Highly improved uplink transmission in bidirectional ... - IEEE Xplore

0 downloads 0 Views 793KB Size Report
central light seeded PONs. By using RZ modulated R-SOA as uplink transmitter, we obtained error- free uplink communications at 10 dB signal-to-crosstalk level.
ECOC 2010, 19-23 September, 2010, Torino, Italy

Mo.1.B.4

Highly improved uplink transmission in bidirectional PONs by using a RZ direct-modulated R-SOA (1)

(1)

(1)

(2)

(1)

L. Banchi , R. Corsini , M. Presi , F. Cavaliere , E. Ciaramella (1) (2)

CEICCP - Scuola Superiore Sant'Anna, via Moruzzi, 1, 56124 – Pisa (ITALY) [email protected] Ericsson, via Moruzzi, 1 – 56124 – Pisa (ITALY)

Abstract We experimentally demonstrate unmatched resilience to carrier and signal backscattering in central light seeded PONs. By using RZ modulated R-SOA as uplink transmitter, we obtained errorfree uplink communications at 10 dB signal-to-crosstalk level. Introduction The availability of potentially low-cost colourless re-modulators such as FP-lasers, Reflective Semiconductor Optical Amplifiers (R-SOA), Reflective Electro-Absorption Modulators (REAMs), can effectively open the way to a massive deployment of WDM-PONs based on Single-feeder loopback Passive Optical Networks (PON) with Centralized Light Seeding (CLS). While this kind of architecture inherently retains several advantages (simple design, easy management, low-cost), it has a fundamental limitation: reflections generated in the feeder fiber produce in-band crosstalk that can severely reduce system performance. Previous literature showed that the in-band crosstalk originated by those reflections can be tolerated above 20 dB [1]. Some sources of reflections (such as the Rayleigh back-scattering into the feeder fiber, or reflections from connectors) are predictable, and thus their impact can be fixed during the system design. However other unpredictable sources of reflections (deriving for example from badsplicing or legacy components) cannot be managed at the design stage. Two main classes of reflections can impair the uplink transmission: carrier backscattering (type-I reflections) and signal back-scattering (type-II reflections), which is directed towards the ONU and is eventually amplified and copropagated together with the upstream [2]. Several techniques based on dithering [2], optimal filtering [3] and line coding [4] can increase the tolerance to reflections allowing to shift the crosstalk tolerance level down to about 18 dB [2-4]. In this paper we demonstrate that by employing R-SOA as ONU transmitter and driving it with RZ pulses, we can further increase the tolerance to the crosstalk, generated both by type-I and type-II reflections. We report error-free operation with limited penalties down 10 dB crosstalk level for type-I reflections and 0 dB crosstalk level for type-II reflections.

978-1-4244-8535-2/10/$26.00 ©2010 IEEE

Operating Principle Because of the gain-phase coupling, a CW light modulated by an R-SOA is accompanied by a phase modulation given by

φ (t ) =

where

α eff

− α eff 2

ln (∆G (t ))

(1)

is the linewidth enhancement factor

parameter. When the R-SOA is driven by a Non Return-to-Zero (NRZ) signal, phase transitions occur only at 0-1 (or 1-0) transitions: a constant symbol sequence (e.g. consecutive 1’s) is not accompanied by any phase change. On the other hand, in a Return-to-Zero (RZ) signal, there are two opposite transitions at every 1’s bit: therefore the signal results to be highly chirped, i.e. with a significantly reduced coherence time if compared to the NRZ signal. This is illustrated in Fig.1, where a comparison of the phase changes induced in the signal modulated by the R-SOA is reported. As can be seen, while consecutive symbols do not generate any phase modulation (chirp) in the NRZ stream, the RZ stream is characterized by a periodic phase variation, which effectively reduces the carrier coherence time and increases the resilience to the crosstalk [2].

Fig. 1: Chirping Effect of the NRZ and RZ modulation.

Mo.1.B.4

ONU

25 km SMF

OLT Polarisation scrambler

PPG Driver

OC-1

CW

50/50

Laser VOA-2 OTF

VOA-1

R-SOA

PC

BERT APD

VOA-3 ODL

VOA-4 OC-2

VARIABLE REFLECTOR

Fig. 2: Experimental setup.

into the R-SOA (type-II reflections). Considering the losses due to VOA-2, the OTF and the 3 dB coupler, the amount of type-II reflections when the variable reflector module is disconnected are estimated between -48 and -52 dB. In this configuration the type-II reflections are therefore negligible (as they reduce the signal-to-crosstalk ratio at around 30 dB). The impact of type-II reflections is evaluated by removing the 25-km fiber spool and connecting the variable reflector module. It emulates a concentrated reflection located very close to the upstream transmitter (few meters). It implements an extremely high coherent reflection point, which is known to be far more detrimental than a distributed Rayleigh reflector [5]. By acting on VOA-3 we can set type-II reflections values between -55 and -20 dB. Both the PC and the variable ODL contained in the reflector module are set to maximise the impact of these reflections. We note that in all the measurements reported in the following, the receiver threshold is adapted to the optimal position for each value of the SCR set.

Sensitivity vs. SCR -22 RZ NRZ

-24 Sensitivity (dBm)

Experiment In order to evaluate the impact of both reflections type-I and reflections type-II with the proposed modulation format, we realized the experimental setup reported in Fig. 2. A laser emitting at λ=1535 nm provides for the CW feeding light. The CW light is sent through a polarization scrambler, driven by a 6 kHz random signal, followed by a variable optical attenuator (VOA-1) that controls the CW feeder power level launched into the fiber. After an optical circulator (OC-1), the CW feeder is launched into a 25 km single-mode fiber spool (6 dB loss). At the end of the fiber, the seeding light reaches the R-SOA after passing through VOA-2, a tunable optical filter (OTF, Gaussian shape, 0.8 nm 3 dB bandwidth and 4 dB insertion loss) and a 3-dB passive splitter. The R-SOA is a commercial device, providing 26 dB small-signal gain and 6 dBm output saturated power at 75 mA bias current. It is driven both by 11 a 1.25 Gb/s NRZ or RZ 2 -1 PRBS sequence. We choose a quite high driving voltage (7 Vp-p) for two main reasons: first, it increases the gain modulation thus chirping significantly the signal; secondly, the high driving voltage allows to compensate for the limited E/O bandwidth (approx. 1.5 GHz) of the R-SOA. In the case of RZ format we further implement an 8B10B line code to maximise the number of transitions. The R-SOA is operated at 20°C and has 1 dB polarization dependent gain. The polarization scrambler allows to average any effect due to the polarization dependency of the R-SOA, emulating an installed fibre. The light remodulated by the R-SOA is both coupled to the fiber and to a variable reflector module (Fig. 2). It is composed by a polarization controller (PC), a variable optical delay line (ODL), VOA-3 and a mirror, realized by means of an optical circulator (OC-2) in a closed-loop configuration. In order to evaluate only the impact of type-I reflections the variable reflector is disconnected. The amount of type-I reflections is varied by acting on the VOA-1, i.e. on the launched power in the feeder fiber. We fix this value at 0 dBm (corresponding to about 34 dB return loss from the fiber). By acting on VOA-2, we set the power delivered to the R-SOA. This changes the gain of the R-SOA and in turn the power received at the upstream receiver. In this case, the signalto-crosstalk ratio (SCR) is measured as the ratio between the upstream received power and the fiber return loss (-34 dB): by setting the R-SOA input power in a range between -18 and -24 dBm, the SCR varies between 24 and 10 dB. Of course, the fiber provides the same amount of reflection from the upstream signal

-26

-28

-30

-32 10

12

14

16

18

20

22

24

Signal-to-Crosstalk Ratio (dB)

Fig. 3: Resilience to Type-I Reflections for different modulation formats: RZ (circles) and NRZ (squares).

Mo.1.B.4

Sensitivity vs. Return Loss

RZ

NRZ

-26 NRZ (-16dBm) NRZ (-24dBm) RZ (-16dBm) RZ (-24dBm)

Sensitivity (dBm)

-27

-16 dBm

-28 -29

Return Loss=-24.22 dB

Return Loss=-32.58 dB

Return Loss=-26.56 dB

Return Loss=-38.06 dB

-30 -31

-24 dBm

-32 -33 -20

-25

-30

-35

-40

-45

-50

-55

Return Loss (dB)

Fig. 4: Resilience to Type-II Reflections for different modulation formats: RZ (circles) and NRZ (squares), and different input power levels.

Results and Discussions We first present the impact of type-I reflections. In Fig. 3 we report the measured upstream -9 receiver sensitivity (measured at BER=10 ) as a function of the SCR for two different modulation formats: RZ and NRZ. In this case the reflector module is disconnected. As it can be seen the NRZ signal tolerates a SCR levels as high as 16 dB, with a power penalty of about 2 dB. Above this value, the sensitivity increases quickly. On the other hand, when using the RZ format we are able to increase dramatically the -9 resilience to crosstalk: a BER of 10 is observed up to 10 dB SCR level, with a power penalty of less than 3 dB. The resilience against type-II reflections is reported in Fig. 4. In this case, the resilience is indicated as a function of the return loss (RL) defined as the ratio between the RSOA output power and the reflected power from the variable reflector. We perform the measurements using two different input power levels at the R-SOA: -16 and -24 dBm. These values correspond respectively to the maximum power that can be delivered to the R-SOA (for 0 dBm launched power from the CO) and the minimum power required to obtain error-free remodulation at the R-SOA. In both cases the RZ format outperforms the NRZ, which cannot tolerate RL level greater than -30 dB. At this value of RL where the NRZ modulation format cannot work, the RZ format still shows a limited penalty in terms of sensitivity (less than 2 dB) for both input power levels. However, we note that in both cases, when the R-SOA is highly saturated (input power set to -16 dBm), the tolerance to type-II reflections increases. It is worth to notice that in the case of RZ format with

Fig. 5: Eye diagrams recorded with type II reflections for RZ (left) and NRZ (right) formats

-24 dBm R-SOA CW input power and a RL value of -26.5 dB, the power of the reflections exactly matches the seeding power (0 dB crosstalk level). Notwithstanding only a limited penalty in terms of sensitivity is observed (less than 3 dB). This surprising behaviour is due mainly to the noise limiting amplification given by the strong saturation of the R-SOA. To better appreciate this behaviour, Fig. 5 shows some eye diagram for RZ and NRZ modulation format recorded at different return loss values, for -16 dBm and -24 dBm input power. Conclusions We demonstrate a technique producing a high resilience to both carrier and signal reflections for upstream transmission in CW seeded PONs. This result is obtained exploiting the chirp properties of directly modulated R-SOAs. By driving the R-SOA with RZ modulation format and high driving voltage, we obtain good performance at so high crosstalk levels that have never been reported before. In particular we obtained 3 dB penalty in terms of sensitivity with SCR values as low as 10 dB for reflections induced by the seeding carrier and less than 2 dB penalty when coupling a reflector module that produces 25 dB of return loss close to the upstream transmitter. This work has been partially supported by Ericsson under a grant. References 1 E. Goldstein et al., IEEE Photon. Technol. Lett. 6, 657 (1994). 2 P. J. Urban et al., J. Lightwave Technol. 22, 4943 (2009). 3 C. Marki, et al., Electr. Lett. 43, 644 (2007). 4 A. Chiuchiarelli, et al., IEEE Photon. Technol. Lett. 22, 85 (2010). 5 Y. J. Lee, et al., Proc. OFC’08, OTuH5 (2008).