Open Access
Add to BibSonomyAbstract
A novel scheme based on 40Gb/s vestigial sideband modulation for optical payload and label multiplex and separation in all optical label switching (AOLS) networks is firstly proposed and experimentally demonstrated. The payload is combined and separated with wavelength labels by optical filters. The experiment results show that after label separation, the power penalties of payload and label are both very little. The influence of the wavelength difference between label and payload is also discussed. The power penalty of payload can be less than 1dB as long as the wavelength difference is larger than 0.1nm. This scheme highly reduces the channel bandwidth of payload and label and is proposing to be used in future optical Internet.
©2005 Optical Society of America
1. Introduction
The next generation optical network needs to support packet routing and forwarding operations at high wire rates to ensure the future Internet traffic. All-optical label switching (AOLS) is a potential technique which is introduced to fully use the bandwidth capacity in optical networks and the packet operations are mostly carried out in the optical layer [1]–[2]. The key issue in the AOLS approach is the method of coding the optical label onto the packet. It directly determines the structure and the performance of the optical core router as well as the channel bandwidth efficiency and the transmission quality of the packet and its label. The optical subcarrier multiplexing (SCM) and wavelength labeling are important techniques to multiplex the payload and the label. Both of them can preserve the transparency of optical payload and offers the low bit rates of the label signal [3]–[5]. But in order to preserve the payload when the label is erased, it is necessary to keep a protection band between the label and payload in frequency domain. This will increase the signal bandwidth and will decrease the spectral efficiency which is a critical issue in wavelength division multiplexing (WDM) systems. When the bit rate goes up to 40Gb/s, there are some modified SCM methods to be used. A method using optical carrier suppression and separation is demonstrated in 10Gb/s and 40Gb/s [6]–[7]. With wavelength conversion technique, multi-node label swapping is also shown in [8]. However, the bandwidth of payload and label is still too large for 40Gb/s WDM system and the subcarrier frequency is 30GHz which may cause some nonlinear distortions in radio frequency (RF) domain.
In this letter, a novel scheme using vestigial sideband modulation (VSB) is given out to realize 40Gb/s all-optical label switching with optical wavelength labeling for the first time. The payload is modulated to be VSB signal by optical filtering and the label signal in another wavelength is added on the VSB payload. So the total bandwidth of payload and label is much less than the traditional SCM and wavelength labeling schemes and there is less crosstalk between them. The payload and label can be combined and separated easily using optical filters. The influence of optical label erasing induced by filters is also investigated in this letter.
2. Principle
The principle of this scheme is shown in Fig. 1. The payload signal is modulated onto a continuous wave with the wavelength λ1, then after optical filtering it becomes the vestigial sideband signal. At the same time, the label signal is modulated onto another continuous wave with the wavelength λ2 close to λ1. The payload and label are kept in different wavelengths in order to be combined together for transmission and can be easily separated by an optical filter. The main advantage of this scheme is that the bandwidth of labeled optical signal is much narrower than the traditional SCM and wavelength labeling schemes. Besides, this scheme is also independent of the payload and label formats.
The wavelength difference between the VSB payload and the label is the key parameter, which will influence the system performance especially when the label is separated from the VSB payload. If the label wavelength λ2 is close enough to the carrier wavelength λ1, the payload signal quality will deteriorate. That is because the label may overlap the vestigial band of payload which will cause the intraband crosstalk. Furthermore, when the label is separated from payload, the optical filter ramp will distort the payload signal. The narrower the wavelength difference between the label and the payload, the greater this distortion is. On the other hand, if the label wavelength is far from the carrier, the channel bandwidth will increase. So there is a trade-off between the signal quality and channel bandwidth. The choice of label wavelength becomes very important, which will be discussed later.
For easy realization, we use tunable optical filter to generate vestigial sideband signals. Compare to nonreturn-to-zero (NRZ) format, carrier-suppressed return-to-zero (CSRZ) is suitable for vestigial sideband modulation and the VSB-CSRZ format also has the good performance for optically-routed networks [9]. So in our scheme the payload signal is modulated as the CSRZ format firstly and then filtered to be vestigial sideband CSRZ (VSB-CSRZ) format, while the label is modulated as traditional NRZ format to combine with the payload.
3. Experiment and results
Fig. 2. Experiment setup. CWL: continuous wavelength laser, T-CWL: tunable CWL, MZM: Mach-Zender modulator, TOF: tunable optical filter, OC: optical coupler, OBPF: optical bandpass filter, Cir: circulator, T-FBG: tunable FBG, BERT: BER tester.
The experiment setup of our proposed scheme is shown in Fig. 2. In this experiment, optical signal from the continuous wavelength laser (CWL) at 1557.36nm is modulated by a Mach-Zender modulator (MZM1) biased at Vπ/2 to generate the 40Gb/s non-return-to-zero (NRZ) code. And the second MZM (MZM2) is biased at the minimal intensity-output point and driven with a 20GHz sine-clock signal. The phase deviation θ of the two modulator arms equals to π. Thus the output of MZM2 is 40Gb/s CSRZ code. Then the payload passes after a tunable optical filter (TOF) to become VSB-CSRZ code. The center wavelength of the optical filter is tuned to 1557.52nm in order to filter out one of the main frequency peaks of the CSRZ code. The 3dB bandwidth of TOF is 0.3nm and 20dB bandwidth is 0.5nm.
Another optical signal from a tunable CWL (T-CWL) at wavelength 1557.26nm is sent into MZM3. The label is modulated as NRZ code with 1.25Gb/s label signal. Then the payload and label are combined together by a 3dB coupler for transmission. At the receiver, we use a circulator (Cir) and a tunable fiber Bragg grating (T-FBG) to separate the payload and label. The FBG is exactly tuned to have a reflection peak wavelength of 1557.26 and 3dB bandwidth is 25GHz. The label is reflected by FBG and detected while the payload passes through the FBG and is detected by 43Gbps optical receiver with an optical preamplifier.
Fig. 3. Optical Spectra, the resolution of all is 0.01nm (a) CSRZ, (b) VSB-CSRZ combined with label, (c) payload passing through FBG, (d) label separated by FBG reflecting.
Fig. 4. Measured eye diagrams (a) VSB-CSRZ back-to-back (b) VSB-CSRZ after label separated (c) label back-to-back (d) label after separation.
After EDFA there is an optical bandpass filter (OBPF) to block the amplified spontaneous emission (ASE) noise. The bandwidths of the receivers for the payload and label signals are 30GHz and 1GHz respectively.
Figure 3 shows the optical spectra of different sections in our scheme. Figure 3(a) is the optical spectra of VSB-CSRZ after TOF. One of CSRZ tones is suppressed over 25dB. Figure 3(b) shows the optical spectra of VSB-CSRZ combined with label after OC. The peak powers of label and payload are almost the same which ensure the received signal qualities. Figure 3(c) shows the spectra of separated payload signal passing through the FBG, measured just before the EDFA. The label signal is suppressed over 30dBm. And Fig.3 (d) is the spectra of separated label signal reflected by FBG, the payload signal is suppressed over 15dBm. Figure 4 (a) shows the eye diagram of the VSB-CSRZ signal after TOF and Fig. 4(b) shows the eye diagram of VSB-CSRZ signal after label separation. Figure 4(c), 4(d) show the measured eye diagrams of optical label back-to-back and after separation. The payload and label distortions induced by FBG are both very small. The BER performance of the proposed technique is shown in Fig. 5. The back-to-back (BTB) VSB signal BER performance was measured as a reference for comparison. After FBG separation, the power penalties of both payload and label are less than 0.5dB (at BER=10-12). This result shows that this scheme is suitable for 40Gb/s optical networks.
Compare to traditional SCM and wavelength labeling schemes, our scheme can highly reduce the optical transmission bandwidth. But if the label wavelength is close to the payload carrier wavelength, the label will impair the payload signal and may cause great power penalty for reception, which is discussed in chapter II. The influence of the wavelength difference between the label and payload is also investigated in our experiment by tuning the label wavelength and the FBG reflection peak from 1557.20nm to 1557.30nm. Figure 6 shows the payload BER curves versus the label wavelengths. It’s shown that the payload power penalty (BER=10-12) at λ2=1557.22nm is the smallest and the penalty at λ2=1557.30nm is much greater than the others. When the label wavelength is much close to the carrier wavelength, e.g., 1557.30nm, the power penalty is above 8dB, which is a bad case for payload reception. If the label wavelength is shorter than λ1-0.1nm, which is 1557.26nm in our experiment, the separated payload power penalty is less than 1dB which is a good performance. Therefore there is a critical value of wavelength difference of payload and label, which is 0.1nm in our experiment. When the wavelength difference is longer than 0.1nm, the separated payload power penalty will increase dramatically.
Considering different channel bandwidths, the payload power penalties are shown in Fig. 7. The channel bandwidth is defined as the 20dB bandwidth from the peak power in spectra. From Fig. 7 we can see that when the wavelength difference is 0.16nm, the power penalty is above 8dB. However this case has the narrowest channel bandwidth of 0.42nm. And the payload power penalties are kept below 1dB as long as the channel bandwidth is larger than 0.46nm. So the least channel bandwidth of payload and label is 0.46nm. Furthermore, this scheme can be employed in 160Gb/s optical networks, and the 20dB channel bandwidth may be less than 0.8nm which cam maintain the traditional DWDM channel spacing paths and devices.
In summary, this scheme has many advantages. 1) It highly reduces the bandwidth of optical signal. For 40Gb/s systems the efficient bandwidth is just less than 50GHz which is the best frequency efficiency as we know. 2) The label wavelength can be close to the carrier wavelength and can be modulated in various formats in low speed, e.g. 1.25Gb/s to be used in 40Gb/s optical networks which can not be done in traditional SCM and wavelength labeling schemes. 3) The payload and the label are generated separately and combined by optical coupler, and separated by optical filters, so there is little crosstalk between the payload and label. We also discussed the impairment induced by the various label wavelengths; the experiment results show that at range below 1557.26nm, the power penalty of the separated payload signal is less than 1dB which is suitable for 40Gb/s optical networks.
4. Conclusion
A novel optical label scheme with vestigial sideband payload is proposed in this paper and experimentally demonstrated. The experiment results show that little power penalty is induced after the separation of payload and label all-optically. This scheme highly reduces the bandwidth of optical payload and label and has good spectral efficiency. The crosstalk between label and payload is very little. So it is promising to realize 40Gb/s or even higher bit rates all-optical label switching networks with low speed optical labeling.
Acknowledgment
This work is supported by National Nature Science Foundation of China (NSFC) under Grant 90104003 and the National High-tech R&D Program of China under Grant 2002AA122031.
Reference and Links
1. D. J. Blumenthal, B. E. Olsson, G. Rossi, T. E. Dimmick, L. Rau, M. Masanovic, O. Lavrova, R. Doshi, O. Jerphagnon, J. E. Bowers, V. Kaman, L. A. Coldren, and J. Barton, “All-optical label swapping networks and technologies,” J. Lightwave Technol. 18, 2058–2075 (2000). [CrossRef]
2. Tarek S. El-Bawab and Jong-Dug Shin, “Optical Packet Switching in Core Networks: Between Vision and Reality,” IEEE Communication Magazine 40, 61–65 (2002). [CrossRef]
3. Y. M. Lin, W. I. Way, and G. K. Chang, “A novel optical label swapping technique using erasable optical single-sideband subcarrier label,” IEEE Photon. Technol. Lett. 12, 1088–1090 (2000). [CrossRef]
4. Z.S. Jia, M.H. Chen, and S.Z. Xie, “Label erasing employing Lyot-Sagnac filter,” Electron. Lett. 38, 1563–1564(2002). [CrossRef]
5. Nan Chi, Jianfeng Zhang, and Palle Jeppesen, “All-Optical Subcarrier Labeling Based on the Carrier Suppression of the Payload,” IEEE Photon. Technol. Lett. 15, 781–78 (2003). [CrossRef]
6. Jianjun Yu and Gee-kung Chang, “A Novel Technique for Optical Label and Payload Generation and Multiplexing Using Optical Carrier Suppression and Separation,” IEEE Photon. Technol. Lett. 16, 320–322 (2004). [CrossRef]
7. G.K. Chang and J. Yu, “40 Gbit/s payload and 2.5 Gbit/s label generation using optical carrier suppression and separation,” Electron. Lett. 40, 442–444 (2004). [CrossRef]
8. Jianjun Yu, Gee-kung Chang, and Qimin Yang, “Optical Label Swapping in a Packet-Switched Optical Network Using Optical Carrier Suppression, Separation, and Wavelength Conversion,” IEEE Photon. Technol. Lett. 16, 2156–2158 (2004). [CrossRef]
9. A. Agarwal, S. Chandrasekhar, and R.-J. Essiambre, “VSB-CSRZ for Spectrally Efficient Optically-Routed Networks,” ECOC 2004 Proceedings, vol.3, Paper We3.4.4.
