Optics Communications 282 (2009) 4693–4698
Contents lists available at ScienceDirect
Optics Communications journal homepage: www.elsevier.com/locate/optcom
The bit pattern effect on all-optical clock recovery using tank circuits and birefringent resonators E. Parsons *, H. Chaouch, F. Küppers College of Optical Sciences, University of Arizona, Tucson, AZ 85721, USA
a r t i c l e
i n f o
Article history: Received 30 March 2009 Received in revised form 3 September 2009 Accepted 3 September 2009
Keywords: Synchronization Tank circuit Fabry-Pérot resonators Wavelength division multiplexing
a b s t r a c t In this paper we discuss the bit pattern effect on optical clock signals recovered using tank circuits and using birefringent resonators. For the first time we compare the bit pattern dependence of the clock signals recovered from both schemes. We show that a birefringent resonator can recover the clock for the same number of consecutive zeros as the optical tank circuit. We also show quantitatively that the fluctuation of the clock signal amplitude is less for the birefringent resonator than the optical tank circuit. These results are obtained numerically and confirmed experimentally. Ó 2009 Elsevier B.V. All rights reserved.
1. Introduction Clock recovery is a fundamental operation in all digital transmission systems, including optical telecommunications. The decision circuit in a receiver requires the data rate and phase of the incoming signal to process the data without errors. This information is provided by clock recovery. In optical communication systems the clock recovery module is traditionally electronic, but much attention has been given to finding an all-optical replacement. A broad variety of concepts and implementations have been proposed [1], among which one popular optical clock recovery device is the optical tank circuit [2]. The optical tank circuit consists of a simple Fabry-Pérot resonator. This device suffers from a bit pattern effect: the clock signal power depends on the input data pattern [2,3]. Modern optical communication systems take advantage of parallelism through dense wave division multiplexing (DWDM) to increase data rates. It is desirable to extend this multi-channel parallelism to clock recovery by replacing several parallel electronic clock recovery devices with a single optical device. In a previous publication [4] we proposed a simple clock recovery device based on a birefringent resonator and a polarizer. This device offers the advantage of parallelism through multi-wavelength operation. A proof of concept experiment was carried out where the clock signals for 20 channels operating at 10 Gb/s and one channel operating at 40 Gb/s were recovered simultaneously. Further analysis of this device is needed. Similar to the optical tank circuit, this device suffers from a bit pattern effect which * Corresponding author. Tel.: +1 520 626 4612. E-mail address:
[email protected] (E. Parsons). 0030-4018/$ - see front matter Ó 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.optcom.2009.09.003
has been observed previously [4–6], but never quantified. A detailed understanding of how the clock signal amplitude fluctuates is necessary before this device can be used in real systems. Several methods have been proposed to eliminate the bit pattern effect in recovered clock signals. The most promising ones take advantage of the self-phase modulation (SPM) induced by a nonlinear medium. The nonlinear medium can consist of a highly nonlinear fiber [7] or a gain saturated semiconductor optical amplifier [8]. Unfortunately, these methods involve nonlinear processes and are not suited for multi-channel operation. Better understanding of the bit pattern effect will lead to solutions that support multi-channel operation. For the first time, to the best of our knowledge, we not only compare the bit pattern effect on the clock signals recovered from a tank circuit and a birefringent resonator qualitatively, but we quantify the difference in terms of the relative standard deviation (RSTD as defined in Section 2.4). We show that both the optical tank circuit and the birefringent resonator can maintain the clock signal for the same number of consecutive 0s. We also show with numerical analysis and experiments that the birefringent resonator produces a clock signal with less amplitude fluctuation than that produced by an optical tank circuit. The clock signal from the birefringent resonator has less amplitude fluctuation because the birefringent resonator passes fewer line spectral components.
2. Principle of operation The spectrum of a return-to-zero on–off-keyed (RZ–OOK) signal contains, as shown in Fig. 1a, equidistant spectral components
4694
E. Parsons et al. / Optics Communications 282 (2009) 4693–4698
The birefringent resonator can recover the clock for multiple channels and data rates simultaneously. For example, if the FSR is 50 GHz and the separation of spectral signal components is 10 GHz, the clock may be recovered for 10, 40, and 160 Gb/s, simultaneously. 2.3. Bit pattern effect The clock signals from both the optical tank circuit and the birefringent resonator suffer from a bit pattern effect [2,4]. At the output facet of the resonator, a pulse representing a binary 1 is partially transmitted and reflected. The reflected portion of the pulse then reflects off the input facet and returns to the output facet. A second pulse delayed by a time 2nl=c and attenuated by a factor of R2, relative to the first pulse, is transmitted. The impulse response of a Fabry-Pérot resonator is Fig. 1. Operation principle of passive all-optical CR using a birefringent fiber resonator. (a) Spectral decomposition of input signal at 40 and 10 Gb/s. (b) Suitably tuned birefringent resonator may filter both clock signals simultaneously. (c) Suitably polarized output signal beats at the clock frequency. (d) The CR device consisting of a birefringent resonator and a polarizer.
separated by the bit rate’s corresponding frequency fb. When the carrier and first clock tone frequencies are selectively filtered and heterodyned, the result is a sinusoid with the same frequency fb. This sinusoid is the recovered clock signal. This clock recovery method has been demonstrated with several types of optical filters [2,3]. In this work we consider clock recovery using an optical tank circuit and a birefringent resonator. 2.1. Optical tank circuit An optical tank circuit consisting of a Fabry-Pérot resonator can be used as clock recovery device [2,3]. The filter is tuned such that its free spectral range (FSR) corresponds to fb. The carrier frequency and clock tones are filtered by the tank circuit, and the resulting heterodyned output is the clock signal. When employing multiwavelength clock recovery with a simple Fabry-Pérot resonator, the resonator length l has to simultaneously fulfill the requirements for data rate and channel spacing: l ¼ c=2nfb ¼ mc=2nDf . Where Df is the channel spacing multiple of the FSR, c is the speed of light in vacuum, m is an integer, and n is the index of refraction of the resonator. 2.2. Birefringent resonator A birefringent resonator and a polarizer can generate a sinusoidal signal with frequency f ¼ cDn=ðk0 nÞ, where c is the speed of light and k0 the wavelength in vacuum, Dn is the difference of refractive indices of the polarization modes, and n is the average index of refraction ðDn 0.4. The result obtained experimentally for the birefringent resonator leads to a lesser bit pattern effect than the numerical results for the optical tank circuit for a resonator of equal finesse. With clock recovery using a birefringent resonator the bit pattern effect is less than with an optical tank circuit. This is due to the fact that the birefringent resonator passes fewer line spectral components than the optical tank circuit that would beat with the clock signal as explained in Section 3.2 and shown in Fig. 5.
The same setup shown in Fig. 6 was used to measure the number of consecutive 0s recoverable using a birefringent resonator. The data signal driving the second modulator was a string of 33 1s followed by a string of 95 0s. The recovered clock signal from this data signal is shown in Fig. 9. The 1=e level is marked in the Fig., and arrows point to the pulses corresponding to the first and seventh 0s. An envelope depicting the exponential decay is marked in the Fig. 9. The 1=e level is passed after 7 consecutive 0s are input into the resonator. This matches the numerical results in Fig. 2b. The clock signal amplitude after a long string of 0s decays according to the photon lifetime of the resonator. These results match the results presented in the literature for an optical tank circuit. 5. Conclusion In this paper we compared the bit pattern effect for the optical tank circuit and the birefringent resonator. We explained the principle of operation and the cause of the bit pattern effect for both schemes. We analyzed the bit pattern effect using numerical simulations and experiments. We showed that the birefringent resonator can recover the clock for the same number of consecutive zeros as the optical tank circuit. We also show that the clock signal amplitude fluctuation is less for the birefringent resonator than for the optical tank circuit.
4698
E. Parsons et al. / Optics Communications 282 (2009) 4693–4698
Acknowledgments The authors would like to acknowledge support from the National Science Foundation through NSF CAREER under Grant EECCS-0644799 and through NSF ERC CIAN under Grant EEC0812072. References [1] T. von Lerber, S. Honkanen, A. Tervonen, H. Ludvigsen, F. Küppers, Opt. Fiber Technol. 15 (2009) 363.
[2] M. Jinno, T. Matsumoto, J. Quant. Electron. 28 (1992) 895. [3] X. Zhou, C. Lu, P. Shum, H.H.M. Shalaby, T.H. Cheng, P. Ye, J. Lightwave Technol. 19 (2001) 603. [4] T. von Lerber, J. Tuominen, H. Ludvigsen, S. Honkanen, F. Küppers, Photonic Technol. Lett. 18 (2006) 1395. [5] T. von Lerber, H. Ludvigsen, A. Romann, Opt. Express 12 (2004) 1363. [6] T. von Lerber, J. Tuominen, H. Ludvigsen, H. Honkanen, F. Küppers, Opt. Commun. 271 (2007) 87. [7] C. Kouloumentas, A. Tzamakaki, I. Tomkos, Photonic Technol. Lett. 18 (2006) 2356. [8] Q. Wang, T. von Lerber, S. Honkanen, F. Küppers, J. Opt. Commun. 28 (2007) 242. [9] T. Wang, C. Lou, L. Huo, Z. Wang, Y. Gao, Photonic Technol. Lett. 16 (2004) 614.
