Open Access
Add to BibSonomyAbstract
We have experimentally investigated the impact of codirectional Raman gains on the performance of distributed fiber Raman amplified systems. The effects of various noise sources, such as optical signal-to-noise ratio (OSNR) degradation, fiber nonlinearities and surviving channel gain variation in different Raman pumping schemes, were evaluated as a function of input power into a fiber span. For measurements, distributed Raman gain was generated by pumping the fiber span with different combinations of Raman pump power between co- and counterdirections. From the results, we found that a large counterdirectional Raman gain assisted by a small codirectional gain could improve the overall static and dynamic performance of distributed Raman amplified systems.
©2007 Optical Society of America
1. Introduction
Distributed fiber Raman amplifiers have been widely used to improve the system’s performance in high-capacity, long-haul dense wavelength-division multiplexed (DWDM) systems [1]. In order to improve the performance of Raman amplified systems efficiently; many studies have been performed to understand pros and cons of various pumping schemes, such as the copumping scheme, the counterpumping scheme, and the bidirectional pumping scheme. It was recently reported that the bidirectional pumping scheme could be a preferred choice among the three schemes from a perspective of overall static system’s performance including both optical signal-to-noise ratio (OSNR) and fiber nonlinearity limitations [2–4] although the counterpumping scheme is commonly used in most of system demonstrations. A good balance between codirectional and counterdirectional Raman gains was shown to allow for combined benefits of improved OSNR and relaxed fiber nonlinearity limitations, which cannot be obtained by use of the unidirectional pumping schemes. It has been also shown that the optimum Raman pumping scheme was a 40% copumping with a net gain of -3 dB when both amplified spontaneous emission (ASE) noise and double Rayleigh backscattering were included [4]. However, previous studies have taken into account only static characteristics, such as OSNR, fiber nonlinearities and double Rayleigh backscattering. In modern DWDM systems, WDM signals would be added/dropped frequently at optical add/drop multiplexers (OADMs) and/or optical cross-connects (OXCs). Therefore, in order to ensure the transmission quality of the signal, the dynamic gain performance of Raman amplifiers has to be addressed. Previously, it was reported that gain transients in a copumping scheme are more pronounced than in a counterpumping scheme [5]. Moreover, several gain clamping methods have been demonstrated to mitigate the gain fluctuations in fiber Raman amplifiers [6–8]. However, unlike erbium doped fiber amplifiers (EDFAs), it is not so straightforward to clamp Raman gains in a long length of optical fiber due to interchannel stimulated Raman scattering as well as pump depletion [5, 7]. Therefore, it would be desirable to design a distributed Raman amplifier with minimum gain fluctuations without using complicated gain clamping methods.
In this paper, we experimentally investigate the impact of codirectional Raman gain on the static and dynamic characteristics of distributed Raman amplifier. For the system’s static performance evaluation, the receiver sensitivity penalties of 10 Gb/s non-return-to-zero (NRZ) signal in various Raman pumping schemes are measured as a function of input power into the fiber span. Moreover, the gain variation of the surviving channel during the add/drop multiplexing of 31 out of 32 WDM signals are also evaluated in order to compare the dynamic gain characteristics in various Raman pumping schemes. From the results, we find that a large counterdirectional Raman gain assisted by a small codirectional gain could improve the overall performance of distributed Raman amplified systems.
2. Measurements and discussion
Figure 1 shows the apparatus used to measure the performance of distributed Raman amplifiers with different pumping schemes. We used thirty-two WDM channels in which the thirty-one channels were added or dropped by using an acousto-optic modulator (AOM). The signal wavelengths were ranging from 1533.47 nm (CH1) to 1558.47 nm (CH32) with a channel spacing of 100 GHz. One surviving channel (1 λ) was combined with the other thirty-one channels and transmitted through a 90-km span of non-zero dispersion shifted fiber (NZDSF) having a span loss of 21 dB (including two WDMs for both signal and pump lasers). The output power levels of 32 WDM channels were set to be almost identical by using two EDFAs. Optical attenuator ATT was also used to adjust the powers of 32 WDM channels launched into the fiber span. Distributed Raman gain was generated by pumping the fiber span with copump and counterpump modules. The copump module consisted of four Bragggrating- stabilized Fabry-Perot semiconductor pumps that were polarization-multiplexed at 1455 nm and 1465 nm. The counterpump module was a fiber Raman laser operating at wavelength of 1455 nm. The Raman gain characteristics of the surviving channel were measured by using an optical spectrum analyzer (OSA). The transmitted WDM signals were demultiplexed by using an arrayed waveguide grating (AWG) and detected using an optically preamplified receiver having an adjustable decision threshold and adjustable timing. Prior to measuring each bit-error rate (BER) curve, the decision threshold and timing were optimized at a BER of 10-8.
Fig. 1 Apparatus used for performance measurement in different distributed Raman amplifiers. Acronyms are acousto-optic modulator, AOM; variable attenuator, ATT; arrayed-waveguide grating, AWG; optical band-pass filter, OBPF, optical spectrum analyzer, OSA; receiver, Rx; and bit-error-rate test sets, BERTS.
Figure 2 shows the Raman on/off gains (measured with only one surviving channel, CH17) as a function of input power into the fiber span for eight different pumping schemes. For all measurements, the total launched power of two Raman pump modules into fiber span was kept to be around 500 mW, and then the power ratio between co- and counterpump modules was adjusted by controlling the output power of each pump module. The Raman on/off gains measured for different pumping ratios were ranging from 20.5 dB to 22 dB, which were almost equal to the span loss of transmission fiber. It can be seen from Fig. 2 that the Raman gain of counterpumping scheme (CO 0%) is kept constantly even when the input power is as high as +7 dBm. However, as the copumping ratio increases, the Raman gains start to decrease at the region of high input power. For copumping scheme (CO 100%), the Raman gain was deceased by 8 dB when the input power was changed from -15 dBm to + 5 dBm. This gain compression at high input power would be caused by the stimulated Brillouin scattering (SBS) of optical signal as well as pump depletion.
Fig. 2 Raman on/off gains as a function of input power into fiber span. The co-pumping ratio was varied from 0% to 100%.
Figure 3 shows the receiver sensitivity penalties (BER = 10-9) of one surviving channel (CH17) as a function of input power for eight different Raman pumping schemes. Only surviving channel was modulated with a 10-Gb/s NRZ pseudo-random pattern of length 231-1 using a LiNbO3 Mach-Zehnder modulator and the other thirty-one channels were not connected for these measurements. The back-to-back sensitivity of our optically preamplified receiver was measured to be -35.9 dBm at BER of 10-9. A 1.3-dB penalty was observed when the signal was transmitted over a 90-km span of NZDSF without a Raman gain. This degradation was most likely caused by a chromatic dispersion of NZDSF (8 ps/nm/km @1550 nm). The penalties shown in Fig. 3 were estimated by the differences between the sensitivities measured with and without the Raman gain. For counterpumping scheme, the penalty decreased as the input power into the fiber span increased. However, for copumping scheme, the system’s performance was improved first as the input power increased. Then, the penalty increased abruptly when the input power was as high as -5 dBm. It has been well known that the performance of Raman amplified systems would be limited by the OSNR at low input power and the Kerr nonlinearities such as self-phase modulation (SPM) at high input power [3–4]. In our single-span measurement, the best receiver sensitivity was obtained when the copumping ratio was 40% with an input power of +4.72 dBm. In order to take into account the accumulation of sensitivity penalty in a multi-span system, we calculated the ratio of output SNRs in 40-spans distributed Raman amplified system, as shown in Fig. 4. The equation in [4] was used with our system’s parameters in this calculation. From this calculation, we confirmed that the best system performance was obtained with ~40% copumping ratio even in a multi-span system when the transmission loss was equal to Raman on/off gain (i.e., a net gain was 0 dB).
Fig. 4 Ratio of output SNRs (in dB) in a 40-spans distributed Raman amplified system by taking into account the effect of ASE and double Rayleigh backscattering..
In order to compare the dynamic gain characteristics of different pumping schemes, the gain variations of CH17 during the add/drop multiplexing of 31 out of 32 WDM channels were measured as shown in Fig. 5(a). From the results, we clearly found that the higher codirectional gain increases the amount of gain fluctuation. These results agree well with the previous result [5], which found that the gain variation in Raman amplifier was determined by pump depletion level; therefore the counterpumping scheme had smaller gain variation than copumping scheme. However, one thing we have found is that a bidirectional pumping scheme with a 20% copumping ratio had almost identical gain variation with a counterpumping scheme. In both 20% copumping and counterpumping schemes, the gain variation was less than 1 dB even when the input powers of WDM channels were as high as -4 dBm. It has been well known that the gain transients in Raman amplifier depend not only on the pump depletion level, but also on the interchannel stimulated Raman scattering [5, 7]. Therefore, the shortest wavelength signal has a larger gain variation than the longer wavelength signal. Figure 5(b) shows the measured gain variation of CH1 which is the shortest wavelength in our measurement. As expected, CH1 had a larger gain variation than CH17 due to interchannel stimulated Raman scattering in both 20% copumping and counterpumping schemes. However, no difference between CH1 and CH17 was observed in a high (>40%) copumping ratio. This is because the amount of gain variation was mainly determined by the pump depletion, but not by the interchannel stimulated Raman scattering in these cases.
Fig. 5 Gain variation of (a) CH17 and (b) CH1 during the add/drop multiplexing of 31 out of 32 WDM channels in four different Raman amplifiers.
Figure 6 shows the receiver sensitivity penalties (BER = 10-9) of CH17 as a function of input power for four different Raman pumping schemes. The other thirty-one added/dropped channels (each added/dropped channel had an equal input power to CH17) were also modulated and transmitted through a 90 km of NZDSF for these measurements. For 40% and 80% copumping ratios, the best system’s performances were obtained at the input powers of 2.1 dBm and 0 dBm, respectively. However, as shown in Fig. 5(a), the gain variations of CH 17 at these input powers were 4 dB and 4.1 dB, respectively. For a 20% copumping ratio, the input power of +4 dBm provided the best system’s performance with the gain variation of 3.4 dB. It can be seen from Fig. 5(a) that the gain variation could be less than 1 dB as long as the input power would decrease to be less than -4 dBm for a 20% copumping ratio. However, at this input power level, the receiver sensitivity of a 20% copumping ratio was degraded by 0.2 dB, as compared to the input power of +4 dBm. It is evident that both sensitivity penalty and gain variation of surviving channel are accumulated in multi-span transmission systems. Therefore, the copumping ratio of bidirectional Raman amplifier has to be optimized by taking into account the system’s operating condition. For example, it would be better to decrease the copumping ratio of bidirectional Raman amplifier located at signal add/drop sites.
3. Summary
The static and dynamic performance of various pumping schemes in fiber Raman amplified systems were measured as a function of input power into a fiber span. Distributed Raman gain was generated by co-, counter- and bidirectional pumping the fiber span with different combination of copump and counterpump modules. In static performance measurements, the best system’s performance was obtained with a 40% copumping ratio. The dynamic gain characteristics of different Raman pumping schemes were also evaluated when 31 out of 32 WDM channels were add/drop multiplexed. From the results, we found that the small codirectional gain (about 20% copumping ratio) could improve the gain variation characteristics and system’s static performance, simultaneously. Therefore, we believe that the complicated gain clamping technique in distributed Raman amplifiers might be avoided by choosing the appropriate input power level and copumping ratio.
Acknowledgments
This work was supported by the University of Seoul under Grant 200605011003.
References and Links
1. J. Broamage, “Raman amplification for fiber communications systems,” J. Lightwave Technol. 22, 79–93 (2004). [CrossRef]
2. P. M. Krummrich, R. E. Neuhauser, H. Bock, W. Fischler, and C. Glingener, “System performance improvements by codirectional Raman pumping of the transmission fiber,” in proceedings of European Conference on Optical Communication 2001.
3. J. Bromage, J.-C. Bouteiller, H. J. Thiele, K. Brar, L. E. Nelson, S. Stulz, C. Headley, R. Boncek, J. Kim, A. Klein, G. Baynham, L. V. Jorgensen, L. Gruner-Nielson, R. L. Lingle Jr., and D. J. DiGiovanni, “WDM transmission over multiple long spans with bidirectional Raman pumping,” J. Lightwave Technol. 22, 225–232 (2004). [CrossRef]
4. R.-J. Essiambre, P. Winzer, J. Bromage, and C. H. Kim, “Design of bidirectionally pumped fiber amplifiers generating double Rayleigh backscattering,” IEEE Photon. Technol. Lett. 14, 914–916 (2002). [CrossRef]
5. L. L. Wang, B. C. Hwang, and L. M. Yang, “Gain transients in copumped and counterpumped Raman amplifiers,” IEEE Photon. Technol. Lett. 15, 664–666 (2003). [CrossRef]
6. C. -J. Chen, J. Ye, W. S. Song, Y.-W. Lu, M.-C. Ho, Y. Cao, M.J. Gassner, J. S. Pease, H.-S. Tsai, H. K. Lee, S. Cabot, and Y. Sun, “Control of transient effects in distributed and lumped Raman amplifiers,” IEE Electron. Lett. 37, 1304–1305 (2001). [CrossRef]
7. P. Kim and N. Park, “Semianalytic dynamic gain-clamping method for the fiber Raman amplifiers,” IEEE Photon. Technol. Lett. 17, 768–770 (2005). [CrossRef]
8. S. H. Chang, H. S. Chung, H. J. Lee, and K. Kim, “Suppression of transient phenomena in hybrid Raman/EDF amplifier,” IEEE Photon. Technol. Lett. 17, 1004–1006 (2005). [CrossRef]
