This manuscript has been accepted for rapid publication in Optical Engineering Letters. This version has not been edited or formatted for publication; the final printed article may differ due to copyediting of the manuscript. The final version of the manuscript is scheduled to appear in the July 2004 issue of Optical Engineering (Volume 43,Number 7)Optical label switching by using DPSK and in-band sub-carrier multiplexing modulation formatI. Tafur Monroy, J. J. Vegas Olmos, A. M. J. Koonen, F. M. Huijskens, H. de Waardt, and G-D. KhoeEindhoven University of TechnologyCOBRA InstituteP.O. Box 513EH 11:08, 5600 MB EindhovenThe NetherlandsE-mail: i.tafur@tue.nl________Manuscript OEL 04002 received January 9, 2004; revised manuscript received March 5, 2004; accepted for publication March 9, 2004; appeared online March 10, 2004.© 2004 Society of Photo-Optical Instrumentation EngineersOptical label switching by using DPSK and in-band sub-carrier multiplexing modulation formatI. Tafur Monroy, J. J. Vegas Olmos, A.M.J. Koonen, F.M. Huijskens,H. de Waardt, G-D. KhoeCOBRA Institute, Eindhoven University of Technology,P.O. Box 513, EH 11:08, 5600 MB Eindhoven, The NetherlandsFax: +31-40-2455197, E-mail: i.tafur@tue.nlAbstractOptical label switching based on combined DPSK modulation and sub-carrier multiplexing is experimentally demonstrated at 10 Gbit/s DPSK encoded payload and 100 Mbit/s AM-SCM labeling. This scheme is spectral efficient and robust to fiber dispersion.I.INTRODUCTIONOptical label switching (OLS) has received much attention recently as a promising technique for optical networking. OLS reconciles the mismatch between high speed data transmission over wavelength division multiplexing (WDM) links and the limitations of current technology for optical memory, buffering, and optical signal processing for look-up table routing decisions. By using short fixed-length labels the core nodes of the network forward/switch packets fast and efficiently while keeping the payload data entirely in the optical domain. OLS also offers the advantage of being protocol transparent, compatible with legacy and emerging networking technologies by adopting the Generalized Multi-Protocol Label Switching (GMPLS) framework for a unified control plane [1]. One important issue in OLS networks is the technique used to label optical signals. Our previous published work has been focused on the combined modulation scheme [2]. However, other techniques have been proposed until now, among those, sub-carrier multiplexing (SCM) with frequencies above the base-band spectrum of the payload data, encoded in intensity modulation [3] or the use of combined angle and intensity modulation formats [4]. Our proposed labeling technique combines angle modulation and sub-carrier multiplexing. Namely, the payload data is encoded in DPSK modulation format and the label information is transported by using SCM techniques. This scheme shows several advantageous features for optical networking and data transmission. For instance, the sub-carrier frequency may be chosen to be in the same frequency band than the payload data, which is appropriate for spectral efficiency while keeping the complexity of the RF signal generation/detection low. The sub-carrier signal may be detected by simple photodetection and its is not affected by the DPSK information as the photodiode does not react to the optical signal phase. DPSKmodulation has received much attention recently as a spectrally efficient format for high bit rate transmission. It has been shown to be robust against chromatic dispersion and polarization mode dispersion (PMD), and cross-phase modulation (XPM) effects during transmission along optical fibers [5]. In fact, record high capacity and link reach optical transmission has been recently reported by using DPSK modulation format [6]. Moreover, if a different value for the sub-carrier frequency in each wavelength channel is used, by tapping a small percentage of the optical power out of a fiber link, core nodes have access to the label information of all the channels transported on a fiber.II.DPSK/SCM LABELING SCHEMEThe label information is imposed into a sub-carrier frequency while DPSK modulation is used for transmitting the high bit rate payload data. The label information, commonly at a lower bit rate than the payload data, can be imposed onto the sub-carrier by using an arbitrary modulation format. The modulated SCM signal can be imposed into the optical carrier by analog, direct laser modulation, or by using an external modulator. At the receiver side, a portion of the signal power is detected and the SCM signal is recovered. The payload is DPSK encoded by using an optical phase modulator. At the receiver side, DPSK data is recovered by using a standard one-bit delay Mach-Zehnder interferometer (MZI) as demodulator followed by photodetection. The use of high bit rates favours DPSK by relaxing the tolerances for instability of the MZI demodulator and phase noise than compared to the case of operation at lower bit rates [5].III.EXPERIMENTAL SET-UPIn the experimental setup shown in Fig. 1, a distributed feed-back (DFB) laser source with an integrated EAM modulator is used for inserting a RF signal at 2 GHz with a 100 Mbit/s amplitude modulation onto the optical carrier operating at the 1555.4 nm wavelength. By using a low frequency sub-carrier we avoid the use of optical filtering for recover the SCM. Optical filtering for the SCM create constraints concerning to the alignment of the signal and the filter. The DFB section is biased at 90 mA while the Electro-Absorbtion Modulator (EAM) section is reversed biased at 50 mV. A LiNbO3 phase modulator is used to impose DPSK modulation at 10 Gbit/s using a 27–1 PRBS pattern. We employed a single photodetector receiver for DPSK detection. In Fig. 1 is shown the eye pattern of the recovered DPSK signal (without transmission over the SOA-modules shown in Fig. 1) with its corresponding Q factor. The received optical power level for this measurement was set to –18 dBm and no optical or electrical amplification was used. When no SCM is used a Q factor of 6.6 (Fig. 1a) is measured suggesting an achievable bit-error rate (BER) in the order of 10–9. As the modulation index of the SCM signal increases from m=4.0% to m=7.6% the Q value for the DPSK signal decreases from 6.6 to 5.9 for a 2 GHz RF with 100 Mbit/s AM label data. Although, a degradation of the measured Q is observed at a received optical power of –18 dBm, the receiver performance can be improved by properly selecting the modulation index of the RF signal, modulation format for the label data such as BPSK or FSK instead of AM, and by improving the DPSK receiver design including balanced detection and electrical filtering.In OLS networks, core nodes perform the function of label reading, erasure of old label and insertion of a new label to deliver the burst/packet to the next hop. Label erasure of the SCM signal may be performed by using the gain saturation effect in a SOA with a speed in the order of the sub-carrier frequency. However, during the label erasure process, the phase information should be preserved for proper DPSK detection. For the label insertion operation, an EAM may be used, similarly to the one used in our experiment for label generation. Wavelength conversion is often required in core nodes and it may be performed by phase-maintaining wavelength conversion techniques based on four-wave mixing effects. In this way the integrity of the DPSK encoded data is preserved. We have performed label erasure of a100 Mbit/s AM-SCM signal at 2 GHz by using three-SOA integrated circuit as shown in Fig. 1. The use of an SOA as eraser was predicted and mathematically developed in [7]. The net fiber-to-fiber gain of the SOA integrated chip is 6 dB. The DPSK/SCM signal after transmission over the SOA-chip is filtered for remove the SOA noise by an optical band-pass filter with a 0.73 nm 3-dB bandwidth and subsequently detected to study the label erasure capability of the SOAs and to assess the effect of the SOA introduced phase noise on the DPSK payload data. The results presented in Fig. 2 are all measured at an average received optical power of –6.2 dBm and without electrical filtering. In Fig 2a is shown the measured DPSK eye diagram with no superimposed RF signal. Fig 2.b shows the DPSK eye diagram when the 2 GHz RF is present and Fig. 2.c the case for a 100 Mbit/s AM-SCM signal with a value m=7.6%.Although the eye diagram presented in Fig. 2b, and 2c are well open, we can observe that the amplitude variations introduced by AM-SCM result in a slight closure of the eye diagram. This is attributed to the AM-SCM imposed intensity envelope on the DPSK signalas well as to added phase transitions by the SOA in response to changes in its input signal amplitude. A more detailed study should be done to quantify this effect on the performance of the proposed labeling scheme. In Fig. 2e is presented the pattern of the original 100 Mb/s AM-SCM signal. In Fig. 2f is shown the effect of transmission through the three-SOA module. As it can be seen, the AM-SCM signal is effectively suppressed. In Fig. 2d is shown the RF spectrum of the original AM-SCM and its 'erased' version after SOA transmission.IV.CONCLUSIONSWe have presented an optical label switching technique based on the use of DPSK modulation for the high-rate payload data and SCM for the label information. We have experimentally demonstrated its feasibility for 10 Gbit/s DPSK and 100 Mbit/s AM-SCM at 2 GHz label data, including label erasure by transmission through a three-SOA integrated circuit. The SOA effectively suppresses the AM-SCM signal while introducing no substantial degradation to the DPSK receiver performance. The proposed DPSK/SCM labeling scheme presents advantageous key features such as spectral efficiency, robustness against chromatic dispersion and PMD, XPM impairments.REFERENCES[1] C. Qiao, and Y.Chen, "The potentials of optical burst switching", In proceedings ofOFC, paper TuJ5, Vol. 1, 219-220, 2003.[2] Sulur, Ton Koonen, Jean Jennen, Huug de Waardt, Idelfonso Tafur Monroy, GeertMorthier, “Optical label coding by using combined ASK, DPSK and FSK modulation formats”, In proceedings of ONDM 2002.[3] D. J. Blumenthal, et al."All-optical label swapping networks and technologies",Journal of Lightwave Technology, Vol. 18, No. 12 , 2058–2075. 2000[4] N. Chi, L. Xu, L. J. Christiansen, K. Yvind, J. Zhang, P. V. Holm-Nielsen, C.Peucheret, C. Zhang and P. Jeppesen, "Optical label swapping and packet transmission based on ASK/DPSK orthogonal modulation format in IP-over-WDM networks", In proceedings of OFC, paper FS2, Vol. 2, 792-793, 2003.[5] M. Rohde,et al., “Robustness of DPSK direct detection transmission format instandard fibre WDM systems”, Electronics Letters,Vol. 36 Issue: 17 ,1483–1484, 2000.[6] H. Bissessur, et al., ”1.6 Tbit/s (40x40 Gbit/s) DPSK transmission over 3x100 km ofTeraLight fibre with direct detection”,IEEE Electronics Lett., Vol. 39, Issue 2, 192-193, 2003.[7] M. T. Hill, et al., “Carrier recovery time in semiconductor optical amplifiers thatemploy holding beams”, Opt. Letters, nr.18, pp 1625-1627, 2002.Figure captions :Figure 1: Block diagram of the experimental set-up for DPSK/SCM labeling a) measured DPSK eye diagram for 10 Gbit/s payload data with no RF b) with 100 Mbit/s AM-SCM and m=4.0%; and c) m =7.6%.Figure 2: DPSK eye pattern after transmission through a 3 SOAs. a) without superimposed RF signal; b) with a 2 GHz RF; c) with a 2 GHz RF with 100 Mbit/s AM modulation and m=7.6%. d) Original 2 GHz RF signal with 100 Mbit/s AM. e) AM-SCM suppression after propagation through 3-SOAs and its RF signal spectrum shown in f)Figure 1Figure 2a)b)c) d)e)f)。
