We have proposed a stable, wideband, and tunable directly modulated fiber ring laser (TDMFRL) by using a reflective semiconductor optical amplifier (RSOA) and an optical tunable filter (OTF). For use in a bidirectional access network, the TDMFRL not only generates downstream data traffic but also serves as the wavelength-selecting injection light source for the Fabry-Pérot laser diode (FP-LD) located at the subscriber site. We experimentally demonstrated a bidirectional transmission at 1.25-Gb/s direct modulation over a 25-km single-mode fiber (SMF), thereby showing good performance in a wavelength division multiplexing (WDM) access network.
© 2010 OSA
Wavelength division multiplexing-passive optical network (WDM-PON) technology has been considered to be a powerful means to realize an access network that is capable of handing the ever increasing demand of higher data bandwidth, enhanced security, and greater scalability desired by local subscribers [1,2]. Implementation of WDM techniques for a local access network requires cost-effective and tunable sources. Recently, various WDM-PON systems were proposed having a remote-seeding optical source located at a central office (CO) to control the wavelength of an upstream-transmitting FP-LD source in the optical network unit (ONU) located at the user end. Several types of light sources have been proposed to injection-lock the FP-LD, including a broadband light source (BLS) based on amplified spontaneous emission (ASE) [3,4], distributed feedback laser diodes (DFB-LDs) , a spectrally sliced free-running FP-LD [6,7], and a tunable laser source (TLS) . A BLS with a wide spectral bandwidth can be easily obtained, but the fact that the spectral bandwidth decreases as the output power increases makes it very difficult to achieve practical applications. DFB-LDs are still expensive solutions even though they can emit over a wide wavelength range and thus can enable a large number of channels. The performance of spectrally sliced free-running FP-LDs is inherently limited by the intensity noise induced by the mode partition and/or the mode fluctuation. TLS are complex to implement because they need separate external modulators. Moreover, a conventional fiber laser also uses an external modulator for signal modulation. This approach, despite its obvious advantages for some applications, is costly and cumbersome due to the cost of a modulator and the associated high-voltage, high-speed electronic circuitry. To overcome these problems, we proposed a tunable directly modulated fiber ring laser as a cost-effective solution to reduce the installation and management costs.
In this paper, we used our proposed TDMFRL light source as an injection-locking source for the FP-LD located in an ONU to reduce the cost of the system, to effectively manage the selection of upstream wavelengths, and to implement a centralized network control system. More specifically, the TDMFRL has a wide tunable range of 45 nm (1535 – 1580 nm), an average output power of −5.5 dBm, and a good side-mode suppression ratio (SMSR) above 38 dB. Moreover, we demonstrated efficient WDM access network using direct modulation technique in both downstream and upstream bidirectional transmission at a data rate of 1.25-Gb/s over a 25-km SMF. Good performances of bit error rate (BER) and a low power penalty were obtained in our system. The implementation of an injection-locked FP-LD by using a TDMFRL could be relatively simple and cost-effective compared with other reported light source schemes.
2. Proposed scheme and experimental setup
2.1 Proposed scheme
Figure 1 shows the schematic diagram of proposed WDM access networks. Firstly, downstream data channels can be provided by using a TDMFRL at the CO. The TDMFRL also serves a secondary role as the master laser to injection-lock onto the WDM grid a directly modulated slave FP-LD located at the ONU. The locked wavelength of the FP-LD allows a specific port to be assigned at the CO for each ONU without the need for complicated wavelength stabilization circuitry. The TDMFRL can also be used as a spare/inventory/testing transmitter for various wavelength channels in WDM systems. Furthermore, the FP-LD modulation is not strongly affected by the master data and thus both downstream and upstream transmissions can achieve good performance. Next, the upstream transmission in our scheme is performed with a wavelength identical to that used for the downstream transmission for each ONU. To avoid crosstalk and backscattering effects, we employ an optical circulator (OC) and adopt a two-wire single-wavelength configuration for simplicity. Two spans of 25-km standard SMF are used as transmission links between the CO and the ONU. The remote node (RN) comprises demultiplexer/multiplexer (DMUX/MUX) to route the wavelength channel from the CO to the ONU via the fiber links. The TDMFRL has a wide tuning range covering all wavelengths corresponding to those selected by the DMUX/MUX. The signal light of a selected wavelength is injected into a specific FP-LD. When the signal wavelength is close to one of the FP longitudinal modes, injection-locking to that mode occurs and results in an intensity suppression of the other FP comb modes. This intensity suppression is related to the wavelength detuning between the signal wavelength and FP mode. It is worthwhile to note that a FP-LD with larger detuning range is more suitable for applications to cases of larger wavelength/temperature variation. Furthermore, while one of the FP modes is assigned for a specific ONU, our TDMFRL can be used to detect its state or test its network performance. In our experiment, the temperature variation of the peak wavelength of FP-LD was around 0.5 nm/°C and no difficulty was found to injection-lock the FP-LD mode using our TDMFRL.
2.2 Experimental setup
The experimental setup for our proposed scheme using the TDMFRL in WDM access networks is shown in Fig. 2 . Figure 2(a) shows the TDMFRL architecture. This ring was constructed by a RSOA that was driven into a state of saturation, thereby having the RSOA act as a noise suppresser and a gain medium [9,10]. The light polarization input state to the RSOA was adjusted by an optical polarization controller (OPC1). The unidirectional operation of the laser was ensured by an optical circulator (OC1). An OTF with a 3-dB bandwidth of 0.3 nm and a tuning range of 1535 nm to 1610 nm was inserted in the cavity as a wavelength-selective element. The output of the laser was characterized by an optical spectrum analyzer (OSA) through a 10/90 optical coupler. For downstream transmission, the TDMFRL acting as a downstream signal source was directly modulated with an 1.25-Gb/s (215-1) non-return-to-zero (NRZ) pseudo random bit sequence (PRBS) signal (λD1) from a pulse pattern generator (PPG). The TDMFRL also serves a secondary role as the master laser to injection-lock onto the WDM grid a directly modulated slave FP-LD located at the ONU. After 25-km of SMF transmission, the data passed through the de-multiplexer DMUX located at the RN. The DMUX has a channel spacing of 100 GHz (0.8 nm) to act as a multi-channel filter and was directed to a receiver at the ONU. The transmission performances were measured by a digital communication analyzer (DCA) and a bit error rate tester (BERT).
The locked wavelength of the FP-LD allows a specific DMUX port to be assigned at the CO for each ONU without the need for complicated wavelength stabilization circuitry. The commercial MQW FP-LD with a free-spectral range of 1.36 nm was injection-locked by the light from the TDMFRL at a specific channel. The FP-LD was driven by a NRZ PRBS signal with a word length of 215-1 at 1.25-Gb/s provided by a PPG. After the injection by the TDMFRL at a specific channel with enough injection power, the cavity mode located at that wavelength was injection-locked and enhanced in intensity, while other modes were suppressed significantly. The modulated output (λU1) was transmitted upstream through a MUX via a 25-km SMF. A characterization system that includes an OSA, a DCA, and a BERT was used to evaluate the transmission performance of the network links.
3. Experimental results and discussions
3.1 TDMFRL output characteristics
To describe our demonstration, first we show the output characteristics of the TDMFRL. For operation in a saturation state, the RSOA was driven by a 60 mA current. Figure 3 shows the superimposition spectra of multiple single-wavelength outputs of the TDMFRL. An extended tunable range over 45 nm (1535 – 1580 nm) covering both the C-band and L-band was observed. The TDMFRL output peak power is 8.2 dBm at 1555 nm measured by power meter. Because of OSA input limit, an optical variable attenuator was used to reduce the laser output power to −5.5 dBm during our spectral measurement. The optical SMSR is above 38 dB. In order to realize and investigate the performances of SMSR and wavelength stability, the long-term stability measurement of the TDMFRL was performed, as shown in Fig. 4 . The initial 1545.82 nm lasing wavelength was monitored over a 90-minute time period. We observed that the output power fluctuation and wavelength variation of the lasing central lightwave are smaller than 0.2 dB and 0.03 nm, respectively. As a result, the proposed fiber laser not only can generate continuously tunable wavelengths, but also has good long-term output stability.
The gain saturation characteristic of the RSOA is shown as a function of bias current in Fig. 5(a) . The curve can be divided into three regions, region I for the current below 15 mA, region II between 15 and 30 mA, and region III above 30 mA. Gain saturation has achieved in the high gain region (region III). The RSOA uses CIP’s mode expander technology coupled with a curved waveguide and angled front to achieve its high gain. At 20 °C, a gain of 44 dB was observed for the 50-mA bias current and −40-dBm input power at 1555-nm wavelength, as shown in Fig. 5. Under a similar condition, the gain can be reduced to 30 dB if the input power is −30 dBm.
The quality of the TDMFRL output depends also on the bias current of the RSOA and can be depicted by the RF electrical spectrum of the laser, which was obtained by using an optical-to-electrical converter connected to a RF electrical spectrum analyzer (ESA). The RF electrical spectra were measured with a 1-MHz resolution at saturation and low-bias currents, respectively. For a low bias current of the RSOA [regions I and II in Fig. 5(a)], poor suppression of laser noise was observed. A relative intensity noise (RIN) of −118.6 dB/Hz was observed, as shown in Fig. 5(b). If the RSOA current is increased to a state of saturation (region III), the suppression of laser noise is significant, since the RSOA acts as a high-pass filter. The measured RIN is −170.48 dB/Hz, as shown in Fig. 5(b). Its performance can be compared with commercial tunable semiconductor laser sources (TSLSs) in terms of lasing stability and the number of lasing wavelengths.
Furthermore, the quality of the TDMFRL output can be indicated by eye diagrams. The insets in Fig. 5(c) show eye diagrams for data transmissions over 25-km SMF, using the TDMFRL direct modulated by an 1.25-Gb/s NRZ data. Three eye diagrams are shown for the TDMFRL with the RSOA biased at three current levels (in regions I, II, and III of Fig. 5(a), respectively). In region I, the eye diagram is not open because of laser noises. When the RSOA current is between 15 and 30 mA in region II, the eye diagram is blurred. It is believed that the fiber laser might operate in multimode region and may not suitable for an 1.25-Gb/s and 25-km SMF data transmission. When the RSOA current is higher than 30 mA in region III, good signal qualities in Q-value and extinction ratio (ER) are above 10. The eye diagram is open and clear. Thus, the TDMFRL is suitable for high-speed data communications.
3.2 Injection-locked behavior of FP-LD by TDMFRL
Before describing the system performance of bidirectional transmissions using our wideband TDMFRL and an injection-locked FP-LD, we first show the effect of injected power on the wavelength locking of the FP-LD at wavelengths in the C-band. The TDMFRL was directly modulated by a downstream 1.25-Gb/s NRZ PRBS signal and was used to injection-lock a FP-LD. The FP-LD was also directly modulated by an upstream NRZ PRBS signal at 1.25-Gb/s. After transmission over a 25-km SMF, the signal was detected by a receiver (Rx) in the CO, with a built-in avalanche photodetector followed by a BERT. Without applying any additional amplification device to our experimental setup shown in Fig. 2, the TDMFRL output power is 7.2 dBm at 1547.82 nm. Because of the SMF loss of 0.2 dB/km, the 1 × 8 DMUX loss about 3 dB per channel, the 50/50 coupler loss of 3 dB, and the OC2 insertion loss of 0.6 dB, the injection power input to the FP-LD is −4.83 dBm. Figure 6 shows the measured optical spectra for a C-band FP-LD with and without the injection from the TDMFRL. The C-band FP-LD was biased at 24.5 mA (with around a 1 mW output power) and operated at 21 °C with a temperature controller. Its temperature coefficient is 0.53 nm/°C. Without optical injection, multiple cavity modes oscillated with wavelengths such as those shown by the dotted curve in Fig. 6. After the injection of TDMFRL light at 1547.82 nm with an injection power of −4.83 dBm, the cavity mode located at that specific wavelength was injection-locked and enhanced in intensity by 5.47 dB, while other modes were suppressed significantly and had reduced linewidths. The output peak power was much larger than that without injection. A SMSR of greater than 38 dB was observed.
The injection-locked behavior depends on the “0” bits of the upstream OOK data. To investigate further the effect of the number of “0” bits of the upstream data on the injection-locking of the FP-LD which is directly modulated by a PPG, we show the value of ER as a function of continuous “0” pattern length in Fig. 7 . The FP-LD was biased at 24.5 mA and operated at 21°C. The peak-to-peak driving voltage of the PPG was 0.77 V. As the continuous “0” pattern length increases from 10 to 100 bits, the ER decreases from 10.19 to 4.07 dB. When the “0” pattern length is less than 70 bits, the eye diagram is clear with an ER larger than 6.38 dB, A typical eye diagram is shown in the inset of Fig. 7(a) for a “0” pattern length of 40 bits. On the other hands, when the “0” pattern length of longer than 70 bits, a wide “0” level rail appears, as shown in the inset of Fig. 7(b) for 80 bits. Thus, it is suggested that the maximum number of continuous “0” bits of the upstream data is 70 bits for the use of the TDMFRL and the injection-locking scheme.
The injection-locked behavior depends also on the injection optical power. Figure 8 shows the measured SMSR of the FP-LD after injection and the Q-value for 25-km SMF upstream transmission as functions of injection optical power at two wavelengths (1547.82 nm and 1550.25 nm). When the injection power is less than a value of −8 dBm, the SMSR and Q-value can be improved significantly with the increase of the injection power. When the injection power is between −8 dBm and 0 dBm, the SMSR remains almost at 38 dB and the Q-values better than 10. An injection optical power larger than 0 dBm can make the level “1” of the upstream signal degrade and the Q-value and SMSR of the upstream signal decrease. Thus, it is concluded that good wavelength-locking performance can be obtained for the injection power range from −8 dBm to 0 dBm.
As to the injection-locking stability, we observed the system performance over 60 minutes. The output spectra are shown in Fig. 9(a) . It shows that the wavelength remains stable and the temporal power fluctuation is less than 0.75 dB. Possible sources of the fluctuation could come from variations in the temperature of the FP-LD and the polarization states of the OPC2. It is worthwhile to note that a fine adjustment of the OPC2 can help to obtain small linewidth and small power fluctuation. Apart from the OSA measurement, we also made a dynamic study of the system stability. The injection-locked FP-LD was directly modulated with an upstream signal at 1.25-Gb/s and SMSR and Q-value were measured for 60 minutes. In order to measure SMSR and Q-value at the same time by using an OSA and a DCA, we connected a 50/50 coupler to the port 3 of OC2. From the results shown in Fig. 9(b), the SMSR is above 38.26 dB and the fluctuation is less than 0.75 dB. The Q-values are about 9.75. They indicate the injection-locked FP-LD can operate stably and is good for an 1.25-Gb/s upstream transmission.
3.3 Performance test of bidirectional transmissions
To test the bidirectional transmission performance of the proposed TDMFRL system, we firstly demonstrate in Fig. 10 that a single FP-LD can be used as a light source for many wavelength channels selectable using the before-mentioned TDMFRL. Three wavelength channels at 1538.05 nm, 1545.83 nm, and 1550.25 nm were observed, as shown in Fig. 10. The measured SMSR are larger than 38 dB when the injection power is above −8 dBm. Next, performance test results are presented for two channels at 1545.83 nm and 1550.25 nm. To verify that the proposed scheme can be implemented in bidirectional WDM access networks, a BER measurement was carried out using direct-modulation method and injection-locking technique. Data were transmitted in both downstream and upstream directions simultaneously through two rolls of 25-km SMF. Figure 11 shows the BER curves at 1545.83 nm and 1550.25 nm and typical eye diagrams at 1545.83 nm. Circle and triangle keys represent results of 1545.83 nm and 1550.25 nm, respectively. Solid and hollow keys are for back to back (BTB) and 25-km transmission, respectively. Almost error-free transmissions were observed for both channels. For the downstream transmission at 1545.83 nm, the Q-values are 9.1 and 7.31 for BTB and 25-km SMF transmissions, respectively. The BTB receiver sensitivity is −27.91 dBm and the power penalty is 0.11 dB for 25-km SMF transmission at a BER of 10−9. For the upstream transmission with injection locking light at 1545.83 nm, the Q-values are 10 and 8.9 for BTB and 25-km SMF transmissions, respectively. The BTB receiver sensitivity is −30.7 dBm and the power penalty is 0.86 dB for 25-km SMF transmission at a BER of 10−9. Similar results were observed for other channels.
It is interesting to study the system performance of a few more wavelengths across the WDM grids. We have carried out a transmission experiment using two light sources with different wavelengths located at the CO in Fig. 1. They are a TDMFRL and a DFB laser diode. The BER data of the wavelengths at 1545.83 nm and 1550.25 nm are shown as square keys in Fig. 11. Solid and hollow keys are for BTB and 25-km transmission, respectively. Comparing the curves with solid square (for 1545.83 nm and 1550.25 nm simultaneously) and solid circle (for 1545.83 nm only) keys for the upstream BTB case at a BER of 10−9, we can reach a conclusion that the crosstalk level due to 1550.25 nm is relatively small. Similar conclusion can also be reached for square (for 1545.83 nm and 1550.25 nm simultaneously) and triangle (for 1550.25 nm only) keys.
A wavelength tunable laser source is proposed for a bidirectional WDM access network in which injection-locked FP-LDs at the ONUs are used for upstream transmissions. We have investigated and experimentally demonstrated a simple TDMFRL using a RSOA and an OTF as a source located at the CO. This source is not only capable of direct intensity modulation and has a wide tuning range to generate stable wavelengths for downstream transmission, but also serves as the injection-locking source for upstream transmission wavelengths. Using a proposed simple architecture, the bidirectional transmission over 25-km of SMF link can be successfully achieved at a data rate of 1.25-Gb/s. We believe that our proposed scheme should prove to be a cost-effective, short-reach WDM system solution for the implementation of high-capacity access networks.
This work was supported in part by the National Science Council of the R.O.C. through NSC-98-3114-P-011-001-Y, NSC-97-2219-E011-003 and NSC-97-2218-E011-007-MY3 projects.
References and links
1. S. J. Park, C. H. Lee, K. T. Jeong, H. J. Park, J. G. Ahn, and K. H. Song, “Fiber-to-the-home services based on wavelength-division-multiplexing passive optical network,” J. Lightwave. Technol. 22(11), 2582–2591 (2004). [CrossRef]
2. K. Iwatsuki, J. I. Kani, H. Suzuki, and M. Fujiwara, “Access and metro networks based on WDM technologies,” J. Lightwave. Technol. 22(11), 2623–2630 (2004). [CrossRef]
3. H. D. Kim, S.-G. Kang, and C.-H. Lee, “A low-cost WDM source with an ASE injected Fabry-Pérot semiconductor laser,” IEEE Photon. Technol. Lett. 12(8), 1067–1069 (2000). [CrossRef]
4. X. Cheng, Y. J. Wen, Z. Xu, and Y. Wang, “Characterization of Fabry-Pérot laser diodes injection locked by spectrum sliced ASE noise in WDM-PON,” Opt. Fiber Technol. 15(2), 161–164 (2009). [CrossRef]
5. 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]
6. H. C. Kwon and S. K. Han, “Performance analysis of a wavelength-locked Fabry-Perot laser diode by light injection of an external spectrally sliced Fabry-Perot laser diode,” Appl. Opt. 45(24), 6175–6179 (2006). [CrossRef] [PubMed]
7. Y.-S. Liao, H.-C. Kuo, Y.-J. Chen, and G.-R. Lin, “Side-mode transmission diagnosis of a multichannel selectable injection-locked Fabry-Perot Laser Diode with anti-reflection coated front facet,” Opt. Express 17(6), 4859–4867 (2009). [CrossRef] [PubMed]
8. C.-L. Tseng, C.-K. Liu, J.-J. Jou, W.-Y. Lin, C.-W. Shih, S.-C. Lin, S.-L. Lee, and G. Keiser, “Bidirectional transmission using tunable fiber lasers and injection-locked Fabry-Pérot laser diodes for WDM access networks,” IEEE Photon. Technol. Lett. 20(10), 794–796 (2008). [CrossRef]
9. 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]
10. L. Xu, I. Glesk, D. Rand, V. Baby, and P. R. Prucnal, “Suppression of beating noise of narrow-linewidth erbium-doped fiber ring lasers by use of a semiconductor optical amplifier,” Opt. Lett. 28(10), 780–782 (2003). [CrossRef] [PubMed]