Open Access
Add to BibSonomyAbstract
A novel WDM-PON system delivering bidirectional baseband data and broadcasting data is proposed and demonstrated. A subcarrier multiplexing signal is broadcasted to all users by modulating a broadband optical source based on a Fabry-Pérot laser diode. Reflective semiconductor optical amplifiers are used as colorless modulators for the baseband data at both optical line terminal and remote optical network units. Transmission performance including bit error rate of bidirectional gigabit data and error vector magnitude of broadcasting data of many optical channels is investigated. Additionally, the data rate for the broadcasting signal was improved by using an external modulator.
©2009 Optical Society of America
1. Introduction
Wavelength-division-multiplexed passive optical networks (WDM-PON) have been considered as attractive broadband access networks with many advantages such as large bandwidths, bit rate independency, easy upgradeability and excellent security [1]–[4]. However, a virtual point-to-point connection based on an Arrayed Waveguide Grating (AWG) causes the wavelength-selectivity characteristic of optical line terminal (OLT) and the remote optical networks (ONUs) [1], [3]. It requires expensive wavelength-specified transmitters at OLT and ONUs. Consequently, for successful deployment of WDM-PON in real networks, the implementation of cost-effective and colorless optical devices is one of the most important issues. Furthermore, with the increase of the demand for diverse services from users, multimedia services such as voice, internet and broadcasting services are expected to be converged on a single WDM-PON [3]. Nevertheless, the wavelength-selectivity characteristic of ONUs also causes a cheap method to simultaneously transmit both broadcasting data and bidirectional baseband data to be challenging.
Recently, some WDM-PON structures have been proposed to address broadcasting problem. The overlay method utilizes a broadband light source (BLS) and cyclic property of an AWG [5]–[8]. These BLSs are based on Fabry-Pérot laser diode (F-P LD) [5], [6] or the ASE source of a an erbium doped fiber amplifier (EDFA) [7] or semiconductor optical amplifier (SOA) [8], All of these schemes, however, are costly and not optimum in term of provisioning bandwidth because a wavelength is dedicated for the broadcasting service of each ONU. Additionally, at each ONU, a WDM device is needed to separate wavelengths and another PD is required to recover the broadcasting data.
On the other methods, broadcasting data and baseband data share a single wavelength. A WDM-PON architecture, in which the combined signal directly modulates a Distributed Feedback (DFB) laser diode at OLT, was proposed [9]. A BLS based on mutually-injected F-P LDs was proposed to transmit both baseband data and broadcasting data [10]. However, this BLS employs two well-controlled F-P LDs with matched mode spacing to generate the low relative-intensity-noise (RIN) BLS. Furthermore, the noise peaks generated by the long cavity between the F-P LDs make the bandwidth for data transmission limited.
In this paper, a cost-effective, colorless WDM-PON architecture to transmit bidirectional baseband data and broadcasting data in a single wavelength is proposed. We have demonstrated for the first time the implementation of gain-saturated reflective semiconductor optical amplifiers (RSOAs) as the colorless modulators at the OLT for a WDM-PON supporting wired/wireless signals [14], [15]. High data rate broadcasting service is embedded in the system to realize a multimedia-converged system. In this proposed system, a single F-P LD based optical source modulated by a broadcasting subcarrier multiplexing (SCM) signal is shared among all bidirectional links. Colorless OLT and ONUs are realized by using RSOAs as modulators for bidirectional baseband data. Consequently, the number of light source for the whole system is dramatically reduced. From the experiment results including bit-error rate (BER) of 1.25-Gbps baseband data and error vector magnitude (EVM) of the 20Mbps/channel of the broadcasting SCM data, we have confirmed the validity of the proposed WDM-PON system for both baseband and broadcasting data transmission. A method to increase the broadcasting data rate is also investigated.
2. Principle of operation
The proposed colorless, bidirectional WDM-PON architecture supporting broadcasting function is schematically illustrated in Fig. 1 . At the OLT, a F-P LD is used as a BLS for many WDM channels. The broadcasting SCM signals modulate optical carriers of the F-P LD directly or externally. Each mode of the F-P LD is separated by an AWG and injected into a gain-saturated RSOA (RSOA1). At this RSOA, first of all, mode partition noise (MPN) generated due to spectrally slicing the F-P LD optical output is reduced by high pass filter (HPF) effect of the gain-saturated RSOA [13]- [15] and then the optical carrier is modulated by downlink baseband data. The optical output of RSOAs of all the channels are recombined by using another AWG and transmitted to the AWG-based remote node. At this remote node, the optical signal of each channel is separated again and transmitted to a specific ONU. At each ONU, a portion of downlink optical signal is detected by a photo-detector (PD) to recover both the baseband and broadcasting data. The other portion is injected into another RSOA (RSOA2), remodulated by uplink baseband data and then transmitted back to the OLT [10], [11].
3. Experimental setup
Figure 2 shows the experimental setup for the proposed scheme using direct modulation of the broadcasting signal. The F-P LD had the mode-spacing of 1.08 nm and the threshold current of 11 mA. The operating temperature and bias current of F-P LD were 25°C and 33 mA, respectively. Its optical spectrum is shown is Fig. 3(a) . The SCM signal with 16-QAM modulation format, 5 Msps at 2.5-GHz carrier directly modulated the optical carriers of the F-P LD with the modulation index (MI) of 40%. Because of high spectrum-slicing loss, before separated and injected into the RSOAs, the optical output of the F-P LD was amplified by using an EDFA to meet the requirement of the optical power for gain saturation of the RSOAs. Instead of the AWGs, two tunable optical filters (TOF) were used. The non-return-to-zero (NRZ) baseband data at the rate of 1.25-Gbps and peak-to-peak voltage (Vp-p) of 2V modulated the optical carrier at the RSOAs for both uplink and downlink. The extinction ratio (ER) was approximately 5.8dB. To avoid interferences between the baseband and the broadcasting signals, low pass filters (LPFs) with cut-off frequency of 1.2 GHz was used to remove the high frequency components of the baseband signal before modulating the optical carrier. The optical signal was transmitted through 23-km standard single mode fiber (SSMF). The optical signal of each downlink channel was divided into two paths by using a 3-dB coupler. For all the channels, the optical power of each path was about −12 dBm. The recovered broadcasting signal and the baseband signals were amplified using low noise amplifiers (LNA), filtered out and measured the performance. The characteristics of the RSOAs used in this experiment are presented in [14], [15]. Modulation frequency response was 1.1 GHz and the required injection power to operate in the gain saturation region was over −20 dBm. Their temperature and bias current were 25°C and 30 mA, respectively.
4. Results and discussion
Up to 10 channels, which are numbered from 1 to 10 as presented in Fig. 3(a), were selected for performance measurement. Optical powers of all channels with and without the EDFA, before and after RSOA1 are shown in Fig. 3(b). With the assistance of the EFDA, the injected optical power satisfied the requirement of gain-saturation condition for all the channels and the output optical power was about −1.5 dBm regardless of the injection power. The spectra of the detected RF signals of channel 9 are described in Fig. 4 including before and after RSOA1, with/without the data. As shown in this figure, the spectrally-sliced optical carrier contained MPN at low frequency. The MPN was also up-converted to the frequency region of the SCM signal (2.5-GHz band). After injected to the gain-saturated RSOA1, the MPN was reduced perfectly at both low frequency band and 2.5-GHz band. The baseband data also embedded in the subcarrier because the SCM data and the baseband data modulated the optical carrier in two consecutive steps. It would make the performance of the broadcasting data degraded.
4.1 Performance of the broadcasting SCM data
Without a gain-saturated RSOA, EVM of the detected SCM signal is unstable and relative high due to MPN [16]. Figure 5(a) shows the EVM curve of the SCM signal at 2.5-GHz subcarrier at back to back (BtoB) and after 23-km transmission using the gain-saturated RSOA. The EVM values of about 2.5% were achieved at −21 dBm at BtoB and it was about 1-dB power penalty to achieve the same EVM after 23-km transmission. This power penalty was originated from dispersion and reflection noises after transmission. The performance of the SCM data was stable regardless of the existence of the downlink baseband data with 5.8-dB ER. It means that the performance degradation was insignificant in this condition since the signal to noise ratio (SNR) of the SCM signal was still high. However, when the power of the SCM signal decreased, or the MI decreased, EVM increased. This influence on the EVM of the SCM signal for 10 channels is described in Fig. 5(b) in case of after 23-km transmission and at −20-dBm optical power by using an optical attenuator. We chose this optical power point because it is the minimum requirement of the optical power to satisfy gain-saturation condition of the RSOA for the uplink. High performance of the SCM data for all the channels was achieved when the MI was high in spite of the existence of the baseband data also with 5.8-dB ER.
Fig. 5 Performance of SCM signal: (a) EVM curve, (b) EVM of all channels, (c) effect of ER of downlink baseband signal and (d) EVM at different subcarrier frequencies.
The influence of the ER of the downlink baseband signal on the EVM with the MI of 30% and the sensitivity for error-free transmission of this data is illustrated in Fig. 5(c). This modulation index was chosen to easily analyze the effect of the baseband data because the lower modulation index is, the higher effect of the baseband data on the broadcasting data. With the modulation index of 40%, the difference of the EVM values in cases of with and without the baseband data was very small, less than 0.5%. It is because the SNRs of the broadcasting signal of this channel were very high in both cases. As presented in the figure, when the ER increased, the optical power to achieve error-free transmission (BER of 10−9) decreased. The EVM also increased because the increased baseband data embedded in the subcarrier made the SNR of the SCM signal reduced. The filled region shows the optimum region to achieve the excellent performance for both the baseband signal and the broadcasting signal at −20-dBm optical power. Error-free transmission of the baseband data and the EVM of less than 5% could be accomplished although the ER varied from 2.3 to 5.8 dB (Vp-p varied from 0.9 to 2V). The effect of the Vp-p on the performance of the baseband signal in up/downlink has been analyzed in a previous work [9]. We chose the maximum value of the Vp-p (2V) to analyze the performance of the SCM data in other cases because it had the worse effect on the SCM data.
The EVM values of the SCM signal at different subcarrier frequencies are shown is Fig. 5(d). When the subcarrier frequencies were far away from the unfiltered band of the baseband signal, the EVM is maintained at 2.5%, otherwise the EVM increased dramatically. There are two causes for this phenomenon. First, the interference from the baseband signal caused the SNR of the SCM signal to be reduced. Second, the SCM signal at low frequency was more suppressed than a signal at high frequency due to the HPF characteristic of the gain-saturated RSOA. From the figure it can be concluded that any frequencies out of the unfiltered band of the baseband signal can be used for the SCM data.
4.2 Performance of the baseband data
As presented by authors in other works, with the use of a F-P LD as a BLS and RSOAs as modulators, error-free transmission for both the uplink and downlink baseband data is achievable for many channels without any effect of the MPN [14]- [15]. In this work, only BER curves for both the downlink and uplink data with 5.8-dB ER of channel 9 were measured which are shown in Fig. 6 . Error-free (BER<10−9) transmission at BtoB was achieved at −25.8 dBm for the downlink and −23 dBm for the uplink. The sensitivity of the uplink baseband data was about 3 dB worse than that of the downlink because remained downlink data was the noise to the uplink data. The effect of MPN on the uplink baseband data could be neglected because the optical signal of each channel underwent gain-saturation effect twice and MPN was already suppressed totally after the first gain-saturated RSOA. The performance after 23-km transmission was about 1.5 dB worse than BtoB. Large dispersion due to the wide line width of the F-P LD based light source is a main cause of such high power penalty. Other factors that contribute to the high power penalty are reflection noises and measurement error. The existence of the SCM signal can causes the degradation of the performance of the baseband data due to two reasons.
First, the non-linearity of the F-P LD generates inter-modulation distortion (IMD) products. Any IMD products or the broadcasting signal itself appearing in the main frequency lobe of the baseband signal make the SNR of the baseband signal reduced. However as illustrated in this figure, there is insignificant difference of the BER characteristics between the cases of with and without the existence of the SCM signal. This is because the frequency of this signal is far away from the main lobe of the baseband data which was perfectly removed by using a LPF with the cut-off frequency of 1.2 GHz and the gain-saturated RSOA also suppressed any IMD components in the low frequency region.
Second, as mentioned above, the baseband data was also up-converted to the subcarrier frequency band so if the subcarrier frequency is close to the frequency band of the baseband signal, the left sideband of the modulated subcarrier can overlap the main lobe of the baseband signal and it acts as noise to the baseband signal. It makes the performance of the baseband signal degraded. In Fig. 7 , power penalty, the power difference to achieve the same BER of 10−9 for the baseband data with 5.8-dB ER in cases of with and without the SCM data, increased when the subcarrier frequency reduced. However, it was just about 1.4 dB when SCM frequency was 1.2 GHz. From Fig. 5(d) and Fig. 7, it can be concluded that a broad frequency band out of the main lobe of the baseband data could be used for the SCM data with a small effect between them.
4.3 Improvement of the data rate for the broadcasting service
In the experimental setup, one SCM signal directly modulated the optical source of the F-P LD. To achieve high performance for all the channels, a high MI up to 40% was required. As a result, the number of subcarrier or data rate for the broadcasting service is restricted. To increase the bit rate for the broadcasting service, another experiment was conducted. Instead of direct modulation, a Mach-Zehnder modulator (MZM) was used. The bias point of the MZM was 6V. This is the averaged quadrate point (Vπ/2) of the MZM for the wavelength range from 1533 to 1545 nm [14]. As discussed in the previous section, the subcarrier is also modulated by the baseband signal so when the subcarrier frequency is close to the main lobe of the baseband signal, the modulated subcarrier acts as noise to the baseband signal and reduce the performance of the baseband signal. In the case of using an external modulator, which has higher frequency response than the directly modulated F-P LD, it is better to use high frequency because this effect can be eliminated. From this point of view, we used 16-QAM, 5-Msps SCM broadcasting signal at 2.5-GHz subcarrier. It was combined with a RF tone and then modulated the optical carriers from the F-P LD at the MZM. The frequency of RF tone was 2.48 GHz. This frequency was chosen because it was close to the subcarrier. When it was modulated by the baseband data, it had worse effect on the SCM signal than RF tones at other frequencies. The RF power of the SCM signal was 0 dBm and the RF power of the RF tone was chose at 0, 3, 6, 9, 12 dBm. It mean that total RF power was respectively equivalent to the RF power of 2, 3, 5, 9 and 17 channels with the RF power of 0 dBm per channel.
Together with optical channel 9, optical channel 1 was chosen for performance analysis. This optical channel had the worst performance in case of direct modulation. Figure 9 shows the dependence of the EVM of the SCM data on the number of the broadcasting channel and the ER of the downlink baseband data for the two optical channels. The EVM also measured at −20-dBm optical power after 23-km transmission. As illustrated in the figure, when the ER or the number of channels increased, the EVM increased. In the first case, when the ER increased, the level of the baseband data embedded in the RF tone and the subcarrier increased. In the second case, when the number of channel increased, in this experiment it mean that the RF tone power increased, the level of the baseband data embedded in the RF tone also increased. In the both two conditions, the noise level of the SCM signal increase or in the other words, the SNR of the SCM signal decreased. However, when the ER increased, the sensitivity of the downlink baseband data increased.
Fig. 9 The dependence of EVM on ER of downlink signal and number of broadcasting channel a) optical channel 9 and b) optical channel 1.
Additionally, the nonlinear transfer characteristic of the MZM also generates the nonlinear products that deteriorate the SNR at the receiver [17]. However in this scheme, the bias voltage of the MZM was fixed at the quadratic point where the second IMD was suppressed and both the second and third IMD products also can be further suppressed by using the gain-saturated RSOA [17].
Similarly to the case of direct modulation, using multiple subcarriers to increase the data rate of the broadcasting signal made the performance of the baseband signal deteriorate because of the generation of IMD components and the interference from the modulated subcarriers. These effects can be mitigated by using proper parameters including the bias point of the MZM and frequencies of subcarriers as mentioned before. The increase of the number of broadcasting channels had other effect on the performance of the baseband signal. The proportion optical signal contained the broadcasting data increase but the total optical power is the same because of gain saturation. As a result of the statistical beating process of the PD, the received signal power at SCM frequency band increases but that at baseband frequency decreases. However, high transmission performance with up to 17 broadcasting channels for optical channel 1 and 9 was still achieved when the ER of the downlink baseband data was 2.1 dB (0.8 Vp-p). It means that it also could be achieved for the other optical channels. At 2.1-dB ER, sensitivity of the baseband data for BER of 10−9 of optical channel 1 and 9 after 23-km transmission was −15.5 and −17 dBm, respectively. Using such low ER of the downlink data, error-free transmission is achievable for this data and not only performance of the broadcasting data is maintained but also the performance of the uplink baseband data is enhanced [11].
5. Conclusion
We have proposed and demonstrated a novel WDM-PON system to deliver bidirectional baseband and broadcasting signals on a single wavelength. This cost-effective, colorless WDM-PON overcomes the wavelength-selectivity feature by using only one F-P LD as a BLS and gain-saturated RSOAs as optical modulators at both the OLT and the ONUs. We demonstrated the transmission of bidirectional 1.25-Gbps baseband data and 20-Mbps broadcasting data for many channels through 23-km SSMF. The effects between the two data were also analyzed. Almost all high data performance was achieved at −20dBm optical power therefore with up to 8 dB power budget for both the uplink and downlink, the transmission distance of the proposed WDM-PON system can be extended while the performance is maintained. The data rate of the broadcasting signal could be significantly increased by adopting external modulation.
Acknowledgments
This work was supported by Ministry of Knowledge Economy and Yonsei University Institute of TMS Information Technology, a BK21 program, Korea.
References and links
1. N. J. Frigo, P. P. Iannone, P. D. Magill, T. E. Darcie, M. M. Downs, B. N. Desai, U. Koren, T. L. Koch, C. Dragone, H. M. Presby, and G. E. Bodeep, “A wavelength-division multiplexed passive optical network with cost-shared components,” J. Lightwave Technol. 6, 1365–1367 (1994).
2. S. L. Woodward, P. P. Iannone, K. C. Reichmann, and N. J. Frigo, “A spectrally sliced PON employing Fabry-Perot lasers,” IEEE Photon. Technol. Lett. 10(9), 1337–1339 (1998). [CrossRef]
3. Y. C. Chung, “Challenges toward practical WDM PON,” in Proceedings of Optoelectronics and Communications Conf. (OECC2006), Paper 6C4–1, (2006).
4. 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]
5. J. H. Moon, K. M. Choi and C. H. Lee, “Overlay of broadcasting signal in a WDM-PON,” in Tech. Dig. Optical Fiber Communications Conf. (OFC2006), Paper OThK8, (2006).
6. H. C. Kwon, Y. Y. Won, and S. K. Han, “Noise suppressed Fabry-Perot laser diode with gain-saturated semiconductor optical amplifier for hybrid WDM/SCM-PON link,” IEEE Photon. Technol. Lett. 18(4), 640–642 (2006). [CrossRef]
7. J. Cho, J. Kim, D. Gutierrez and L. G. Kazovsky, “Broadcast transmission in WDM-PON using a broadband light source,” in Tech. Dig. Optical Fiber Communications Conf. (OFC2006), Paper OWS7, (2006).
8. J. M. Kang, S. H. Lee, H. C. Kwon, and S. K. Han, “WDM-PON with broadcasting function using direct ASE modulation of reflective SOA,” in Tech. Dig. Optical Fiber Communications Conf. (OFC2006), Paper P.160, (2006).
9. T. Y. Kim and S. K. Han, “Reflective SOA-based bidirectional WDM-PON sharing optical source for up/downlink data and broadcasting transmission,” IEEE Photon. Technol. Lett. 18(22), 2350–2352 (2006). [CrossRef]
10. H. C. Ji, I. Yamashita, and K. I. Kitayama, “Cost-effective colorless WDM-PON delivering up/down-stream data and broadcast services on a single wavelength using mutually injected Fabry-Perot laser diodes,” Opt. Express 16(7), 4520–4528 (2008). [CrossRef] [PubMed]
11. H. Takesue and T. Sugie, “Wavelength channel data rewrite using saturated SOA modulator for WDM metworks with centralized light source,” J. Lightwave Technol. 21(11), 2546–2556 (2003). [CrossRef]
12. J. J. Koponen and M. J. Soderlund, “A duplex WDM passive optical network with 1:16 power split using reflective SOA,” in Tech. Dig. Optical Fiber Communications Conf. (OFC2004) 1, 23–27 (2004).
13. K. Sato and H. Toba, “Reduction of mode partition noise by using semiconductor optical amplifiers,” IEEE J. Sel. Top. Quantum Electron. 7(2), 328–333 (2001). [CrossRef]
14. T. T. Pham, H. S. Kim, Y. Y. Won, and S. K. Han, “Bidirectional 1.25-Gbps wired/wireless optical transmission based on single sideband carriers in Fabry-Perot laser diode by multi-mode injection locking,” J. Lightwave Technol. 27(13), 2457–2464 (2009). [CrossRef]
15. H. S. Kim, T. T. Pham, Y. Y. Won, and S. K. Han, “Simultaneous wired and wireless 1.25-Gb/s bidirectional WDM-RoF transmission using multi optical carrier Suppression in FP LD,” J. Lightwave Technol. (to be published).
16. H. C. Kwon, Y. Y. Won, and S. K. Han, “Noise suppressed Fabry-Perot laser diode with gain-saturated semiconductor optical amplifier for hybrid WDM/SCM-PON link,” IEEE Photon. Technol. Lett. 18(4), 640–642 (2006). [CrossRef]
17. D. H. Jeon, H. D. Jung, and S. K. Han, “Mitigation of Dispersion-Induced Effects Using SOA in Analog Optical Transmission,” IEEE Photon. Technol. Lett. 14(8), 1166–1168 (2002). [CrossRef]
