Patent Application
20080158656
1. Field of the Invention
The invention relates to optical communications, and, in particular, to optical buffers employing four-wave mixing.
2. Description of the Related Art
All-optical communication systems have the potential to transmit information at bit rates higher than 10 Gb/s. Such systems employ tunable optical buffers, i.e., delay elements, with bit-level control, to implement bit-interleaved or packet-interleaved multiplexing, and to prevent conflicts among different bit streams at optical switches. M. Burzio, P. Cinato, R. Finotti, P. Gambini, M. Puleo, E. Vezzoni, and L. Zucchelli, “Optical cell synchronisation in an ATM optical switch,” Proc. ECOC 1994, pp. 581-584, the teachings of which are incorporated herein by reference, describe an optical buffer employing opto-electronic frequency conversion. J. E. Sharping, Y. Okawachi, J. van Howe, C. Xu, Y. Wang, A. E. Willner, and A. L. Gaeta, “All-optical, wavelength and bandwidth preserving, pulse delay based on parametric wave-length conversion and dispersion,” Opt. Express 13, 7872-7877 (2005), the teachings of which are incorporated herein by reference, describe an optical buffer employing modulation interaction (MI), a particular type of four-wave mixing (FWM), in a fiber.
Four-wave mixing is a parametric interaction in which two input photons are destroyed and two different, output photons are created.
In MI, a single pump wave ω1 interacts with the input signal ω1+ according to 2γ1→γ1−+γ1+, such that two pump photons 2γ1 are destroyed, and one signal photon γ1+ and one idler photon γ1− are created, where ω1−=(ω1˜ω). In PC, two pump waves ω1 and ω2 interact with the input signal ω1+ according to γ1+γ2→γ1++γ2−, such that one two pump photons γ1 and γ2 are destroyed, and one signal photon γ1+ and one idler photon γ2− are created, where ω2−=(ω2˜ω). In BS, two pump waves ω1 and ω2 interact with the input signal ω1+ according to γ1++γ2→γ1+γ2+, such that one signal photon γ1+ and one pump photon γ2 are destroyed, and one pump photon γ1 and one idler photon γ2+ are created, where ω2+=(ω2+ω). The value of the zero-dispersion frequency (ZDF) of the fiber in which the parametric interaction occurs, relative to the pump and signal frequencies, determines which type or types of FWM occur within the fiber.
The MI-based optical buffering described by Sharping et al. is a three-step process. First, MI is used to generate an idler pulse (e.g., ω1−) that is the frequency converted (FC) image of the input signal pulse (e.g., ω1+). Second, the idler is sent through a dispersive medium, such as a fiber. Third, MI is again used, this time to generate an output idler with the same frequency as the input signal. Because of dispersion, the output idler is delayed by an amount that is proportional to the frequency difference (i.e., ω1+˜ω1−) between the signal and intermediate idler.
There are (at least) three reasons why MI-based optical buffers are poor choices for communication systems: (1) their ability to controllably vary the delays is limited, (2) they are not suitable for multiple-channel systems, e.g., wavelength-division-multiplexed (WDM) systems, and (3) they generate low-quality output idlers.
In principle, one can vary the idler delay by varying the frequency difference between the input signal and the idler. In practice, the extent to which one can do this in MI-based optical buffers is limited. Suppose that the signal frequency ω1+ is fixed. If one were to vary the pump frequency ω1+, one would vary the idler frequency ω1−=2ω1˜ω1+. However, one would also vary the MI gain, which depends sensitively on the difference between the pump frequency and the ZDF of the fiber. For typical parameters, the constraint of nearly constant gain limits the pump and idler wavelengths to tuning ranges of a few nm.
Because the idler frequency ω1− equals 2ω1˜ω1+, the input signal-idler frequency difference ω1+˜ω1− equals 2(ω1+˜ω1), which depends on the signal frequency ω1+. As a result, signals in different channels experience different frequency shifts and, hence, different time delays. Hence, MI buffers are not suitable for use in WDM systems.
In many parametric devices, the pump powers exceed the stimulated Brillouin scattering (SBS) threshold by a wide margin. To circumvent SBS, pumps are phase-modulated to broaden their spectra and reduce the powers of their spectral component below the SBS threshold. Let the pump frequency be ω1+δω1, where ω1 is the average pump frequency, which is constant, and δω1 is the slowly varying perturbation to the pump frequency associated with phase modulation. Then, the idler frequency ω1− equals 2(ω1+δω1)˜ω1+. The MI idler inherits twice the phase modulation of the pump, and the dispersive fiber converts phase modulation into amplitude modulation. As a result, MI buffers generate low-quality output idlers.
MI amplifies the signal and generates a strong idler. However, with amplification comes noise. The addition of noise photons to a classical (i.e., many-photon) signal would not perturb the signal significantly and, hence, would not prevent the use of an MI buffer in a classical communication system. However, the addition of noise photons would perturb significantly the state of a quantal (i.e., few-photon) signal. Hence, MI buffers are not suitable for use in quantal communication systems.
Problems in the prior art are addressed in accordance with the principles of the invention by an optical buffer employing Bragg scattering, a particular type of four-wave mixing (FWM).
In one embodiment, an input signal and two pump signals are injected into an FWM medium. Bragg scattering within the FWM medium generates an idler signal having a frequency different from that of the input signal. The idler signal is then applied to a dispersive medium, which propagates the idler signal at a different, e.g., slower, speed than the propagation speed of the input signal through the dispersive medium, thereby resulting in a delay of the idler signal relative to the input signal as it progresses through, and especially at the far end of the dispersive medium. If desired, after transmission through the dispersive medium, the idler signal can be frequency converted back to the frequency of the original input signal via Bragg scattering in a second FWM medium.
Embodiments of the present invention can be used to perform bit-level switching of data encoded in the input signal. By using Bragg scattering (BS), embodiments of the present invention can provide optical buffering, i.e., delay, functions having one or more advantages over optical buffering achieved using the alternative FWM mechanism of modulation interaction (MI). In particular, unlike MI-based optical buffers:
In one embodiment, the present invention is an optical buffer comprising first and second pumps, a four-wave mixing (FWM) medium, and a dispersive medium. The first pump is adapted to generate a first pump signal at a first pump signal frequency, and the second pump is adapted to generate a second pump signal at a second pump signal frequency different from the first pump signal frequency. The FWM medium is adapted to (i) receive an input signal and the first and second pump signals and (ii) perform Bragg scattering (BS) so that at least some of the input signal is frequency converted into an idler signal having a frequency different from a frequency of the input signal. The dispersive medium is adapted to receive the idler signal generated in the FWM medium, whereby the idler signal propagates at a speed different from the propagation speed of the input signal through the dispersive medium.
In another embodiment, the present invention is a method and apparatus for buffering an optical input signal. Bragg scattering is used to frequency convert an input signal into an idler signal having a frequency different from the frequency of the input signal. The idler signal is transmitted through a dispersive medium, whereby the idler signal propagates at a speed different from the propagation speed of the input signal through the dispersive medium.
Other aspects, features, and advantages of the invention will become more fully apparent from the following detailed description, the appended claims, and the accompanying drawings in which like reference numerals identify similar or identical elements.
Within FWM fiber 208, BS converts some (if not substantially all) of input (data) signal ω1+ into idler signal ω2+, where ω2+=(ω1++ω2˜ω1). Filter 210 selects idler signal ω2+ for transmission through dispersive fiber 212. As known in the art, the speed at which an optical signal propagates through a dispersive fiber is a function of the signal frequency. As a result of the signal-to-idler frequency conversion from ω1+ to ω2+, the time that it takes for the idler signal ω2+ to propogate through dispersive fiber 212 will be different from the time that it would have taken the original input signal ω1+ to propagate through dispersive fiber 212, with the magnitude of the time difference (i.e., delay) being substantially proportional to the frequency difference (ω2+˜ω1+).
If desired, idler signal ω2+ can be frequency converted back to the frequency of input signal ω1+ using frequency converter 214, which can (but does not have to) employ Bragg scattering using elements analogous to elements 204-210. In that case, the resulting output signal applied to output fiber 216 would be an identical, albeit relatively time-shifted, version of the original input signal. Note that, in general, whether the time-shifted output signal is a delayed version or an advanced version of the input signal depends on (1) the sign of the signal-to-idler frequency change and (2) the sign of the dispersion coefficient (i.e., whether higher-frequency signals propagate faster or slower through the dispersive fiber). Note further that, in applications where the output signal can remain in the idler frequency ω2+, frequency converter 214 can be omitted.
In one possible operating scenario, one of pumps 204 is a continuous-wave (CW) pump that is operated continuously, and the other pump is a pulsed pump that is operated intermittently, where the BS-based frequency conversion occurs only when both pumps are on. In this way, optical buffer 200 can be used to selectively delay portions of input signal ω1+ (i.e., bits of information) relative to other portions of that input signal by selectively turning on and off the pulsed pump. Note that, in alternative implementations, both pumps can be pulsed pumps.
Bit-level switching employing FWM was demonstrated at 10 and 40 Gb/s by S. Radic, C. J. McKinstrie, R. M. Jopson, A. H. Gnauck, J. C. Centanni, and A. R. Chraplyvy, “Multi-band bit-level switching in two-pump fiber parametric devices,” IEEE Photon. Technol. Lett. 16, 852-854 (2004), and by Q. Lin, R. Jiang, C. F. Marki, C. J. McKinstrie, R. M. Jopson, J. E. Ford, G. P. Agrawal, and S. Radic, “40-Gb/s optical switching and multicasting in a two-pump parametric device,” IEEE Photon. Technol. Lett. 17, 2376-2378 (2005), the teachings of both of which are incorporated herein by reference. The experiments described in these papers involved a CW pump, whose frequency was in the normal dispersion regime, and a pulsed pump, whose frequency was in the anomalous dispersion regime. When the anomalous pump was off, the signal was not amplified and no idler was generated. When the anomalous pump was on, the signal was amplified and three idlers were generated. Because the anomalous pump can be turned on and off in a small fraction of the bit period (e.g., a few ps), selected bits can be frequency converted without affecting their neighbors. Similar remarks apply if the anomalous pump is CW and the normal pump is pulsed, or if both pumps are pulsed.
Referring again to
(1) Pulsed pump 204-2 is off during Bit A in input signal 202 to allow Bit A to pass through FWM fiber 208 without being frequency converted and then propagate through dispersive fiber 212 at signal frequency ω1+. Note that, in this implementation, filter 210 is designed to select signal frequency ω1+ as well as idler frequency ω2+ (although no idler is generated in this situation); and
(2) Pulsed pump 204-2 is on during Bit B in input signal 202 to frequency convert Bit B from signal frequency ω1+ to idler frequency ω2+ in FWM fiber 208, in which case Bit B propagates through dispersive fiber 212 at a different speed from Bit A. For this application, dispersive fiber 212 is designed to propagate higher frequencies faster than lower frequencies, such that Bit B propagating at idler frequency ω2+ overtakes Bit A propagating at signal frequency ω1+ within dispersive fiber 212 and arrives at frequency converter 214 ahead of Bit A.
Note that, if dispersive fiber 212 propagates higher frequencies slower than lower frequencies, then Bits A and B can be re-ordered by turning on pulsed pump 204-2 during Bit A and turning off pulsed pump 204-2 during Bit B. Note further that, if the time shifting is sufficiently large, then two groups of bits, each having two or more consecutive bits, can be reordered within optical buffer 200.
BS-based optical buffers, such as optical buffer 200 of
In one implementation, dispersive fiber 212 may be a single-mode fiber (SMF) having a positive delay coefficient of 17 ps/nm-Km. In another implementation, dispersive fiber 212 may be a dispersion-compensating fiber (DCF) having a delay coefficient of −200 ps/nm-Km. Assuming an available range of signal-to-idler wavelength differences from 0 to 30 nm, optical buffer 200 can be controlled (by varying pump frequency ω1) to produce delays ranging from 0 to 6 ns/Km, where the amount of delay depends on the length of the dispersive fiber. Larger signal-to-idler wavelength differences would result in even larger delays.
Pump frequency ω1 can be varied using a controllable pump, such as a tunable laser, for pump 204-1. Alternatively, pump 204-1 can be implemented using a plurality of pulsed pumps (as indicated in
Since the idler frequency ω2+ is equal to (ω1++ω2˜ω1), signal-to-idler frequency difference (ω2+˜ω1+) is equal to (ω2˜ω1), which does not depend on the signal frequency. As a result, signals in neighboring channels will experience (substantially) the same frequency shift in FWM fiber 208 and therefore (substantially) the same time delay in dispersive fiber 212. As such, unlike MI-based optical buffers, BS-based optical buffers are suitable for use in WDM or other multiple-channel communication systems.
As mentioned previously, in many parametric devices, the pumps are phase-modulated to suppress stimulated Brillouin scattering (SBS). For such phase-modulated pumps, let the pump frequencies be (ω1+δω1) and (ω2+δω2), where ω1 and ω2 are the average pump frequencies, which are constant, and δω1 and δω2 are the slowly-varying perturbations to the pump frequencies associated with phase modulation. Then, the idler frequency ω2+ equals ω1++(ω2+δω2)˜(ω1+δω1). Spectral broadening of the BS idler can be prevented by co-phasing the phase-modulation schemes, such that δω1=δω2. See S. Radic, C. J. McKinstrie, R. M. Jopson, J. C. Centanni, A. R. Chraplyvy, C. G. Jorgensen, K. Brar, and C. Headley, “Selective suppression of idler spectral broadening in two-pump parametric architectures,” IEEE Photonics Technology Letters, Vol. 15, No. 5, May 2003, the teachings of which are incorporated herein by reference. As a result, unlike MI-based optical buffers, BS-based optical buffers can be designed to generate high-quality output idlers.
BS converts signal photons into idler photons without adding noise photons. Under ideal conditions (i.e., 100% FC), the state of the output idler (i.e., after frequency conversion) is identical to the state of the input signal, albeit shifted in time. As such, unlike MI-based optical buffers, BS-based optical buffers are suitable for use in quantal as well as classical communication systems.
There are special (niche) systems in which signal polarizations are controlled. In such systems, the pump and signal polarizations can be aligned to maximize the coupling between them and simplify the buffer designs. However, because transmission fibers are not polarization-maintaining, buffers for general (mainstream) systems should operate in a manner that does not depend on the polarizations of the input signals.
Most current FWM experiments involve highly nonlinear fibers (HNFs), which are single-mode fibers with small effective areas. They are randomly birefringent and not polarization-maintaining. BS in an HNF has been found to be a signal-polarization-dependent process. BS driven by co-rotating, circularly polarized pumps in a rapidly-spun fiber (RSF) has been shown to be a polarization-independent process. Some current experiments involve micro-structured fibers (MSFs), which are strongly birefringent and polarization-maintaining. BS in an MSF is a polarization-dependent process.
For niche systems, any of the aforementioned fibers could be used to produce BS. In contrast, for mainstream systems, only RSFs provide polarization invariance. If HNFs or MSFs are used to produce BS, then a polarization-diversity scheme should be used to provide polarization invariance.
At least two such polarization-diversity schemes originally designed for MI should also work for BS. In one scheme (described by T. Hasegawa, K. Inoue, and K. Oda, “Polarization independent frequency conversion by fiber four-wave mixing with a polarization diversity technique,” IEEE Photon. Technol. Lett. 5, 947-949 (1993), the teachings of which are incorporated herein by reference), the pump and signal are sent through a polarization beam splitter (PBS), which splits the pump into two components of equal power (because the pump is polarized at 45 degrees to the polarization axes of the PBS) and sends them into a fiber loop, in opposite directions. The signal is also split into co- and counter-propagating components, each of which experiences the same amplification. When the signal components are recombined by the PBS, an output signal is produced, whose power does not depend on the polarization of the input signal.
In a related scheme (described by K. K. Chow, C. Shu, C. Lin, and A. Bjarklev, “Polarization-insensitive widely tunable wavelength converter based on four-wave mixing in a dispersion-flattened nonlinear photonic crystal fiber,” IEEE Photon. Technol. Lett. 17, 624-626 (2005), the teachings of which are incorporated herein by reference), the loop is made of birefringent fiber, whose polarization axes are aligned with those of the PBS. In this scheme the co- and counter-propagating components of the pump and signal are aligned with the polarization axes of the birefringent fiber.
A similar outcome can be achieved by a pump that is polarized at 45 degrees to the axes of an open section (not a closed loop) of birefringent fiber. Birefringence tends to suppress interactions between waves with different polarization components (as discussed by C. J. McKinstrie, S. Radic, and C. Xie, “Phase conjugation driven by orthogonal pump waves in a birefringent fiber,” J. Opt. Soc. Am. B 20, 1437-1446, (2003), the teachings of which are incorporated herein by reference). Each component of the pump drives an MI process with a co-polarized signal component. Because the coupling between these processes is weak, the power of the output signal depends only weakly on the polarization of the input signal.
Although the present invention has been described in the context of an input signal being at frequency ω1+ and an idler signal being at frequency ω2+, those skilled in the art will understand that the invention is not so limited. In general, as long as the BS condition (i.e., that the average of the input signal frequency and the “opposing” pump frequency is approximately equal to the FWM fiber's ZDF) is satisfied, an optical buffer can be designed such that any one of the four sideband frequencies (ω1−, ω1+, ω2−, and ω2+ of
Although the present invention has been described in the context of optical buffers having an FWM fiber and a dispersive fiber, optical buffers of the present invention could, in principle, be implemented using FWM media and/or dispersive media other than fibers. As used herein, the term “FWM medium” refers to any suitable medium that supports Bragg scattering. Similarly, as used herein, the term “dispersive medium” refers to any suitable medium in which the light propagation speed varies with signal frequency.
Unless explicitly stated otherwise, each numerical value and range should be interpreted as being approximate as if the word “about” or “approximately” preceded the value of the value or range.
It will be further understood that various changes in the details, materials, and arrangements of the parts which have been described and illustrated in order to explain the nature of this invention may be made by those skilled in the art without departing from the scope of the invention as expressed in the following claims.
It should be understood that the steps of the exemplary methods set forth herein are not necessarily required to be performed in the order described, and the order of the steps of such methods should be understood to be merely exemplary. Likewise, additional steps may be included in such methods, and certain steps may be omitted or combined, in methods consistent with various embodiments of the present invention.
Although the elements in the following method claims, if any, are recited in a particular sequence with corresponding labeling, unless the claim recitations otherwise imply a particular sequence for implementing some or all of those elements, those elements are not necessarily intended to be limited to being implemented in that particular sequence.
Reference herein to “one embodiment” or “an embodiment” means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the invention. The appearances of the phrase “in one embodiment” in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments necessarily mutually exclusive of other embodiments. The same applies to the term “implementation.”
