W2A.66.pdf OFC 2016 © OSA 2016

4x10 Gb/s Coherent WDM-PON System over 110 km Single Mode Fibre and with 55 dB ODN Power Budget

M. Artiglia(1), F. Bottoni(1), R. Corsini(1), M. Presi(1), M. Rannello(1), M. Valvo(2), E. Ciaramella(1)

(1) Scuola Superiore Sant'Anna, V. G. Moruzzi, 1, Pisa, Italy (2) Telecom Italia Lab, Via G. Reiss Romoli, 274, Torino, Italy

e.ciaramella@sssup.it

Abstract: Simplified coherent detection of directly-modulated DFB transmitters allows for 55 dB ODN power budget in downstream of a 40 Gb/s DML-based WDM-PON with 110 km reach. The solution can enhance NGPON2, addressing long reach and high power budget links (>50 dB). OCIS codes: (060.2360) Fiber Optics Links and Subsystems; (060.1660) Coherent Communications;

1. Introduction The latest ITU-T Passive Optical Network (PON) standard that implements the NG-PON2 concept (ITU-T G.989.x series of Recommendations) includes the Time and Wavelength Division Multiplexed PON (TWDM-PON)[1] that stacks 4 pairs of wavelengths to achieve an aggregate capacity of 40 Gb/s downstream and 10 Gb/s upstream with a typical reach and ODN (Optical Distribution Network) power budget of up to 40 km and 35 dB, respectively.

However, soon it will be important to increase the PON optical power budget at reduced costs and ensuring backward system compatibility. Indeed, increasing the optical power budget will support a higher number of ONUs per ODN and will allow to deploy PON of longer reach in favour of a higher degree of infrastructure sharing and node consolidation. Network consolidation is addressed within PON ITU-T standards through the introduction of Reach-Extenders (ITU-T G.984.6, G.987.4), i.e. devices able to guarantee PON operation over distance up to 60 km. However, as pointed out by the ongoing EU FP7 Project DISCUS [2], much higher reach (up to 100 km) is required to better exploit the advantages of network consolidation (Long Reach PON, LR-PON). Higher loss budgets might be achieved by introducing coherent detection and digital signal processing (DSP) in the access segment [3][4]. Yet, the current coherent receiver technology, based on high-speed A/D converters and advanced DSP and expensive local oscillator (LO), is hardly suitable for access applications, where receivers are expected to be low-cost, simple and robust [5]. Recently, within the COCONUT EU FP7 project, we demonstrated a 10 Gb/s Directly Modulated Laser (DML)-based LR-PON, where an effect similar to chirp management [6] is obtained by using a simplified coherent receiver with proper post-detection filtering [7].

In this paper, we show that the approach of [7] can be extended to a WDM scenario (4's). We chose a configuration that can be seen as an extension of the TWDM-PON concept since it can support both higher sensitivity at the receiver and higher launch power allowed by the DMLs. With this approach we successfully demonstrated a 40 Gb/s downstream bit-rate, with 55 dB of optical power budget over an optical distribution network (ODN) consisting of 110 km of standard Single Mode Fibre (SMF) and capable of supporting a total split ratio up to 1:256. Thanks to the specific approach, no optical dispersion compensation unit is needed.

2. Experiment and Results The proposed WDM-PON architecture is shown in Fig. 1. It emulates the downstream path of a long-reach, high ODN loss (>50 dB) PON. At the OLT side, we used four 10 Gbit/s directly modulated DFB lasers (DMLs) as downstream transmitters. Laser emission frequencies were 100 GHz spaced, in a range from 193.4 to 193.7 THz. Typical laser line-width was around 10 MHz. Their output signals were multiplexed by an arrayed waveguide grating (AWG) to achieve the 40 Gb/s aggregate rate.

Fig. 1: Experimental Setup. PPG: Programmable Pattern Generator; AWG: wavelength multiplexer; EDFA: Erbium Doped Fibre Amplifier; EFs: Electrical Filters; RTO: real time oscilloscope; CDR: clock and data recovery; BER: Bit Errror Ratio evaluation

W2A.66.pdf OFC 2016 © OSA 2016

In each transmitter unit, the DMLs were driven by 10 Gb/s pseudorandom bit sequence (PRBS) data with a word length of 27-1. Neighbour channels were uncorrelated by adding patch-cords of different lengths before the AWG. Then, the signals were amplified up to a total power of 23 dBm by an erbium-doped fibre amplifier (EDFA) booster. The ODN consisted of spools of SMF (G.652, D=16ps/nm/km; =0.2dB/km) of various lengths combined to obtain a fibre line of 110 km, and a Variable Optical Attenuator (VOA) simulating the loss of passive splitters.

At the ONU side, a phase-diverse coherent receiver based on a 3x3 symmetrical fused-fibre optical coupler (120° hybrid) was used [8]. No optical filter is required neither any special source (as LO) in this receiver: a set of three common DFBs (DFB-LO, average linewidth ~10 MHz) were used as LOs to select the four downstream channels. The DFB-LOs were operated free-running, i.e. without any frequency control system (only controlled by usual thermal and intensity controllers). The signal and the respective LO were mixed by the 3x3 symmetrical coupler and then received by three photodiodes (PD) each having a responsivity of 0.6 A/W, 15 GHz bandwidth and an integrated trans-impedance amplifier. In this specific implementation, the polarization states of the LO and the received signal were mutually aligned by a manual polarization controller (PC). The three output photocurrents were then acquired by a real-time oscilloscope (RTO, 13 GHz analogue bandwidth, 40 GSa/s) and processed off line as described below. In the usual implementation of the receiver, ASK demodulation simply requires squaring and summing the three photocurrents [8][9]. Here, however, a key element, is added which is emulating in the electrical domain the well-known CM approach [6]. In the CM approach, an OSR (Optical Spectrum Reshaper) is required, which is usually realized by an optical filter embedded in the transmitter. Here a similar CM effect is very simply obtained by matching the modulation depth of the DMLs (yielding about 5 GHZ dynamic chirp) with electrical low-pass filters of suitable slope in each of the three photodiode current paths (see EFs in Fig.1) [7]. In this proof-of-concept experiment, the EFs were implemented by using a built-in digital low pass filter function (3 GHz bandwidth, Gaussian shape, and linear phase) in the RTO before envelope reconstruction, but they might be simply realized by analogue electronics.

Fig. 2: Optical Spectra in back-to-back (dashed red) and after

transmission (continuous). RBW=0.1 nm.

Fig. 3: Eye diagrams of the four 10 Gbit/s channels after 110 km transmission (17 dBm/channel).

Similarly to the CM case, the signal-LO detuning was set so that the signal spectrum falls on the right EF slope.

The ASK signal envelope was finally recovered by first squaring the three filtered signals and then summing them as shown in Fig. 1. The bit error ratio (BER) was also computed off-line comparing bit-by-bit the received sequence against the transmitted sequence, by means of a software routine run by the RTO processor.

Fig. 2 reports the optical spectra taken at the input of the link at 17 dBm/ch input power (dashed red curve) and the corresponding output (continuous blue curve). These data were taken for maximal polarization alignment: we noticed some Four Wave Mixing (FWM), but it was around -40 dB below the channels, which is expectedly well tolerated. Indeed, in the same conditions, we took the four eye diagrams, at the receiver end: all of them are well open and no significant distortion, either due to dispersion nor FWM, can be seen, when compared to the back-to-back condition. We outline that using chirped DMLs allowed us to reach the above high launch power levels without adopting countermeasures for suppressing Stimulated Brillouin Scattering. This is a intrinsic key strength of our system, enabling to enhance the overall power budget.

In Fig. 4, we then report the BER curves taken for the four channels. Fig. 4a shows the curves in back-to-back, whilst Fig. 4b reports the curves taken after propagation, with the maximum input power (17 dBm/channel, higher levels were not possible with the used EDFA). As we see, around 1÷2 dB penalty is observed at FEC level

W2A.66.pdf OFC 2016 © OSA 2016

(1.4x10-3). Similar curves were taken also for lower launch power values, but they are not directly shown here for the sake of clarity. Rather, based on Figs. 4a and 4b we could draw the main result of this paper, i.e., the average ODN loss curve for each channel, which is shown in Fig 5. As we see, even reducing the power/channel at around 11 dBm, the system could still support almost 50 dB ODN loss, while up to 55 dB ODN loss can be achieved by increasing the channel launch power.

2

3

4

5

6

789

10-42 -40 -38 -36 -34 -32 -30 -28

193.400 THz193.500 THz193.600 THz193.700 THz

-log

(BER

)

Power (dBm) a)

2

3

4

5

6

789

10-42 -40 -38 -36 -34 -32 -30 -28

193.400 THz193.500 THz193.600 THz193.700 THz

-log

(BER

)Power (dBm) (b)

Fig. 4: BER curves for the four WDM channels:(a) back-to-back; (b) after 110 km propagation with 17 dBm launch power per channel.

45

50

55

60

10 11 12 13 14 15 16 17 18

ODN

Los

s (d

B)

Launch Power /ch (dBm) Fig. 5: Maximum supported ODN loss as a function of input power/channel.

3. Conclusions

We showed the feasibility of a WDM-PON system solution capable to deliver 4x10 Gbit/s downstream capacity over a PON infrastructure that supports up to 55 dB ODN loss and 110 km reach. The solution does not require a reach extender nor dispersion compensation. The system combines simplified coherent detection at the receiver side (with common DFBs as LOs) and simple Directly Modulated Lasers at the transmitter side. Signal demodulation at the receiver can be performed in principle by very simple analogue electrical processing, which can help to keep implementation costs low. Furthermore, the receiver is suitable for photonic integration. These results can open the way to implement future TWDM-PON solutions where higher performance is achieved in a cost-effective way.

This work has been partly supported by the EU FP-7 project COCONUT (GA 318515) and PRIN ROAD-NGN.

References [1] L. Yuanqiu et al., "Time- and Wavelength-Division Multiplexed Passive Optical Network (TWDM-PON) for Next-Generation PON Stage 2

(NG-PON2)," J. Light. Techn.vol.31, no.4, pp.587-593, Feb.15, 2013

[2] M. Ruffini, et al., "DISCUS: an end-to-end solution for ubiquitous broadband optical access" IEEE Comm. Magazine, , vol. 52, no. 2, pp. S24-S32, February 2014

[3] D. Lavery, et al., "Digital coherent receivers for long-reach optical access networks," J. Lightw. Techn., vol. 31, no.4, pp. 609-620, 2013.

[4] S. Smolorz, et al., "Demonstration of a coherent UDWDM-PON with real-time processing," Proc. OFC/NFOEC 2011, Paper PDPD4, 2011

[5] J. Prat, et al., "Towards ultra-dense wavelength-to-the-user: The approach of the COCONUT project," Proc. ICTON 2013, 23-27 June 2013

[6] D. Mahgerefteh, et al., “"Chirp Managed Laser and Ap-plications" J. Sel. Top. Quant. Elect., vol. 5, pp. 1126-1139, 2010

[7] R. Corsini, et al., “"Simple and low cost 10 Gb/s coherent transmission for long reach PON," Proc. of ECOC 2014, Paper P.7.3, Sept. 2014

[8] L. Kazovsky, et al., “"ASK multiport optical homodyne receivers," J. Lightw. Techn., vol. JLT5, no. 6, pp.770-791, Jun 1987

[9] M. Artiglia, et al. “Coherent Systems for Low-Cost 10 Gbit/s Optical Access Networks,” J. Ligh. Techn., 33, 15, pp. 3338-3344, Aug. 2015.