We demonstrate that WDM burst signals are amplified with suppressed gain transients in cascaded erbium-doped fiber amplifiers (EDFAs) with fast electrical feedback control. In a recirculating loop experiment, the gain excursion is suppressed to 5.1 dBp-p after 45 EDFAs while the excursion is 28 dBp-p after 5 EDFAs without the gain control. Spectral hole burning is a main limiting factor for the transient suppression. The effect of the gain control is also verified in a transmission experiment at 10 Gb/s.
© 2006 Optical Society of America
Optical burst switching (OBS) is one of the technologies which enable flexible and efficient optical network [1, 2]. In an OBS network, optical amplifiers will be needed at WDM sections to compensate for the loss at optical nodes and fibers. One of the issues for realizing OBS is the transient response of optical amplifiers as optical input power varies with turning on and off the bursts. The transient response is caused by the inherent gain dynamics of the optical amplifiers operating in the saturation region [3, 4].
Several approaches have been proposed to handle this issue, such as using optical feedback to clamp the amplifier gain [5, 6], adding dummy optical signal at burst intervals to have constant input power [7, 8], or using electrical feedforward and feedback to have constant gain [9, 10]. We have proposed a solution using fast electrical feedback and demonstrated that WDM burst signals were amplified with suppressed transients in a single erbium-doped fiber amplifier (EDFA) [11, 12]. However, the gain excursion is considered to accumulate as optical bursts pass through a number of EDFAs.
In this paper, we investigate the gain variation of the cascaded EDFAs with the transient control. In a recirculating loop experiment, WDM optical bursts are transmitted through up to 45 EDFAs. Although the gain excursion accumulates with a number of EDFAs, it is suppressed to 5.1 dBp-p after 45 EDFAs. The effect of spectral hole burning is discussed. The transmission performance at 10 Gb/s is also demonstrated.
2. EDFA configuration
Figure 1 shows the block diagram of the optical amplifier used in the experiment. The optical configuration is a conventional EDFA pumped at 980 nm. A gain flattening filter (GFF) is used to have a flat gain spectrum at the optimum gain of 23 dB. The gain flatness is about 0.1 dB over the signal wavelength range of 1546.12 nm to 1550.92 nm. The amplifier gain is monitored with two photodiodes (PDs) at the input and output port of the EDFA.
The transient control is achieved in automatic gain control (AGC) mode to have the target gain of 23 dB. The gain error is fed back to the pump current with a digital control unit where proportional-integral control is performed [11, 12]. The amount of the feedback is optimized to have a stable operation of the EDFA. The response time of the feedback is less than 1 µs, including signal processing and the pump laser drive. In calculating the gain error, the power of amplified spontaneous emission (ASE) is taken into account. For comparison, the EDFA is operated also in automatic current control (ACC) mode where the pump current is kept constant.
3. Experimental Setup
Figure 2 shows the experimental setup. Four DFB-LDs with a spacing of 200 GHz are used as light sources. Optical bursts are generated with acousto-optic modulators (AOMs). The multiplexed optical bursts are launched into a recirculating loop consisting of 5 EDFAs, single-mode optical fibers (SMFs), dispersion compensating fibers (DCFs), a variable optical attenuator (VOA), and an optical fiber coupler (CPL). The length of the loop is about 380 km and the propagation time is 1.88 ms. Two AOMs (SW1 and SW2) are controlled by function generators (FG1 and FG2) so that optical bursts pass through the loop 9 times. The waveforms of the bursts after the circulation are measured with photodiodes.
Figure 3 shows the waveforms of the input optical bursts. Ch.1 and Ch.3 are modulated with a 5 kHz square wave. The duty ratio is 50 % and the burst length is 100 µs. Ch.2 is modulated at 3 kHz and the burst length is 167 µs. The burst length of Ch.4, which is determined by SW1, is 1.61 ms. In order to maximize the input power variation, the timing of turning on/off the bursts is synchronized. All channels are turned on simultaneously at t=0 and turned off at t=1.61 ms. Since the propagation time of the loop is 1.88 ms, each EDFA has an idle interval of 0.27 ms, when no signal is input.
The input power into the 1st EDFA is adjusted to -17 dBm per channel during the burst period. The EDFAs are operated either in AGC mode or ACC mode. The loss of the VOA is adjusted so that the net gain of the loop is ~0 dB when EDFAs are in AGC mode. In ACC mode, the pump current is adjusted to have 23 dB gain for the continuous input power of -17 dBm per channel. In order to measure the transmission performance, Ch.4 is further modulated with 10 Gb/s, 231-1 pseudo-random binary sequence by a LiNbO3 modulator. A function generator (FG3) controls the gate timing of the bit-error rate tester (BERT) which measures BER during the burst period of Ch.4 after each circulation. The receiver used in the measurement is a conventional one and is not optimized for receiving optical bursts. The input power to the receiver was adjusted to -3 dBm.
4. Results and Discussion
Figures 4 (a) and (b) shows the measured waveforms of all channels after the 1st circulation (5 EDFAs) with ACC and AGC mode, respectively. With ACC mode, a large optical surge occurs at t=0, when all channels are turned on. The peak power of the surge, not shown in the Fig. 4 (a), is 16 times as high as the average power. In addition to the surge at the rising edge of the burst, the waveform of a channel is distorted due to the cross gain modulation. As a result, the gain excursion of Ch.4 becomes about 28 dBp-p. For ACC mode, no further optical circulation was performed in order to avoid larger optical surge and potential damage to the components. With AGC mode, the gain excursion is suppressed to <1 dBp-p.
Figures 5 (a) and (b) show the measured waveforms with AGC mode after the 7th circulation (35 EDFAs) and the 9th circulation (45 EDFAs), respectively. The gain excursion accumulates with the number of circulations. The excursion for Ch.4 occurs when the other channels are turned on and off.
The largest overshoot occurs at t=1000 µs, when the bursts of Ch.1, 2, and 3 are turned on simultaneously on top of Ch.4. The overshoot at t=1000 µs is larger than the one at t=0, when all the channels are turned on. This is explained by the spectral hole burning (SHB) [13, 14]. The gain spectrum of the EDFA depends on the allocation of active channels because a hole appears around the signal wavelength. When all the channels are present, the gain spectrum is almost flat. However, when only Ch.4 is active, the gains at the other channels are higher than the target gain because they are not depleted by the hole. This is reflected in the ASE spectra around 1548 nm shown in Fig. 6. The ASE power with Ch.4 only is higher than the one with all the channels. As a result of the higher gain, the overshoot occurs at t=1000 µs. In order to verify the mechanism, another experiment was made where Ch.1, 2, and 3 were turned on after a long interval of 500 µs. The result is shown in Fig. 7. The amounts of overshoot are 4.1 dB, 3.9 dB, and 2.6 dB for Ch.1, 2, and 3, respectively. Ch.1 shows the largest overshoot because it has the farthest wavelength from Ch.4 and the least gain depletion due to SHB.
Figure 8 shows the amount of gain excursion at Ch.4 after each circulation. While the excursion is 28 dBp-p with ACC mode after 5 EDFAs, it is suppressed to 5.1 dBp-p with AGC mode after 45 EDFAs owing to the fast electrical feedback. The excursion in dB increases almost linearly with the number of EDFAs.
The measurement was also made with changing the EDFA input power from -17 dBm to -20 dBm per channel. In this case, the excursion was reduced to 3.3 dBp-p after 45 EDFAs. This is because the transient response is more efficiently suppressed with AGC at lower EDFA output power owing to slower gain relaxation.
Figure 9 shows the measured BER at Ch.4 as a function of the number of EDFAs. With ACC mode, BER is higher than 10-3 after 5 EDFAs because of the large gain excursion. With AGC mode, BER is lower than 10–13 up to the 3rd circulation (15 EDFAs). For comparison, BER was also measured with CW operation of Ch.1, 2, and 3, where no cross gain modulation was present. The main cause of the bit error is not the gain excursion but the noise of the EDFA because the measured BER is similar between the two cases although the differences of the BER in each circulation have a statistical variation.
In summary, we have investigated the gain variation of the cascaded EDFAs with WDM burst signals in a recirculating loop experiment. The automatic gain control (AGC) is achieved with a fast electrical feedback loop. While the gain excursion is 28 dBp-p after 5 EDFAs without the gain control, it is suppressed to 5.1 dBp-p after 45 EDFAs with AGC. Spectral hole burning is a main limiting factor for the transient suppression. The effect of the AGC is also verified in a transmission experiment at 10 Gb/s.
This work was supported by the Strategic Information and Communications R&D Programme (SCOPE) of the Ministry of Internal Affairs and Communications of Japan.
References and links
1. M. Yoo, M. Jeong, and C. Qiao, “A high speed protocol for bursty traffic in optical networks,” SPIE’s All-Optical Communication Systems: Architecture, Control and Network Issues , 3230, 79–90, (1997).
2. J. Turner, “Terabit burst switching,” J. High Speed Networks 8, 3–16, (1999).
3. A. Bononi and L. A. Rusch, , “Doped-fiber amplifier dynamics: a system perspective,” J. Lightwave Technol. 16, 945–956, (1998). [CrossRef]
4. L. Tancevski, A. Bononi, and L. A. Rusch, “Output power and SNR swings in cascades of EDFA’s for circuit- and packet-switched optical networks,” J. Lightwave Technol. 17, 733–742, (1999). [CrossRef]
5. M. Karasek, A. Bononi, L. A. Rusch, and M. Menif, “Gain stabilization in gain clamped EDFA cascades fed by WDM burst-mode packet traffic,” J. Lightwave Technol. 18, 308–313, (2000). [CrossRef]
6. M. Karasek, M. Menif, and L. A. Rusch, “Output power excursions in a cascade of EDFAs fed by multichannel burst-mode packet traffic: experimentation and modeling,” J. Lightwave Technol. 19, 933–940, (2001). [CrossRef]
7. T. Shiozaki, M. Fuse, and S. Morikura, “A study of gain dynamics of erbium-doped fiber amplifiers for burst optical signals,” in ECOC 2002, Copenhagen, Denmark, 2002, Paper P4.02.
8. A. V. Tran, C. J. Chae, R. S. Tucker, and Y. J. Wen, “EDFA transient control based on envelope detection for optical burst switched networks,” IEEE Photon. Technol. Lett. 17, 226–228, (2005). [CrossRef]
9. C. Tian and S. Kinoshita, “Analysis and control of transient dynamics of EDFA pumped by 1480- and 980-nm lasers,” J. Lightwave Technol. 21, 1728–1733, (2003). [CrossRef]
10. C. Tian and S. Kinoshita, “Elimination of power and OSNR swings of EDFA in the amplification of burst WDM traffic,” in Technical Digest of 9th OptoElectronics and Communication Conference/3rd International Conference on Optical Internet (2004), pp. 500–501.
11. K. Okamura, E. Otani, T. Yoshikawa, T. Uchino, M. Fukushima, and N. Kagi, “Optical burst amplification using EDFA with fast feedback control,” in Technical Digest of Optical Fiber Communication Conference (2005), Paper OTuN2.
12. E. Otani, T. Okaniwa, K. Okamura, T. Yoshikawa, T. Uchino, M. Fukushima, and N. Kagi, “Optical burst amplification using EDFA with a fast feedback control for WDM Network,” in Proceedings of The 18th Annual Meeting of the IEEE Lasers & Electro-Optics Society, (2005), Paper TuO3.
13. J. W. Sulhoff, A. K. Srivastava, C. Wolf, Y. Sun, and J. L. Zyskind, “Spectral-hole burning in erbiumdoped silica and fluoride fibers,” IEEE Photon. Technol. Lett. 9, 1578–1579, (1997). [CrossRef]
14. E. Rudkevich, D. M. Baney, J. Stimple, D. Derickson, and G. Wang, “Nonresonant spectral hole-burning in erbium-doped fiber amplifiers,” IEEE Photon. Technol. Lett. 11, 542–544, (1999). [CrossRef]