Open Access
Add to BibSonomyAbstract
In this paper, we discuss the multicast efficiency issue in the optical layer multicast. A 1-to-8 phase-preserved different phase-shifted keying (DPSK) wavelength multicast is experimentally demonstrated using four-wave mixing (FWM) in a piece of Bismuth highly nonlinear fiber (Bi-HNLF). DPSK signal is successfully delivered from one wavelength to up to eight different wavelengths using only three pumps. Compared with the existing schemes, the multicast efficiency is greatly improved by delivering the phase information to more destination wavelengths.
©2008 Optical Society of America
1. Introduction
Simultaneous multiple wavelength conversion from one to many different wavelengths, which is referred to as wavelength multicast, is expected to be a key technique for connecting networking with different wavebands in wavelength-division multiplexing (WDM) systems [1, 2]. Wavelength multicast for on-off keying (OOK) signals has previously been studied through the nonlinear effects in semiconductors optical amplifiers (SOA) [3–5], electroabsorption modulator (EAM) [6], and nonlinear fibers [7–8]. More recently, wavelength multicast of differential phase shift keying (DPSK) has also been discussed and experimentally demonstrated using four-wave mixing (FWM) in dispersion-flattened highly nonlinear photonics crystal fiber [9] and Bismuth-oxide high nonlinear fiber (Bi-HNLF) [10–11]. Nevertheless, the multicast efficiency, which is critical in evaluating the achieved multicast gain in terms of network resource consumption in IP layer multicast [12], has never been studied in wavelength multicast. Specially, the multicast efficiency in wavelength multicast could be characterized in terms of the number of probes and pumps employed to deliver the source information to a certain number of destination wavelengths. With higher multicast efficiency, source information could reach more destination wavelengths with the same number of employed probes and pumps. Naturally, it is desired to achieve wavelength multicast at higher multicast efficiency.
In this paper, we experimentally demonstrate an all-optical 1-to-5/1-to-8 DPSK wavelength multicast scheme by using FWM in Bi-HNLF with two or three pumps only. Less than 3.6-dB power penalty is observed for all of the converted channels. Owing to the large stimulated Brillouin scattering (SBS) threshold in the Bi-HNLF [13], high-power CW pumps are launched to the Bi-HNLF to increase the FWM conversion efficiency. Besides, the high nonlinear coefficient up to 1100 W-1km-1 of the Bi-HNLF ensures that more components could be generated in FWM process. Therefore, DPSK data carried in the input signal are effectively delivered to more destination wavelengths just by using two or three CW pumps, which enhances the multicast efficiency. In this paper, we raise, for the first time to the best of our knowledge, the multicast efficiency issue in the wavelength multicast. Compared with the reported DPSK experiments [9–11], a high multicast efficiency (one to eight wavelengths) is demonstrated. Moreover, different from the cross-gain modulation-based OOK multicast scheme in SOA [4] and the cross-absorption modulation-based scheme in EAM [6], no additional CW probe is required at the input, which also indicates the high multicast efficiency of the proposed scheme. Recently, OOK wavelength multicast schemes based on pump-modulated optical parametric amplifier [14–15] have been reported, offering a high multicast efficiency. However, it is not suitable for DPSK multicasting.
2. Operation principle
The proposed DPSK multicast is based on two-/three-pump FWM in Bi-HNLF. Figure 1 illustrates the spectrum of three-pump FWM. As discussed in [9–11], with more pumps the carried data in input signal can be further delivered to more generated FWM components. In addition, different from the work in [9], to effectively avoid crosstalk induced by high-order FWM and beating among the strong pumps, in our experiment three pumps are placed with un-equal spacing, i.e. Δω23≠Δω12. Although the pumps are place un-equidistantly, it is possible to ensure the converted components compliant with the ITU grid through maintaining the separations among input and pumps are multiples of 0.8 nm [16]. Among these generated FWM components, phase information is effectively preserved and transferred to components at ω423, ω143, ω341, ω223, ω123, ω231, ω113, ω132 and ω234. Because of the larger spacing among the interacted components, the FWM conversion efficiencies at ω423, ω223, ω132 and ω234 are relatively low compared with those of other components. With the multiple-pump configuration, the destination wavelength could be dynamically configured by selectively launched different pumps to the HNLF. The selective multicast functionality could effectively enhance the flexibility of the wavelength management in the future optical networks according to the available wavelength resources in networks.
In FWM experiments for OOK, phase dithering is usually performed for pumps to increase the SBS threshold and more pump power can be launched to the nonlinear fiber to improve the FWM conversion efficiency. In contrast, in the multicast scheme for DPSK, CW instead of phase-modulated light is preferred for pumps to avoid the induced phase noise from pumps. Owing to the large SBS threshold and high nonlinearity of the Bi-HNLF, high power CW light can be used as pumps to increase the conversion efficiency.
Figure 2 shows the experiment setup to verify the proposed DPSK multicast scheme. Light beams from tunable lasers at 1544 nm (P1), 1547.3 nm (P2) and 1553.6 nm (P3) were employed as the three pumps. Light from another tunable laser at 1543.1 nm worked as an input signal. The input signal was phase-modulated by 10-Gb/s 231-1 PRBS data. After individual power amplification, P1, P2, P3 and the input signal were combined by couplers and fed to a 2-m length of Bi-HNLF. The launched power to Bi-HNLF of each pump was measured as around 21 dBm, whereas the launched power of input signal was around 11dBm. The measured nonlinear coefficient γ of the Bi-HNLF (Asahi Glass Co., Japan) in the experiment was around 1100 W-1km-1. The fiber propagation loss at 1550nm was 2 dB/m. Both ends of the Bi-HNLF were spliced to SMFs using high NA fibers with input and output splicing losses of 1.5 and 1.8 dB, respectively. The measured group velocity dispersion was-320 ps/nm/km at 1550 nm. The estimated SBS threshold was around 28.4 dBm. After the Bi-HNLF, the generated FWM components were filtered out by a 1-nm optical tunable filter, and sent to a DPSK receiver with an optical pre-amplifier. A Mach-Zehnder delay interferometer (MZDI) with a 100-ps delay between two arms was employed for phase de-modulation. No other SBS suppression was employed in the experiment thanks to the high SBS threshold of the utilized Bi-HNLF.
3. Experiment and results
Fig. 3. Measured Optical spectra after Bi-HNLF with (a) two or (b) three pumps, the converted DPSK signals after multicast were indicated by triangle symbols.
Figure 3 shows the optical spectra after Bi-HNLF. With two/three CW pumps, five/nine FWM components were mainly generated and encoded as DPSK signals. In the case with two pumps at ω1 and ω2 (Fig. 2 (a)), DPSK signals were obtained at components at ω223, ω123, ω231, ω113 and ω132. In the three-pump case, DPSK signals were delivered to additional four components at ω423, ω143, ω341 and ω234 by adding the other pump at ω4 (Fig. 2(b)). Among these nine converted DPSK signals, obtained FWM component at ω223, ω123, ω113, ω423 and ω143 were with conjugated phase compared with the input at ω3, whereas phase non-conjugations were observed at ω231, ω132, ω341 and ω234. Optical phase conjugation could be employed for compensation of the nonlinear distortions and chromatic dispersion for both OOK and DPSK systems.
Note that two components were generated at ω221 and ω223 due to the pump-to-pump beating and high-order FWM. If the three pumps were equally-spaced, i.e., the third pump ω4 was placed at ω221, the component at ω223 would overlap with the converted signal ω143, and cause at least -17-dB crosstalk if we assume that the same efficiency as components at ω123 can be obtained for the converted signal at ω143 in the case with equal-spaced pumps. To avoid the crosstalk, pumps were placed with unequal-spacing.
To characterize the performance of the proposed DPSK multicast, BER of input and converted DPSK signals after multicast were measured and shown in Fig. 4. Error-free operations were obtained for the converted eight channels. The maximum power penalty of the nine obtained signals after conversion, including input signal, is 3.6 dB at BER of 10-9. The corresponding eye diagrams of input and output DPSK signals after demodulation are shown in Fig. 5.
Fig. 5. Eye diagrams of (a) input, and converted DPSK signals at (b)1543.1nm, (c)1545nm, (d)1546.4nm, (e)1548.3nm, (f)1551.7nm, (g)1554.6nm, (h)1539.8nm, (i)1551.7nm, (j)1557.9nm.
Table I. Performance of converted DPSK signals in 1-to-9 DPSK multicast
Table 1 summarizes the OSNR, FWM conversion efficiency, receiver sensitivity and power penalty of the generated FWM components in the case with three unequal-spaced CW pumps. Here, we define FWM efficiency as the ratio of the power of the FWM component to that of the input signal. It is clear that relative high FWM efficiency was obtained for the components at ω113, ω231, ω123, ω341 and ω143, close to the three pumps. Because of the low FWM efficiency at ω423, a relative large power penalty up to 3.6 dB was observed compared with the input DPSK. This can also be verified from the eye diagram shown in Fig. 5(j). For the component at ω234, only around -41-dB FWM conversion efficiency was obtained. It was hard to measure the BER of the converted signal due to the low OSNR. The FWM efficiency was mainly limited by the relative large dispersion of the employed Bi-HNLF. Higher FWM efficiency and better performance can be expected by using a dispersion-shifted Bismuth-oxide photonic crystal fiber [17]. Besides, the performance and FWM efficiency of proposed DPSK multicast scheme can be further improved by further increasing the pump power or reducing the pump spacing.
Different from the case with OOK input, not all of the generated FWM components, which were contributed by input at ω3, carry the same phase information as the input light. For examples, no (0, π) phase modulation was preserved in FWM components at ω331 and ω332 after HNLF. They were produced by cascaded processes satisfying energy relation ω331=2 ω3- ω1, and ω332=2ω3- ω2, respectively. Binary (0, π) phase modulation was eliminated in these converted components after the process, whereas intensity modulation could be preserved with an improved extinction ratio. This indicates that, if the scheme is applied in the multicast for input OOK signal, the input information could be multi-casted to more wavelengths with further-enhanced multicast efficiency.
Table 2. Multicast efficiency comparison of referred works in this letter
As an example, we compare the resources utilization of the proposed work with the mentioned experiments [3–7] by taking into account the number of input probes, pumps and the converted wavelengths, as shown in Table 2. To generally evaluate the multicasting efficiency for different multicasting schemes, an efficiency factor could be defined as a ratio of the number of the employed probes and pumps to the number of the converted wavelengths. A larger ratio implies a higher multicasting efficiency. It is clear that the demonstrated experiment shows high multicast efficiency.
5. Conclusion
In this paper, we discussed the multicast efficiency in DPSK wavelength multicast experiment. A 1-to-8 DPSK wavelength multicast was successfully demonstrated based on FWM in Bi-HNLF. Phase information at input DPSK has been effectively preserved and delivered to up to eight different wavelengths just using three pumps, showing a high multicast efficiency. This could be attributed to the high nonlinearity, short fiber length and large SBS threshold offered by the Bi-HNLF. Less than 3.6-dB power penalty for all of the converted eight channels has been successfully observed.
References and links
1. X. Zhang, J. Wei, and C. Qiao, “On fundamental issues in IP over WDM multicasting,” in Int. Conf. Computer, Communications and Networks (IC3N) (1999), pp.84–90.
2. C.Y. Li, P.K.A. Wai, X.C. Yuan, and V.O.K. Li,“Multicasting in deflection-routed all-optical packet-switched networks,” in Proceeding of Globecom (2002), Taipei, pp.17–21.
3. G. Contestabile, M. Presi, and E. Ciaramella, “Multiple wavelength conversion for WDM multicasting by FWM in an SOA,” IEEE Photonics Technology Letters 16, 1775–1777 (2004). [CrossRef]
4. T. Ito, I. Ogawa, Y. Suzaki, K. Magari, Y. Kawaguchi, Y. Suzuki, and N. Ishihara,“Eight-channel simultaneous wavelength conversion from equal to unequal channel spacing,” IEEE Photon. Technol. Lett. 13, 1106–1108 (2001). [CrossRef]
5. T. Miyazaki, T. Kato, Y. Nagao, and S. Yamamoto, “Multicast demonstration in the optical cross-connect node using the wavelength converter by a semiconductor optical amplifier,” The OSA Trends in Optics and Photonics Series 30, 212–215 (1999).
6. L. Xu, N. Chi, K. Yvind, L. Christiansen, L. Oxenløwe, J. Mørk0, P. Jeppesen, and J. Hanberg, “7×40 Gb/s base-rate RZ all-optical broadcasting utilizing an electroabsorption modulator,” Opt. Express 12, 416–420 (2004). [CrossRef] [PubMed]
7. Q. Lin, R. Jiang, C.F. Marki, C.J. McKinstrie, R. Jopson, J. Ford, G.P. Agrawal, and S. Radic, “40-gb/s optical switching and wavelength multicasting in a two-pump parametric device,” IEEE Photonics Technology Letters 17, 2376–2378 (2005). [CrossRef]
8. Y. Wang, C. Yu, T. Luo, L. Yan, Z. Pan, and A. E. Willner, “Tunable all-optical wavelength conversion and wavelength multicasting using orthogonally polarized fiber FWM,” J. of Lightwave Technol. 23, 3331–3338 (2005). [CrossRef]
9. M. P. Fok and C. Shu, “Multipump four-wave mixing in a photonic crystal fiber for 6x10Gb/s wavelength multicasting of DPSK signals,” IEEE Photon. Technol. Lett. 19, 1166–1168 (2007). [CrossRef]
10. G.-W. Lu, K. S. Abedin, and T. Miyazaki, “Wavelength Multicasting of DPSK Signals Using Dual-Pump FWM in a Bismuth-Oxide Highly Nonlinear Fiber,” in Proc. OFC (2008), paper OMP4.
11. G.-W. Lu, K. S. Abedin, and T. Miyazaki, “DPSK Wavelength Multicasting Using FWM with Three Unequally Spaced Pumps in a Bismuth-Oxide Highly Nonlinear Fiber,” in Proc. CLEO/QELS (2008), paper JWA100.
12. P. V. Mieghem, G. Hooghiemstra, and R. Hofstad, “On the efficiency of multicast,” IEEE/ACM Transaction on networking 9, 719–732 (2001). [CrossRef]
13. J. H. Lee, T. Nagashima, T. Hasegawa, S. Ohara, N. Sugimoto, T. Tanemura, and K. Kikuchi, “Wavelength conversion of 40-Gb/s NRZ signal using four-wave mixing in 40-cm-long Bismuth oxide based highlynonlinear optical fiber,” in Proc. OFC (2005), paper PDP23.
14. G.-W. Lu, K. Abedin, and T. Miyazaki, “All-optical Broadband Tunable Wavelength Multicasting using a Pump-Modulated Wide-Band Fiber Optical Parametric Amplifier with High and Flat Gain,” in Proc. ECOC (2007), paper P027.
15. C. Bres, N. Alic, E. Myslivets, and S. Radic, “1-to-40 Multicasting and Amplification of 40Gbps Channels in Wideband Parametric Amplifier,” in National Fiber Optic Engineers Conference, OSA Technical Digest (CD) (Optical Society of America, 2008), paper PDP16.
16. K. Lau, L. Xu, S. H. Wang, L. F. K. Lui, P. K. W. Wai, C. Lu, and H. Y. Tam, “Selectable multicast using Raman-assisted four-wave mixing in dispersion shifted fiber,” in Proc. OECC/IOCC (2007), paper 12P-14.
17. K. K. Chow, K. Kikuchi, T. Nagashima, T. Hasegawa, S. Ohara, and N. Sugimoto, “Widely tunable wavelength conversion by four-wave mixing in 1-m dispersion-shifted Bismuth-Oxide photonic crystal fiber,” in Proc. OFC (2007), paper OTuI2.
