By controlling the extinction ratio (ER) and overshooting level of the down-stream amplified spontaneous emission (ASE) with a gain-saturation semiconductor optical amplifier (SOA), the down-stream data-erased ASE carrier is re-encoded in an injection-locked weak-resonant-cavity Fabry-Perot laser diode (WRC-FPLD) up-stream transmitter to implement all-ASE based bi-directional WDM-PON system. The effect of ER on the up-stream transmission performance of the down-stream data-erased ASE injection-locked WRC-FPLD is elucidated via the gain-saturation model. It is observed that the communication criterion with a bit-error-rate of <10−9 at 2.488 Gbit/s can be met only when ER is reduced to <3 dB and overshooting level <-5 dB. The up-stream WRC-FPLD re-encoded ASE data-stream could improve its signal-to-noise ratio (SNR) to 6.4 dB by minimizing the ER and overshooting level of the down-stream data-erased ASE to 2.4 dB and −7.8 dB, respectively, with the gain-saturated SOA. The SNR can also be improved with higher power injecting into the up-stream transmitter until saturation occurs and the optimal window of the ASE injection power is between −7 and −3 dBm.
©2011 Optical Society of America
With the development of multimedia entertainment and service, the demand for high capacity fiber-optic transmission becomes larger, and the dense wavelength-division multiplexed passive optical network (DWDM-PON) is considered to satisfy the demand in the near future. However, a wavelength-independent transmitter is usually needed for cost-effectiveness and therefore many types of WDM-PON transmitters have been proposed to meet the requirement . Although the system employing spectrum-sliced incoherent light sources was once attractive, the insufficient output power limits the practical application of such transmitter in WDM-PON communication network [2,3]. Subsequently, several remodulation schemes employing higher power light sources such as the amplified spontaneous emission (ASE) sources to injection-lock the Fabry-Perot laser diode (FPLD)  or the semiconductor optical amplifiers (SOA)  have been proposed. The technique using only the injection-locked SOAs is not cost-effective due to the addition of the external optical modulator, while the injection-locked FPLD could be directly modulated despite the limited bandwidth. Recently, an ultralow surface reflectivity FPLD is employed to be a colorless transmitter at 1.25 Gbit/s over 20-km transmission  for a wavelength-independent operation of WDM-PON. Such device bears no strict mode selectivity as compared to the conventional FPLD and could have low intensity noise.
Our previous work of the weak-resonant-cavity FPLD (WRC-FPLD) based up-stream transmitter  was focused on discussing a preliminary system using an ASE source, which is not encoded by any data-stream and is employed as an injection-locker after channelization. The extinction and signal-to-noise ratios of the WRC-FPLD transmitter injection-locked by the spectrum-sliced pure ASE located at remote node were reported, while the location of the ASE and its effect on the up-stream WRC-FPLD transmission performance have also been compared. At then, only the uni-directional up-stream WDM-PON transmission can be demonstrated and such a channelized ASE source located at the optical network unit (ONU) part is required additionally. In this work, we re-allocate the ASE carrier at central office and encode it with PRBS data to demonstrate a real ASE based down- and up-stream bi-directional transmission via the use of WRC-FPLD as an up-stream ASE data re-encoder. Afterwards, we erase the down-stream data encoded on the ASE carrier before entering the WRC-FPLD, and characterize the relationship between the purity of down-stream data-erased ASE carrier and the transmission performance of the re-encoded up-stream data from the injection-locked WRC-FPLD. We attempt to investigate the possibility of re-encoding the ASE data-stream data in the WRC-FPLD after a suitable wash-out (or data-bleaching) procedure. The signal-to-noise ratio (SNR) saturation effect in the re-encoded down-stream ASE data output from the injection-locked WRC-FPLD is also theoretically elucidated.
Nevertheless, the crosstalk effect inevitably occurs in such remodulation scheme due to the simultaneous injection of the down-stream data. The unsuppressed part of the injected data becomes an excess noise source during the up-stream transmission, and hence reduces the SNR of the up-stream data. Therefore, the SNR needs to be increased by eliminating the extinction ratio (ER) of the down-stream data for injection-locking the up-stream transmitter to alleviate such an unwanted cross-talk. With a SOA operating at gain-saturation regime, the ER and overshooting level of the down-stream data could be greatly reduced before entering the up-stream transmitter and thus the communication quality is further improved. Although the structure combining a wavelength-locked FPLD transmitter and a saturable SOA has been reported at 10 Gbit/s down-stream and 2.5 Gbit/s up-stream transmission rate , the SNR as a deterministic factor has not yet been fully investigated. In this paper, the improvement on the relative intensity noise (RIN) and SNR of the down-stream data to be injection-lock the up-stream WRC-FPLD made by a gain-saturated SOA are investigated, and the effects of the excessive amplitude of the injecting down-stream data on the up-stream transmission performances at bit rate of 2.488 Gbit/s are characterized. Another gain-saturated SOA located at central office in the proposed scheme could greatly suppress the intensity noise of the spectrum-sliced ASE source. Moreover, it is observed that the SNR of the re-encoded up-stream data reveals a saturating trend with increasing down-stream ASE injection power. With a compromise between the injection power level and the SOA biased current, the ASE injection power range is provided to reach an optimized bi-directional WDM-PON communication.
2. Experimental setup
The proposed bi-directional DWDM-PON configuration with SOA based down-stream data eraser is shown in Fig. 1 . First of all, an array waveguide grating (AWG) with 200-GHz channel spacing is employed to spectrally slice the ASE source of the Erbium-doped fiber amplifier (EDFA) in central office. Before entering the external optical modulator, the noise of the channelized ASE source is greatly suppressed through a saturated SOA operated at current up to 400 mA. The ASE power input into the SOA is set as high as 0 dBm to cause the gain-saturation of the SOA. The down-stream data transmitted through 25-km SMF to ONUs is split by a 50/50 1x2 optical coupler for data decoding and remodulation. The WRC-FPLD with longitudinal mode spacing of 0.6 nm and threshold current of about 25 mA is employed as an up-stream transmitter, which exhibits the back and front facet reflectivity of 100% and 1%, respectively. Furthermore, the data stream in the branch for remodulation is completely erased by another gain-saturated SOA and is subsequently remodulated at bit rates of 2.488 Gbit/s with the non-return-to-zero (NRZ) pattern length of 223-1 in the WRC-FPLD.
When evaluating the system performance, we are more concerning on extracting the data-erasing ASE carrier from the down-stream after passing through the SOA. As shown in the right inset of Fig. 1, the on-bit (or high-level) signal experiences a lower gain as compared to that of the off-bit (or low-level) signal at SOA saturation condition. The power difference between low- and high-level is reduced through such a saturated transfer function, which results in a purified ASE carrier without sacrifice the average power during the down-stream data-erasing process. If we define the on/off extinction ratio (hereafter referred as ER) as the ratio of on-bit power to off-bit power, it is found that either using a down-stream with a larger input power or operating the SOA at lower biased current can achieve the data-erasing ASE carrier (with extremely low ER as expected). Even though, it is preferred to operate the SOA at slightly lower biased condition since the saturation power of the SOA can be greatly reduced to an acceptable level when considering the power consumption issue. The wash-out of data from ASE down-stream is the key issue deciding the re-encoded up-stream transmission performance of the down-stream ASE carrier injected WRC-FPLD. Instead of looking the level of “low” or “high” signal separately, the ER and the overshooting level are the key parameters which precisely describe the purity and the data-erasing capability of the down-stream ASE carrier after passing the gain-saturated SOA. To be re-encoded in the injection-locked WRC-FPLD transmitter, a completely data-erased ASE carrier with greatly reduced ER and overshooting level is expected to achieve better up-stream BER performance.
3. Results and discussions
3.1 Relative intensity noise and signal-to-noise ratio of the down-stream ASE source filtered by gain-saturated SOA
The SNR of a laser diode and its relationship with the RIN can be simply described as 
That is, the suppression of the RIN results in a better SNR for the down-stream signal. For a spectral-sliced, SOA-filtered ASE based and noise superimposed source as a down-stream signal, the RIN can be derived as in  considering large power near saturation and small modulation (perturbing signal). Assuming that P0 and N0 are the time-averages of the output power and equivalent carrier density associated with the noise terms, the noise and power ratios between output and input of the SOA filtered ASE source can be described as
For a given SOA, both P0 and Γg(N0) can be enlarged by increasing the input power Pin and injected current Ib, respectively. Equation (4) clearly shows that P0, Γg(N0) and Γa0 need to be as large as possible to get a lower RINout. In our case, the original RIN of the spectrally sliced ASE source is −135 dB/Hz at offset frequency below 6 GHz. Operating the SOA at 400 mA and setting the input ASE power at 0 dBm, the SNR of down-stream data is improved from 2.3 to 8.0 dB. The addition of such an SOA based noise blocker can essentially suppress the RIN of the ASE source by 2 dB at frequency below 3 GHz, and as shown in Fig. 2 .
3.2 Distinguished influence of data-erased ASE power to the SNR and BER performances of the WRC-FPLD transmitted up-stream data
The WRC-FPLD transmitted up-stream data is generated after the injection of the data-erased down-stream signal. The SNR of up-stream data can be significantly affected by changing ER and overshooting level of the data-erased down-stream signal. In the worst case with ER of 5 dB and overshooting level of −2.9 dB, the WRC-FPLD transmitted data shown in Fig. 3(a) indicates a SNR of only 4.6 dB.
By minimizing the overshooting level of the data-erased signal to −7.75 dB and reducing its ER to 2.41 dB, the SNR of the up-stream data can be improved to 6.4 dB. From Fig. 4 , it is straightforward that the data-erased down-stream data with lower ER and overshooting level can essentially improve SNR and BER performances of the up-stream transmitted data from the injection-locked WRC-FPLD. Remaining ER constant, it can be concluded that the increasing overshooting level of the data-erased down-stream data severely degrades the transmitted up-stream data quality. And the plot of BER versus receiving power clearly interprets that only when the down-stream injection with ER of smaller than 3 dB and the overshooting level of smaller than −5 dB can the demand of the data-communication criterion with a BER of <10−9 be met. Otherwise, the SNR of the WRC-FPLD up-stream transmitted data will be reduced to less than 6 dB, which seriously reduces the BER by at least three orders of magnitude at same receiving power.
The trade-off between the SOA biased current and the down-stream signal power need to be further considered. It is understood that either using a down-stream with a larger input power or operating the SOA at lower biased current can achieve the data-erasing ASE carrier (with extremely low ER as expected). In general case for obtaining high output power, the increasing SOA biased current permits us to use higher ASE injection power to improve the transmission performance. As shown in Fig. 5 , we have measured the output power response of the SOA as a function of the input data-stream power, and illustrate the on- and off-level of the output data stream after passing through the gain-saturated SOA with increasing bias. Naturally, the ER and overshooting level on the ASE carrier are reduced when enlarging the input data-stream power to enter the power saturation region of the SOA and simultaneously increasing the SOA biased current. Under complete power-saturation, it is observed that the ER of the data-stream after passing through the SOA is decreased from 1.25 to 1.1 by increasing the SOA power from 200 to 400 mA.
However, it is preferred to operate the SOA at slightly lower biased condition since the saturation power of the SOA can be greatly reduced to an acceptable level when considering the power consumption issue. The saturated power of the SOA can be slightly enlarged with increasing bias, which inevitably sets a higher criterion for the requested input down-stream ASE data power. With a slightly sacrificing ER by 0.5 dB, both the SOA bias and the down-stream signal power can be greatly reduced by 200 mA and 2 dB, respectively. In addition, another restriction is on the power handling of the components, since the biased current over 400 mA may damage the SOA and the WRC-FPLD could only tolerate the ASE injection power <-3 dBm. On the other hand, the SNR could also be improved by increasing ASE injection power, however, a SNR saturation phenomenon concurrently occurs when the saturation power of SOA is enlarged with increasing the injection power. That is, an excessively high injection power would cause SNR to saturate or even decay, as interpreted by Eq. (7) in the section 3-3 below.
3.3 SNR saturation phenomenon of the up-stream transmitted by re-encoding the data-erased ASE carrier in the WRC-FPLD
Furthermore, the SNR of the injection-locked WRC-FPLD in gain-saturation condition is considered. Originally, the SNR of WRC-FPLD under external injection is given by 
As described in , the saturated output power Psat of the WRC-FPLD is no longer a constant under external injection. With the modified carrier lifetime of τ−1 = τc−1 + (ainjPinj /Ahν) and the saturation power of Psat = Psat,0 + ΔPsat = Ahν/τca0 + ainjPinj/a0, respectively, the Eq. (6) can be rewritten as below.
In Eq. (7), Psat,0 is the saturation output power without injection and ΔPsat is the increment of the saturated output power induced by injection. For other parameters, τc denotes the carrier lifetime and ainj denotes the differential gain coefficient of the WRC-FPLD with external ASE injection, respectively. Since the saturated output power increases with the injection power, the output power response in the inset of Fig. 6 does not show obvious gain-saturation effect. In contrast, the SNR tends to saturate at higher Pinj due to the larger Psat in the denominator of the exponential term. Hence, the SNR of WRC-FPLD re-encoded ASE up-stream data can only be optimized by appropriately selecting the optimized ASE injection power range.
When comparing the experimental result of SNR with the simulated curve using Eq. (6), it is observed that the saturated SNR curve with increasing ASE injecting power is in good agreement with the modified SNR model for the injection-locked WRC-FPLD under gain-saturation condition, as shown in Fig. 6. As a result, the improvement of SNR for up-stream transmitted data is mainly attributed to the enlarged ASE injection power. It is thus mandatory to use higher power of the data-erased signal injection-locking WRC-FPLD to obtain better SNR of the up-stream transmitted data. However, the SOA biased current and the down-stream ASE transmitted data power must be increased simultaneously. With higher power of the data-erased signal injection-locking the WRC-FPLD, ER and overshooting level of the signal are also increased and decay the SNR of the up-stream transmitted data. Thus, the SOA bias is required to be enlarged to decrease both ER and overshooting level. In our experimental setup, the SOA biased is operated at 400 mA and the ASE injection power is controlled at near −3 dBm, and the SNR of the up-stream data shows the saturation characteristics under this condition. Although the power handling could be improved by a more sophisticated fabrication technique for the components, the power budget still restrains the source which can be manipulated. To achieve a BER of <10−9, the minimum ASE injection power is at least −4 dBm to achieve an SNR of larger than 6 dB when biasing the SOA at 400 mA, whereas the maximum ASE injection power is limited at −3 dBm by the power handling capability of the WRC-FPLD, as already shown in Figs. 4 and 6. In addition, the critical injection power which induces the SNR saturation effect is about −4 dBm due to the variation on saturation power of WRC-FPLD under external injection. That is, even if the WRC-FPLD could tolerate a higher injection power, the SNR would gradually saturate and even decay with enlarged external injection according to Eq. (6).
As an alternative to the intriguing WDM-PON system based on optical-comb source , there are some unique features in our proposed system needed to be addressed. Since the broadband ASE carrier is directly filtered through the AWG in our system, the wavelength of each channel is designated to fit the ITU-T standard without mismatch. The WRC-FPLD employed in our system takes the advantages of both a broadened gain spectrum (like an SOA) and a weak longitudinal-mode feature (like and FP-LD). The wider gain spectral linewidth of WRC-FPLD than that of a common FPLD is achieved by appropriately reducing the front-facet coating reflectance, which permits us to enhance the channel capacity. The weak and dense mode feature of the WRC-FPLD facilitates the injection-locking of 2-3 modes by the data-erased ASE carrier within each DWDM filtered channel, which easily avoids the wavelength mismatch caused by temperature fluctuation or cavity-length deviation. With long-cavity design, the WRC-FPLD behaves nearly like a colorless transmitter that exhibits a higher tolerance than other architectures to the wavelength drift of the injected light. Due to the relatively weak mode characteristics, the WRC-FPLD can hold more power at same injection level than the reflective SOA that is also considered as an alternative candidate of the WDM-PON transmitter. Owing to these specific features, the WRC-FPLD under DWDM channelized ASE injection is relatively a considerable candidate for serving as the universal transmitter in all DWDM channels in the proposed bi-directional WDM-PON system. The smallest WDM-PON channel spacing of 50 GHz with an AWG based DWDM has already been examined by using such an ASE injected WRC-FPLD up-stream transmitter at bit rate up to 2.5 Gbit/s (already approaching the limitation set by channel cross-talk effect). Either NRZ or RZ format can be considered to demonstrate in the proposed system and the universal WRC-FPLD can be used in any DWDM channels covering the C-band wavelength region. The channelization frequency is completely independent from the encoding bit rate, such a flexibility makes the proposed system a practical candidate comparable with the other approaches.
With the proposed all-ASE based bi-directional WDM-PON system at 2.488 Gbit/s, the down-stream data on the ASE carrier is data-erased by a gain-saturated SOA and subsequently re-encoded in the injection-locked up-stream WRC-FPLD transmitter. In the process of optimizing the transmission performance by controlling the ER of the down-stream data with the saturable SOA and the ASE injecting power, the SNR saturation characteristic of the up-stream data as a function of the injection power to the WRC-FPLD is observed. By adding the noise-filtering SOA at central office with input ASE power of 0 dBm and biased current of 400 mA, the original RIN of −135 dB/Hz for the down-stream ASE data can be reduced by 2 dB at frequency below 3 GHz, such that the SNR of the down-stream data is further improved from 2.3 dB to 8.0 dB. For upstream transmission from the WRC-FPLD after injection-locking by the SOA-bleached down-stream data in the worst case, the SNR is at least 4.6 dB by controlling the ER and the overshooting level of the data-erased down-stream signal at 5 dB and −2.9 dB, respectively. The SNR could increase to 6.4 dB by minimizing ER to 2.41 dB and overshooting level to −7.75 dB. To satisfy the communication criterion that BER is below 10−9, the ER of the down-stream data data-erased by SOA must be reduced to <3 dB and the overshooting level must be <-5 dB, or the SNR of the up-stream data will be decreased to smaller than 6 dB and BER will decrease by more than three orders of magnitude. The relationship between the SNR and the optical gain of the WRC-FPLD operated near saturation region shows that higher down-stream data power is needed to injection-lock the WRC-FPLD for better SNR. Since both ER and overshooting level increase with higher power to degrade SNR, it is mandatory to increase the biased current of SOA and the trade-off between down-stream signal power and SOA bias needs a careful consideration. However, SNR saturation is observed with ASE injecting power at −4 dBm or higher, and the injection power ranged between −7 and −3 dBm is suggested to achieve the optimized bi-directional transmission with a BER of <10−9 at SOA biased current of 400 mA. Finally, the advanced features such as the wide gain spectrum and weak mode spectrum of the WRC-FPLD are elucidated and compared with other candidates for WDM-PON bi-directional transmission.
The work was financially supported by National Science Council and Excellent Research Projects of National Taiwan University under grants NSC97-2221-E-002-055, NSC98-2221-E-002-023-MY3, NSC98-2623-E-002-002-ET, NTU98R0062-07, and 99R80301.
References and links
1. K. Iwatsuki, J. Kani, H. Suzuki, and M. Fujiwara, “Access and metro networks based on WDM technologies,” J. Lightwave Technol. 22(11), 2623–2630 (2004). [CrossRef]
2. K.-Y. Liou and G. Raybon, “Operation of an LED with a single-mode semiconductor amplifier as a broad-band 1.3-μm transmitter source,” IEEE Photon. Technol. Lett. 7(9), 1025–1027 (1995). [CrossRef]
3. D.-K. Jung, H. Kim, K. H. Han, and Y. C. Chung, “Spectrum-sliced bidirectional passive optical network for simultaneous transmission of WDM and digital broadcast video signals,” Electron. Lett. 37(5), 308–309 (2001). [CrossRef]
4. S.-M. Lee, M.-H. Kim, and C.-H. Lee, “Demonstration of a bidirectional 80-km-reach DWDM-PON with 8-Gbs capacity,” IEEE Photon. Technol. Lett. 19(6), 405–407 (2007). [CrossRef]
5. 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]
6. G.-R. Lin, H.-L. Wang, G.-C. Lin, Y.-H. Huang, Y.-H. Lin, and T.-K. Cheng, “Comparison on injection-locked Fabry–Perot laser diode with front-facet reflectivity of 1% and 30% for optical data transmission in WDM-PON system,” J. Lightwave Technol. 27(14), 2779–2785 (2009). [CrossRef]
7. Y.-H. Lin, G.-C. Lin, H.-L. Wang, Y.-C. Chi, and G.-R. Lin, “Compromised extinction and signal-to-noise ratios of weak-resonant-cavity laser diode transmitter injected by channelized and amplitude squeezed spontaneous-emission,” Opt. Express 18(5), 4457–4468 (2010). [CrossRef] [PubMed]
8. N. Kashima, “Injection-locked Fabry–Pérot laser diode transmitters with semiconductor optical amplifier for WDM-PON,” J. Lightwave Technol. 27(12), 2132–2139 (2009). [CrossRef]
9. G. P. Agrawal, Fiber-Optic Communication Systems, (Willy Inter-Science, 2002) Chap. 4–6.
10. M. Zhao, G. Morthier, and R. Baets, “Analysis and optimization of intensity noise reduction in spectrum-sliced WDM systems using a saturated semiconductor optical amplifier,” IEEE Photon. Technol. Lett. 14(3), 390–392 (2002). [CrossRef]
11. M. Yoshino and K. Inoue, “Improvement of saturation output power in a semiconductor laser amplifier through pumping light injection,” IEEE Photon. Technol. Lett. 8(1), 58–59 (1996). [CrossRef]
12. R. Maher, K. Shi, L. P. Barry, J. O’Carroll, B. Kelly, R. Phelan, J. O’Gorman, and P. M. Anandarajah, “Implementation of a cost-effective optical comb source in a WDM-PON with 10.7 Gb/s data to each ONU and 50 km reach,” Opt. Express 18(15), 15672–15681 (2010). [CrossRef] [PubMed]