFF
EDF

31%;

3.4%;

Traffic density:

0.74%

WDM channels

0
-2
-4
1540

1545

1550

1555

Wavelength [nm]

1560

31%;
2

Post-EDF

3.4%;

0.74%

WDM channels

LOA

Traffic density:

Relative gain [dB]

Relative gain [dB]

Traffic density:

Pre-EDF

High power LD

Relative gain [dB]

LD

0
-2
-4
1540

1545

1550

1555

Wavelength [nm]

1560

31%;
2

EDF

LD
3.4%;

0.74%

WDM channels

0
-2
-4
1540

1545

1550

1555

Wavelength [nm]

Fig. 4 comparison of WDM flatness against the change of traffic density


Type (a) conventional EDFA, Type (b) overpumped TS-EDFA with mid-position GFF, Type (c) Hybrid of LOA and TS-EDFA

1560

a889_1.pdf
OThI4.pdf

OSA / OFC/NFOEC 2010

difficult to maintain gain flatness for WDM for broad input power ranges [9]. This is the second issue of burst-mode
EDFA.
In order to measure this second issue, we have proposed two configurations. One was mid-position gain
flattening filter (GFF) with overpumped TS-EDFA [9], and another on was hybrid configuration [10] by using linear
optical amplifier (LOA) [11] and TS-EDFA.
The potential drawbacks of the former configuration is the requirement for a high power laser for pumping and
precise tuning of GFF, and increase of noise figure (NF). The latter configuration is quite simple and potentially
cost-effective to realize burst-mode amplifier for WDM. Fig. 4 shows comparison of WDM flatness against the
change of traffic density.
4. Supplemental combination with optical feedback

Max. measured gain transient (dB)

Initially, the TS-EDFA was specifically


optimized for packet-switched networks
13
Transient
- Burst channels
C-EDFA
where each packet has short (<1s) duration.
12
reduction
- Continuous channels
due to TS11
In order to meet the demand of longer
EDFA
10
duration such as burst-switched or
Trans.
9 Region where no supplementary
-switched
traffic,
we
investigated
Cont. Ch.
control required - TS-EDFA
8
t
supplemental combination of TS-EDFA with
Trans.
7
TS-EDFA
Burst
optical feedback [12]. Fig. 5 shows the
6
t
variation of measured gain transience
tR tR tR
5 No supplementary
control required
C-EDFA+Opt.FB
according to the minimum reconfiguration
4
- C-EDFA
Reconfiguration
3
time (tR) between each switching event. In
time (min)
2
this case we used two wavelengths for
Transient
1
reduction
continuous channel and burst channel.
due to TS0
TS-EDFA+Opt.FB
The results show that the TS-EDFA
EDFA
1
10
100
1000
10000
showed gain transience less than 1dB at tR
Reconfiguration time - tR (s)
up to 100s compared to 10s for
conventional EDFA (C-EDFA) minimizing
Fig. 5 Variation of gain transience
the requirement of any supplementary
control scheme. With optical feedback the TS-EDFA was able to suppress transients over all timescales. However,
the C-EDFA still showed transience up to 3dB for the same level of optical feedback. The combination of TS-EDFA
with optical feedback presents almost ideal characteristic as a burst-mode amplifier for wide dynamic range of
traffic density. That is to say, TS-EDFA can enhance the performance of optical feedback scheme.
3. Conclusion
We gave an overview of a novel transient-suppressed EDFA (TS-EDFA) and relating technologies for burst-mode
amplification. At first, TS-EDFA is the most simple and cost effective solution to be deployed anywhere in an
optical packet-switched network. Secondly, we showed solutions for two scenarios toward burst-mode amplification
in WDM environment. Thirdly, we investigated the combination of TS-EDFA and optical feedback loop which
present almost ideal performance of a burst-mode amplifier.
4. References
[1] A. K. Srivastava, Y. Sun, J. L. Zyskind, J. W. Sulhoff, IEEE Photon. Technol. Lett., 9(3), pp. 386-388, (1997).
[2] C. Tian and S. Kinoshita, IEEE J. Lightwave Technol., 21(8), pp. 1728-1734, (2003).
[3] M. Karsek, A. Bononi, L. A. Rusch, M. Menif, IEEE JLT 18(3), pp. 308-313 (2000).
[4] Y. Sun, J. L. Zyskind, A. K. Srivastava, L. Zhang, Applied optics 38(9), pp. 1682-1685 (1999).
[5] A. A. M. Saleh, R. M. Jopson, J. D. Evankow, J. Aspell, IEEE Photon. Technol. Lett. 2(10), pp. 714-717 (1990).
[6] Y. Awaji, H. Furukawa, N. Wada, P. Chan, R. Man, CLEO2007, JTuA133 (2007).
[7] Y. Awaji, H. Furukawa, N. Wada, E. Kong, P. Chan, and R. Man, OECC2007, PDP2-4.
[8] Y. Awaji, H. Furukawa, N. Wada, E. Kong, P. Chan, R. Man, Computer Networks, vol. 52, Issue 10, pp.2087-2093, July 2008.
[9] Y. Awaji, H. Furukawa, N. Wada, E. Kong, P. Chan, R. Man, Optical and Quantum Electronics, vol. 40, No. 5-6, pp.461-466, Apr 2008.
[10] Y. Awaji, H. Furukawa, N. Wada, OFC2009, OTuH2 (2009).
[11] D. A. Francis, S. P. DiJaili, J. D. Walker, OFC2001, PD13-1, Mar 2001.
[12] B. J. Puttnam, Y. Awaji, N. Wada, ECOC2009, Tu 5.3.2 (2009).

You might also like