A simple polarization division multiplexed (PDM) multiband direct-detection optical orthogonal frequency division multiplexing (DDO-OFDM) long reach passive optical network (LR-PON) with net data rate over 210 Gb/s on single wavelength channel is proposed and experimentally demonstrated with self-polarization diversity technique. The proposed self-polarization diversity function is realized at a powered remote node with all passive components to achieve cost-effectiveness and simultaneously double both the channel capacity and subscriber numbers. Meanwhile, this architecture retains the simplicity of direct-detection single receiver-end without any hardware or software modification at the optical network units. The measured power penalty of the proposed PDM multiband DDO-OFDM LR-PON is 0.8 dB over 100km transmission with respect to that of the ordinary single polarization scheme at a specified forward error correction threshold.
© 2015 Optical Society of America
As the fast soaring of modern communication applications, the extensions in both the transmission capacity and distance of the passive optical network (PON) are inevitable . A popular way to increase the coverage of single central office (CO) is to employ an optical amplifier as a span extender in the remote node (RN). Thus, in such large network coverage, increasing the subscriber numbers, but also simultaneously providing an acceptable data rate to meet the modern communication requirement, i.e. around 1 Gb/s per user , is becoming an important system design issue.
To enhance system capacity, orthogonal frequency division multiplexing (OFDM) with advanced modulation format provides high spectral efficiency (SE) and high bandwidth allocation flexibility for wireless and wired communications . Moreover, the long symbol period property of OFDM signals enhances the signal tolerance to chromatic dispersion (CD) and polarization mode dispersion in assess distances which enables OFDM as a very promising candidate in future PON. As presented in our previous work , multiband direct-detection optical OFDM (DDO-OFDM) can provide 150 Gb/s data rate and uniform quality of service (QoS) for all signal bands in a 50 GHz wavelength division multiplexing (WDM) channel.
Polarization division multiplexing (PDM) technique, which simultaneously transmits two independent data streams on the same wavelength, can double the SE and is also compatible with OFDM systems. Therefore, by applying PDM technique, a high speed data stream can be decomposed into two low speed data which can effectively relieve the bandwidth constrain of both the electric and optical devices. However, considering that the implementation cost at the subscriber ends is usually preferred to be simple and low-cost, the necessity of dynamic polarization tracking  and/or digital signal processing (DSP) enhanced receiver  in conventional PDM system always restricts its feasibility in PON. A self-polarization diversity technique is first introduced for polarization mode dispersion issue in 2008 by Xie . In the same year, this technique was applied to transmit two independent data stream in our previous work . Without adaptive polarization control, the PDM signal has been successfully demodulated with high uniform performance at different polarization angles with respect to the fiber fast axis. However, the drawback is that a complex multiple-input and multiple-output (MIMO) equalizer is necessary in the receiver, which induces more latency.
In this work, we experimentally demonstrate a long reach (LR) self-polarization diversity multiband DDO-OFDM downstream with aggregated raw data rate up to 270 Gb/s over 100 km transmission. Compared with our previous work , PDM de-multiplexing structure is shifted from receiver to RN to reduce the design complexity and preserve the ordinary single polarization (single-pol) DDO-OFDM demodulation processing in the receiver of optical network units (ONUs). Also, to retain low cost and low complexity, the proposed self-polarization diversity function in the RN is implemented with only passive components. In contrast to conventional PDM DDO-OFDM system , the feasibility and cost-effectiveness of this architecture is enhanced. Consequently, the channel capacity and the number of subscribers of the proposed self-polarization diversity scheme are doubled from that of an ordinary single-pol scheme. The experimental result shows that we can provide about net 1 Gb/s data rate per user for over 210 ONUs. Moreover, the bit-error rate (BER) performance of the multiband DDO-OFDM signals on both polarizations is almost identical. A negligible power penalty is observed over 100 km LR transmission.
2. Operation principles
The key concept of this self-polarization diversity is originated from Martinelli’s research , where the Faraday rotator mirror (FRM) was employed as a “polarization orthoconjugator” and presented with Jones matrix formalism. The corresponding matrix of FRM is presented as follow, which can be understood as a linear combination of a mirror and a Faraday rotator with rotation power equal to exactly 45°.
The matrix for back propagation of non-reciprocal Faraday rotator is just changing the sign of the rotation angle. Thus, one can note that, in general, this linear operator transforms any input polarization state into an orthogonally output polarization state. This property is very useful and also has lots of applications in optical interferometer system, sensing system [10, 11] or quantum cryptography [12, 13].
The conceptual diagram of the proposed self-polarization diversity PDM DDO-OFDM structure is depicted in Fig. 1. In transmitter, two OFDM signals are individually modulated by I/Q modulators (IQMs) and orthogonally combined with a polarization beam combiner (PBC). After fiber transmission, the polarization states of the PDM signals are randomly rotated, which is generally an elliptical polarization state. However, even the state of polarization (SOP) is unspecified elliptical polarization, the orthogonality between two signals are preserved. Thus, traditional PDM OFDM receiver in ONUs can usually employ a polarization tracking mechanism and/or DSP algorithm to de-multiplex the PDM signals, which demands additional hardware and/or software and increases the receiver cost.
In this proposed system, we de-multiplex the PDM signals at the RN. For clear explanation, hereinafter, we simply symbolized the two orthogonally polarized signals as x-polarized (x-pol) and y-polarized (y-pol) light, as shown in Fig. 1, respectively. In the transmitter design, only one optical carrier is employed and aligned in the x-polarization by a polarization controller before launched into fiber. Thus, two orthogonally polarized signals are carried by a single polarized carrier. In the designed remote node, the upper path is a direct pass. Thus, signals directly pass through the RN and enter the upper group of ONUs, circled in red in Fig. 1, for receiving x-pol signals. When the dual polarized OFDM signals and single polarized optical carrier detected by a single-end photo-detector (PD) in upper ONU groups, the corresponding square law detection can be expressed as Eq. (2):
where Cx, Sx and Sy are optical carrier in x-pol, OFDM signal in x-pol and OFDM signal in y-pol, respectively, and ∣.∣2 denotes the square-law detection operator. Due to the orthogonality of x-pol and y-pol, the cross beating terms with orthogonal SOP will be zero. The rest terms in Eq. (2) sequentially represent DC, signal-to-signal beating interference (SSBI)  of x-pol and y-pol signals, and the desired x-pol signal. The DC term can be easily removed by a DC block or DSP circuits. However, the SSBI will occupy the same spectral range as that of the baseband OFDM signal and consequently degrade the received signal quality. Therefore, a guard band with bandwidth equal to the total OFDM signal is needed in conventional DDO-OFDM system. With the protection from the guard band, the signal in the desired polarization (x-pol in this case) can be demodulated but with doubled SSBI power compared with that of the ordinary single-pol scheme. Alternatively, the signal in y-pol can also be detected with optical carrier in y-pol by a single-end PD in the same manner.
Based on the aforementioned description, the proposed PDM de-multiplexing mechanism for y-pol is constructed in the lower path of the RN. After an optical 3 dB coupler and an optical circulator, a 25/50 inter-leaver connecting is applied to separate the optical carrier and dual polarized signals. The separated optical carrier is sent into a Faraday rotator mirror (FRM), thus based on Eq. (1), the input SOP of the optical carrier is rotated to its orthogonal SOP (in this case, is y-pol) by FRM. After that, the reflected carrier in y-pol then merges with the dual polarization signals reflected by a normal fiber mirror. The carrier in y-pol and the dual-polarized signals are then delivered into the lower group of ONUs, which are circled in blue, for receiving y-pol signal. An alternative approach, orthogonally converting the SOPs of the dual polarized signals rather than optical carrier, can be achieved with the same design principle by swapping the corresponding FRM and fiber mirror.
3. Experimental setups
Figure 2 depicts the experimental setup of the proposed self-polarization diversity PDM multiband DDO-OFDM in LR-PON. An external cavity laser at 1552.45 nm with 100 kHz linewidth is applied at the transmitter. The laser is first evenly split into two paths: one is applied for multiband OFDM modulation, and the other is prepared for optical carrier insertion mechanism. In the upper path, a frequency comb generator (FCG), which is implemented by cascading two Mach-Zehnder modulators (MZM), is applied to generate 6 optical comb tones for OFDM modulation in the next stage.
The generation of the 6 tones is described as follows: the first MZM is driven by a 14.5-GHz sinusoidal clock source and operated at optical carrier suppression mode by biasing the MZM at its null point; the second MZM is linearly driven by a 5.5-GHz sinusoidal clock source. Thus the generated 6 optical combs are shown as Fig. 2(a). The OFDM signals are offline encoded by MATLABTM and then digital-to-analog converted by a 12 GSa/s arbitrary waveform generator. We designate 180 subcarriers out of 512 FFT size for baseband OFDM signal, which occupies about 4.22 GHz bandwidth. Each OFDM frame is combined with 20 training symbols and 180 data symbols. A 1/16 cyclic prefix (CP) is also inserted for chromatic dispersion compensation and receiving synchronization. The electric OFDM signals are launched into an IQM to load the desired OFDM signals onto 6 optical combs, as shown in Fig. 2(b). Since the chromatic dispersion induced phase mismatch between the optical carrier and OFDM signals is proportional to their frequency differentials, the inner band is considered as a stronger channel than other signal bands because it has better coherence with the optical carrier due to the shortest spectral distance away from the carrier. Thus, an adaptive modulation is employed with 64-, 32- and 32-QAM size for the 3 RF bands of the inner, middle and outer bands located at 9-, 14.5- and 20-GHz, respectively. Then, an aggregated 135 Gb/s raw data stream on single polarization is sent into a polarization multiplexing scheme to double the date capacity to 270 Gb/s. After removing 7% forward error correction (FEC) overhead , training symbols and CP, the net data rate is about 211.8 Gb/s in this PDM multi-band DDO-OFDM system. In the lower path, due to carrier insertion architecture, the amplitude and phase mismatch between optical signals and carrier should be carefully tuned to obtain the best system performance among the 12 OFDM bands on both polarizations. Before combined with the multiband signals, the SOP of the optical carrier is aligned to x-pol to guarantee only signal in x-pol can be down converted and demodulated in the upper group of ONUs. The optical spectrum at the transmitter output is demonstrated in Fig. 2(c).
After 100 km SMF transmission, an EDFA with small signal gain of 28 dB is applied as a span extender to compensate for the transmission loss. As mentioned in the previous section, carrier in different SOP with dual polarization signals are delivered into designated ONUs. In the lower path of RN, the SOP of extracted optical carrier is orthogonally converted and then reflected by an FRM. An optical delay line, which is exactly half of the de-correlating fiber in the polarization multiplexing scheme, is embedded between FRM and optical inter-leaver to compensate for the additional phase mismatch between optical carrier and signal in y-pol. After RN, a variable optical attenuator is applied to manipulate the splitting ratios.
At the receiver frontend, a single-side band (SSB) optical filter is employed to filter out the desired signal bands, which is the outmost band after optical filtering, as shown in Fig. 2(d). The applied SSB filter is manually wavelength and bandwidth tunable with about 1.2 Gaussian order. The filter 3-dB bandwidth settings are 11.1-, 16.6- and 22.1-GHz for inner, middle and outer bands, while filter central wavelength offset to optical carrier are 5.55-, 8.3- and 11.05-GHz, respectively. The detailed double side band receiving mechanism is specified in our previous work , where we can carry OFDM signals on both sides of the carrier without incurring RF fading problem and half reduce the maximum receiver bandwidth requirement by this optical SSB filtering mechanism. After optical filtering, an ordinary DDO-OFDM demodulation scheme is applied. The received OFDM signals are O/E down-converted by a single-end PD and then analog-to-digital converted (ADC) by a real time scope at a sampling rate of 80 GSa/s. It is worth to note that the ADC sampling rate can be reduced by employing physical RF down-conversion mechanisms to demodulate the I and Q signals . The sampled electric signals are then decoded by the offline DSP procedure following a reverse procedure of the encoding process.
4. Experimental results and discussions
In this section, we first verify the polarization orthogonality by electric spectra shown in Fig. 3. When carrier in x-pol and signal in y-pol are employed and detected by PD, only SSBI can be observed from the spectrum, as shown in Fig. 3(a). Figure 3(b) exhibits the detected spectrum of orthogonally-converted carrier in y-pol and three signal bands in y-pol, where the outmost band after optical SSB filter can be free from SSBI contamination, which has been explained and demonstrated in our double sideband DDO-OFDM demodulation principle .
As we mentioned in the principle section, in lower group ONUs, converted carrier in y-pol and the dual polarization signals are O/E converted by single-end PD, and only signal in y-pol can be demodulated. However, since the self-beating of the signal in x-pol is also presented, the SSBI noise power, coming from both x-pol and y-pol signals, is doubled and the signal to noise ratio is thus degraded, as shown in Fig. 3(c).
In system performance evaluation, for a comprehensive comparison, a single-pol scheme with ordinary DDO-OFDM demodulation receiver is also conducted in this experiment. Our previous work  has proved that with the help of SSB optical filter the performance difference is negligible between signals in upper and lower sidebands. Therefore, we only exhibit the BER curves of the upper 3 signal bands for back-to-back (BTB) in Fig. 4. The black lines with open markers are the single polarization results, while the red and blue lines with filled markers are the x-pol and y-pol results, respectively, in PDM scheme. The averaged receiving sensitivity at BER = 3.8x10−3, 7% FEC overhead threshold, is −27.2 dBm for single-pol scheme and −26.5 dBm for dual polarized PDM scheme with self-polarization diversity technique. However, due to the slow spectral roll-off of the inter-leaver employed in RN, part of the inner band signal in x-pol accompanied optical carrier passes through the inter-leaver. Thus, the performance of the inner band in y-pol is degraded by this residual x-pol signal pollution, as shown in Fig. 3(b), and the relatively receiving sensitivity is −24.7 dBm at the threshold. Without considering the y-pol inner band, the average power penalty between single-pol and self-polarization diversity scheme in both polarization is 0.7 dB, which is caused by the unwanted beating interference from the other polarization after PD and is presented as SSBI, as shown in Fig. 3(c). However, it is worth noting that the performance of x-pol and y-pol are almost the same, which attributes to the employment of all passive components in this proposed mechanism. Moreover, it is reasonable that if we can apply a fast spectral roll-off filter or increase the bandwidth gap of the guard band, the performance of the inner band in y-pol will be close to the rest bands and then this PDM DDO-OFDM system can provide equal quality of service for all users.
The averaged receiving sensitivity after 100 km transmission at FEC threshold is −26.6 dBm for single-pol scheme, −25.8 dBm for self-polarization diversity PDM scheme and −23 dBm for inner band in y-pol, as shown in Fig. 5. Again, without considering the inner band, the power penalty for self-polarization diversity scheme is 0.8 dB after 100 km transmission, which is almost the same as that of BTB scheme. This guarantees the feasibility of the proposed self-polarization diversity mechanism even after 100 km transmission. However, the inner band in y-pol exhibits a 2.8 dB power penalty with respect to the rest 5 bands over 100 km transmission, which is increased by 1 dB from BTB scheme. This can be understood as the aforementioned slow spectral roll-off effect of inter-leaver but also the higher level QAM signal is more sensitive to the amplified spontaneous emission noise induced by the span extender. The corresponding constellations are also inserted in Fig. 5.
Figure 6 shows the splitting ratio test of this self-polarization diversity PDM LR-PON scheme. The output power after the span extender is 7 dBm, and therefore the maximum received power in x-pol is about 4 dBm for the case of one subscriber. However, the power splitter always evenly split all signal bands. Thus, without considering per user data requirement, the maximum splitting ratio under the FEC threshold is 1:512 for x-pol scheme which is limited by 64-QAM performance in the inner band. As for the y-pol scheme, on the other hand, it needs additional optical circulator, inter-leaver, and mirrors in PDM de-multiplexing mechanism, the total power loss of these devices is about 5.8 dB. Therefore, the splitting ratio, which is about 4 times less than that of x-pol signals, is 1:128. However, the unbalanced power splitting can be evened by replacing the 3 dB optical coupler in RN with a 1:5 splitter. As we mentioned in the introduction, the future data requirement of each subscriber is about 1 Gb/s. In this experimental demonstration, due to the training symbols, 7% FEC overhead and 1/16 CP, the net data rate is about 105.9 Gb/s per polarization which corresponds to an effective dual polarization SE of 4.79 b/s/Hz. Under this circumstance, the proposed system can serve up to 105 subscribers in single-pol and 210 subscribers in PDM self-polarization diversity scheme. We then set the zero point of power margin at 128 splits, as shown in the upper axis in Fig. 6. There is almost no power margin for inner band in y-pol, while about 7.5 dB power margin remains in x-pol. Again, this unbalance power loss can be compensated by using unbalanced power splitter in RN. Consequently, even y-pol suffer more power loss in this demonstration, self-polarization diversity scheme can still double the number of subscribers from the single-pol OFDM system.
We have successfully demonstrated an aggregated 270 Gb/s self-polarization diversity PDM multiband DDO-OFDM LR-PON over 100 km transmission. The DDO-OFDM receivers in ONUs require no automatic polarization tracking hardware and/or additional MIMO DSP software modification. The channel capacity and number of subscribers can be doubled from that of the single-pol scheme. Moreover, the passive components implemented in RN retain a low operational expenditure nature for system provider and therefore enhances the feasibility of the proposed architecture. Finally, this proposed self-polarization diversity PDM DDO-OFDM architecture is a low cost but with high spectral efficiency solution for future PON systems.
References and links
2. I. P. Kaminow, T. Li, and A. E. Willner, Optical Fiber Telecommunications VIB, 6th ed. (Academic, 2013), Chap. 22.
4. J.-H. Yan, Y.-W. Chen, K.-H. Shen, and K.-M. Feng, “An experimental demonstration for carrier reused bidirectional PON system with adaptive modulation DDO-OFDM downstream and QPSK upstream signals,” Opt. Express 21(23), 28154–28166 (2013). [CrossRef] [PubMed]
6. C.-C. Wei, C.-T. Lin, C.-Y. Wang, and F.-M. Wu, “A novel polarization division multiplexed OFDM system with a direct-detection BLAST-aided receiver,” in Proceedings of Optical Fiber Communication Conference (2013), paper. JTh2A.49. [CrossRef]
7. C. Xie, “PMD Insensitive Direct-Detection Optical OFDM Systems Using Self-Polarization Diversity,” in Proceedings of Optical Fiber Communication Conference (2008), paper. OMM2. [CrossRef]
8. W.-R. Peng, K.-M. Feng, and A. E. Willner, “Direct-Detected Polarization Division Multiplexed OFDM Systems with Self-Polarization Diversity,” in Proceedings of Lasers and Electro-Optics Society (2008), paper. MH3.
9. M. Martinelli, “A universal compensator for polarization changes induced by birefringence on a retracing beam,” Opt. Commun. 72(6), 341–344 (1989). [CrossRef]
10. H. Zhang, Y. Dong, J. Leeson, L. Chen, and X. Bao, “High sensitivity optical fiber current sensor based on polarization diversity and a Faraday rotation mirror cavity,” Appl. Opt. 50(6), 924–929 (2011). [CrossRef] [PubMed]
11. P. Drexler and P. Fiala, “Utilization of Faraday mirror in fiber optic current sensors,” Radioengineering 17(4), 101–107 (2008).
12. M. Bourennane, F. Gibson, A. Karlsson, A. Hening, P. Jonsson, T. Tsegaye, D. Ljunggren, and E. Sundberg, “Experiments on long wavelength (1550 nm) “plug and play” quantum cryptography systems,” Opt. Express 4(10), 383–387 (1999). [CrossRef] [PubMed]
13. D. S. Bethune and W. P. Risk, “An Autocompensating Fiber-Optic Quantum Cryptography System Based on Polarization Splitting of Light,” IEEE J. Quantum Electron. 36(3), 340–347 (2000). [CrossRef]
15. ITU-T Recommendation G.975.1, Appendix I.9, 2004.
16. W.-R. Peng, I. Morita, H. Takahashi, and T. Tsuritani, “Transmission of high-speed (> 100 Gb/s) direct-detection optical OFDM superchannel,” J. Lightwave Technol. 30(12), 2025–2034 (2012). [CrossRef]