A large scale wavelength division multiplexed passive optical network is proposed and experimentally demonstrated. 124 bidirectional optical channels with 10-Gb/s downstream and 1.25-Gb/s upstream transmission are simultaneously distributed by a single 32*32 cyclic AWG. The effect of the extinction ratio and seeding power to BER performance are experimentally investigated. The selection of the subcarrier frequency is also analyzed by simulation.
© 2012 OSA
Wavelength division multiplexed passive optical network (WDM-PON) has been demonstrated as a promising solution for future broadband access network [1,2], due to its features such as large capacity, privacy, format transparent, network security, and per-customer based flexible upgrade. So far, extensive attention has been paid to WDM-PON research and development, including but not limited to, low-cost light sources , protection and fault localization [4,5], high data rate , and long reach [7,8]. Among them, attributes such as higher data rate and longer reach can make a single PON cover a larger service area and thus reduce the total number of the deployed PONs. Therefore, such PONs with large number of subscribers are attractive to telecom service providers.
Today, more and more residents are subscribing to broadband internet access. The rapid increase in the number of new subscribers and the extension of the coverage of the service area may exhaust the wavelength channels of a WDM-PON. To increase the subscriber number, one way is to combine the WDM with time division multiplexing (TDM) technique. However, the TDM technique will reduce the bandwidth for each user and cannot contribute to the system capacity. Another way is to employ more wavelengths. In current WDM-PON implementations, an array waveguide grating (AWG) is used at the remote node (RN) to distribute the wavelength channels for each user. Thus, the total number of wavelength channels assigned is limited by the port count of the AWG. To further increase the wavelength channels, we have proposed to fully utilize all the ports on both sides of the N*N AWG  so that we can nearly double the number of wavelength channels.
Recently the subcarrier modulation technique has been introduced into WDM-PON to generate seeding light for uplink transmission . In this paper, we propose to separate an optical carrier into two subcarriers and each subscarrier can be used by any user for either uplink or downlink transmission. We have enhanced the design of our NxN AWG-based remote node  to enable the newly proposed large scale WDM PON to support 1.24 Tb/s (i.e., 124*10Gb/s) downlink transmission. To the best of our knowledge, the total number of the wavelengths assigned (124) is the largest among all the WDM-PON architectures reported so far.
2. Proposed WDM PON and experimental setup
For a typical WDM-PON remote node, an AWG is used to distribute wavelength channels to each user [11–13]. As a result, the supported user count is limited by the size of the AWG. Figure 1(a) shows a general RN structure. Here all wavelength channels are launched into the AWG from a same port (e.g. I1). And one (N + 1)*(N + 1) AWG can distribute N + 1 wavelength channels. To increase the user count, we propose to fully use all the 2*(N + 1) ports by launching the optical wavelength channels from two ports located at different sides of the AWG, e.g. I1 and I2 as shown in Fig. 1(b). In this case, 2*N wavelength channels can be distributed by the single cyclic AWG and this nearly double the user count. Here, the optical wavelengths input from I1 and I2 can be located in a same waveband or different waveband. In enhanced RN, another wavelength demultiplexer is used to separate optical signals S1 and S2 before launched into AWG. If the input wavelengths are located in different waveband channels, e.g. S1 is in C band channels and S2 is in L band channels, one simple course WDM coupler can be used to separate them before AWG. While using the same waveband can further increase the user count, for example, both input optical signals (S1 and S2) contain 2*N wavelength channels, half in C band and the others in L band. In this case, the total user count can be increased to 4*N. However, the optical separation of S1 and S2 is difficult since they are mixed in the same WDM channel. In this paper, one optical interleaver is used to combine and separate these signals.
The proposed large scale WDM PON architecture is shown in Fig. 2 . Here, 2*N lasers with wavelengths covered two free space ranges (FSRs) of the cyclic AWG are combined and modulated with a clock signal (subcarrier frequency) to generate 4N optical subcarriers. These 4N optical subcarriers are then used as light sources of 4N optical channels. After modulation of data on each optical subcarrier, all 4N optical channels are then multiplexed and transmitted to the RN in one single feeder fibre. Optical amplifier maybe required to compensate the insertion loss of signal modulation in the central office. Please note that the 4N optical lights can also be generated by 4N lasers instead of using 2N lasers and subcarrier modulation to simplify the signal generation scheme. Assuming the subcarrier frequency is fsc, the frequency space between the optical subcarrier pair is 2fsc. One optical interleaver with an FSR of 4fsc, can separate the optical subcarrier pairs for all wavelengths simultaneously. One set of the optical subcarriers is launched into one side of the (N + 1)*(N + 1) AWG (e.g. port I1 in Fig. 1). The other set is launched into the other side (e.g. port I2 in Fig. 1). Since the optical wavelengths cover 2 different FSRs of the AWG, two optical subcarriers with a wavelength spacing of FSRAWG appear in each output port, and thus they can be easily separated using a coarse-WDM coupler. As a result, the 4N optical signals are distributed to end users using only one (N + 1)*(N + 1) AWG and this significantly increases the number of supported end users. Here, we should note that the channel spacing of the WDM channel should be integer times of the subcarrier frequency fsc, and the AWG should have a constant FSR in the range of working wavelengths. For the uplink, one reflective semiconductor optical amplifier (RSOA) is used in each ONU for uplink remoudaltion. To balance the signal performance for both downlink and uplink, some compromise in the extinction ratio (ER) for downlink is introduced [14,15].
Figure 3 shows the experimental setup of the proposed PON architecture. In the central office (CO), 31 DFB lasers with wavelength located from 1534nm to 1558nm are used as C-band light source and one tunable laser is used to simulate the L-band light source. The 32 light sources are then multiplexed by a C/L coarse WDM coupler and they are modulated by a 16.65-GHz sinusoidal signal to generate subcarrier pairs using the optical carrier suppression technique . The frequency separation between each subcarrier pair is 33.3GHz (2fsc), which is much smaller than the 3-dB bandwidth of the AWG, 57GHz. The 32 subcarrier pairs are employed as the light sources of 64 downlink channels. A 10-Gb/s Pseudo Random Bit Sequence (PRBS) with a word length of 223-1 is modulated to all optical channels to simulate downlink data. After transmission through a 25-km standard single mode fiber (SSMF), the optical subcarriers, S0, are then separated into two sets, S1 and S2 in Fig. 4 , by an optical interleaver with an FSR of 66.6GHz. The channel spacing of the DFB lasers is 100GHz, which is three times of fsc, all optical subcarrier pairs can be separated simultaneously as shown in Fig. 4. The separated optical subcarriers, S1 and S2, are then launched into the AWG from port I1 and I2, respectively. One piece of optical fiber is used as a delay to make S1 and S2 non-synchronized and thus to simulate two different PRBS signal. At each of the 62 output ports of the AWG, the two widely spaced optical subcarriers are separated again using a coarse C/L band WDM coupler. Finally all the optical subcarriers are sent to their respective optical network units (ONUs) through a 1-km distribution fiber. At the ONU, half of the downlink optical power is launched into an optical receiver for signal detection. The other half is sent to a reflective semiconductor optical amplifier (RSOA) for uplink signal remodulation. The RSOA here has three functions: (1) compensate the power loss, (2) suppress the power fluctuation of the seeding light due to the gain saturation, and (3) to remodulate the uplink data onto the downlink optical subcarriers. To reduce the crosstalk from downlink, a 3-dB extinction ratio is employed for the downlink signal. The uplink data is 1.25-Gb/s PRBS with a word length of 223-1. In this experiment, the AWG is somewhat temperature dependent, but the wavelength change is not significant in lab environment. For practical implementation, athermal product  can be used to avoid the temperature influence.
3. Experimental results and discussions
Figure 5 shows the optical spectra of the downlink signals. The optical carrier and second order subcarriers have been effectively suppressed (23dB lower than optical subcarriers in our experiment (as shown in Fig. 5(b)).
To evaluate the system performance of the proposed architecture, the bit error rates (BER) and eye diagrams are measured and shown in Fig. 6 . The measured downlink optical subcarriers are the wavelengths shown in Fig. 5(b). All the other optical subcarriers have the similar performance but not shown in this figure. The back-to-back (BTB) receiver sensitivities with a BER of 10−9 are −18.9 dBm for both optical subcarriers. The power penalties after 26-km transmission are 2.0 dB and 2.1 dB for the two optical subcarriers located at shorter wavelength and longer wavelength, respectively. For both subcarriers, a 1.2-dB power penalty is evident and it is due to the fibre dispersion. Another 0.8-dB power penalty is attributed to the crosstalk from the neighboring optical subcarriers. Using a filter to suppress the bandwidth of the optical signal could reduce the crosstalk. The eye diagrams are measured with a 20-GHz APD followed by an oscilloscope. Extinction ratio for BTB signal is about 3 dB. The slight distortion of the eye diagram after the transmission is due to the fiber dispersion.
The uplink BER performance is shown in Fig. 7 . Here, the power of the seeding light is −16dBm. The RSOA is biased at a current of 60mA and modulated by PRBS signal with amplitude of 2Vpp. The BTB receiver sensitivity with a BER of 10−9 is −27.4 dBm. In order to evaluate the crosstalk from the downlink transmission, we measured the uplink BER performance for the case where the downlink modulator is removed and a CW lightwave is used as the seeding light replace of the modulated signal. The result is also shown in Fig. 7 and it is evident that power penalty induced by the crosstalk is about 3.0 dB. The eye diagram (the inset of Fig. 7) also shows the residual downlink data has not been suppressed completely and this results in crosstalk and a deterioration of the uplink BER performance. Increasing the seeding light optical power should further suppress the power fluctuation. Figure 7 also shows that with a 26-km long fiber transmission, the receiver sensitivity is −25.6 dBm. Both fiber dispersion and Rayleigh scattering have contributed to the performance degradation. Further evaluation shows the power penalties due to dispersion and Rayleigh scattering are 1.02dB and 0.78dB, respectively.
Figure 8 shows the measured receiver sensitivities when extinction ratio varies from 1.5dB to 6dB for both downlink and uplink transmission. Apparently, there is an ER trade-off between downstream signal and upstream signal for this wavelength reuse scheme. For downstream signal, receiver sensitivity improves with the increase of the downlink ER. The optical signal with smaller ER has more DC component, which cannot contribute to data detection and results in a worse receiver sensitivity. However, this DC component benefits uplink data modulation. Thus, for upstream signal, the receiver sensitivity improves with the decrease of the downlink ER. In practical systems, the ER can be selected according to the power budget of both downlink and uplink.
Besides the extinction ratio, the seeding power is another important factor which affects the uplink BER performance. As shown in Fig. 9 , the uplink receiver sensitivity improves with the increase of the seeding optical power launched into the RSOA. Here, the ER is 3 dB, the bias current is 40 mA, and the amplitude of the driving single is 2 Vpp. Although larger seeding power can significantly improve the BER performance, the uplink modulation tolerates the seeding power variation very well. Our experimental result shows the uplink receiver sensitivity of −22.9dB can be still maintained when the seeding power is as low as −24dBm.
Figure 10 shows the influence of the subcarrier frequency on the system performance. For a given data rate, larger subcarrier frequency can effectively reduce the crosstalk from neighbor subcarrier channel. However, larger subcarrier frequency has to introduce larger channel spacing. To better understand the effect of the subcarrier frequency, we conducted a simulation with a commercial program, VPI transmission maker. As shown in Fig. 10, the BER performance improves with the increase of the subcarrier frequency, fsc. However, this improvement is only significant when the subcarrier frequency is lower than 15GHz. Further increasing the subcarrier frequency the improvement is not obvious. The reason is that for a given channel spacing, the optical signal with a larger subcarrier frequency suffers a larger distortion due to the filtering characteristic of the AWG. In this simulation, the 3-dB bandwidth and channel spacing of the AWG are 57GHz and 100GHz, respectively. And results show a subcarrier frequency of 15GHz is an optimal choice.
In this paper, a large-scale WDM-PON architecture based on N*N cyclic AWG has been proposed. A 124-channel asymmetrical transmission with 10-Gb/s downlink and 1.25-Gb/s uplink speed has been experimentally demonstrated. Only one 32*32 AWG is required to distribute these 124 bidirectional optical wavelengths channels to the ONUs. BER measurements verified the feasibility of the proposed architecture. The influences of the extinction ratio, seeding power and subcarrier frequency on the BER performance have also been investigated.
References and links
1. J. Yu, O. Akanbi, Y. Luo, Z. Zong, T. Wang, Z. Jia, and G.-K. Chang, “Demonstration of a novel WDM passive optical network architecture with source-free optical network units,” IEEE Photon. Technol. Lett. 19(8), 571–573 (2007). [CrossRef]
2. A. Banerjee, Y. Park, F. Clarke, H. Song, S. Yang, G. Kramer, K. Kim, and B. Mukherjee, “Wavelength-division-multiplexed passive optical network (WDM-PON) technologies for broadband access: a review [Invited],” J. Opt. Netw. 4(11), 737–758 (2005). [CrossRef]
3. S-G. Mun, H-S. Cho, and C-H. Lee, “A cost-effective WDM-PON using a multiple contact Fabry-Perot laser diode,” in proceeding of ECOC2010, (Torino, Italy, 2010), paper Mo.1.B.3.
4. C-K. Chan, L-K. Chen, and C. Lin, “WDM PON for next-generation optical broadband access networks,” in proceeding of OECC2006, (Kaohsiung, Taiwan, 2006) paper 5E2–1-1.
5. J-H. Park, J-S. Baik, and C-H. Lee, “Fault-localization in WDM-PONs,” in the proceeding of OFC2006, 2006, paper JThB79.
6. Z. Xu, Y. J. Wen, W.-D. Zhong, C.-J. Chae, X. F. Cheng, Y. Wang, C. Lu, and J. Shankar, “High-speed WDM-PON using CW injection-locked Fabry-Pérot laser diodes,” Opt. Express 15(6), 2953–2962 (2007). [CrossRef] [PubMed]
7. S. P. Jung, Y. Takushima, and Y.C. Chung, “Generation of 5-Gps QPSK signal using directly modulated RSOA for 100-km coherent WDM-PON” in proceeding of OFC2011, (Los Angeles, California, 2011), paper OTuB3.
8. L. Ana, S. Aleksic, J.A. Lazaro, G.M. Tosi Beleffi, F. Bonada, J. Prat, and A.L.J. Texeira, “Influence of broadcast traffic on energy efficiency of long-reach SARDANA access network, ” in proceeding of OFC2011, (Los Angeles, California, 2011), paper OThB5.
9. Z. Xu, X. Cheng, Y-K. Yeo, L. Zhou, X. Shao, “60-channel bidirectional WDM-PON using a single 32*32 AWGR for 120 wavelengths distribution,” in Proceeding of OFC2011, (Los Angeles, California, 2011), paper JWA65.
10. S. Jang, C-S. Lee, D-M. Seol, E-S. Jung, and B. W. Kim, “A bidirectional RSOA based WDM-PON utilizing a SCM signal for down-link and a baseband signal for up-link,” in proceeding of OFC2007 (Anaheim, California, 2007), Paper JThA78.
11. F. Ponzini, F. Cavaliere, G. Berrettini, M. Presi, E. Ciaramella, N. Calabretta, and A. Bogoni, “Evolution scenario toward WDM-PON [Invited],” J. Opt. Commun. Netw. 1(4), C25–C34 (2009). [CrossRef]
12. J. Ingenhoff, “Athermal AWG devices for WDM-PON architectures,” in the proceeding of LEOS 2006, 26–27 (2006).
13. K. Y. Cho, Y. J. Lee, H. Y. Choi, A. Murakami, A. Agata, Y. Takushima, and Y. C. Chung, “Effects of reflection in RSOA-based WDM PON utilizing remodulation technique,” J. Lightwave Technol. 27(10), 1286–1295 (2009). [CrossRef]
15. H. Takesue and T. Sugie, “Wavelength channel data rewrite using saturated SOA modulator for WDM networks with centralized light sources,” J. Lightwave Technol. 21(11), 2546–2556 (2003). [CrossRef]
16. C. Bock, J. Prat, and S. D. Walker, “Hybrid WDM/TDM PON using the AWG FSR and featuring centralized light generation and dynamic bandwidth allocation,” J. Lightwave Technol. 23(12), 3981–3988 (2005). [CrossRef]