US20020145795A1 - Gain flattening with nonlinear sagnac amplifiers - Google Patents
Gain flattening with nonlinear sagnac amplifiers Download PDFInfo
- Publication number
- US20020145795A1 US20020145795A1 US10/060,868 US6086802A US2002145795A1 US 20020145795 A1 US20020145795 A1 US 20020145795A1 US 6086802 A US6086802 A US 6086802A US 2002145795 A1 US2002145795 A1 US 2002145795A1
- Authority
- US
- United States
- Prior art keywords
- optical
- gain
- amplifier
- signals
- range
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B10/00—Transmission systems employing electromagnetic waves other than radio-waves, e.g. infrared, visible or ultraviolet light, or employing corpuscular radiation, e.g. quantum communication
- H04B10/29—Repeaters
- H04B10/291—Repeaters in which processing or amplification is carried out without conversion of the main signal from optical form
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B10/00—Transmission systems employing electromagnetic waves other than radio-waves, e.g. infrared, visible or ultraviolet light, or employing corpuscular radiation, e.g. quantum communication
- H04B10/29—Repeaters
- H04B10/291—Repeaters in which processing or amplification is carried out without conversion of the main signal from optical form
- H04B10/293—Signal power control
- H04B10/294—Signal power control in a multiwavelength system, e.g. gain equalisation
-
- G—PHYSICS
- G02—OPTICS
- G02F—OPTICAL DEVICES OR ARRANGEMENTS FOR THE CONTROL OF LIGHT BY MODIFICATION OF THE OPTICAL PROPERTIES OF THE MEDIA OF THE ELEMENTS INVOLVED THEREIN; NON-LINEAR OPTICS; FREQUENCY-CHANGING OF LIGHT; OPTICAL LOGIC ELEMENTS; OPTICAL ANALOGUE/DIGITAL CONVERTERS
- G02F1/00—Devices or arrangements for the control of the intensity, colour, phase, polarisation or direction of light arriving from an independent light source, e.g. switching, gating or modulating; Non-linear optics
- G02F1/35—Non-linear optics
- G02F1/3515—All-optical modulation, gating, switching, e.g. control of a light beam by another light beam
- G02F1/3517—All-optical modulation, gating, switching, e.g. control of a light beam by another light beam using an interferometer
- G02F1/3519—All-optical modulation, gating, switching, e.g. control of a light beam by another light beam using an interferometer of Sagnac type, i.e. nonlinear optical loop mirror [NOLM]
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01S—DEVICES USING THE PROCESS OF LIGHT AMPLIFICATION BY STIMULATED EMISSION OF RADIATION [LASER] TO AMPLIFY OR GENERATE LIGHT; DEVICES USING STIMULATED EMISSION OF ELECTROMAGNETIC RADIATION IN WAVE RANGES OTHER THAN OPTICAL
- H01S3/00—Lasers, i.e. devices using stimulated emission of electromagnetic radiation in the infrared, visible or ultraviolet wave range
- H01S3/10—Controlling the intensity, frequency, phase, polarisation or direction of the emitted radiation, e.g. switching, gating, modulating or demodulating
- H01S3/13—Stabilisation of laser output parameters, e.g. frequency or amplitude
- H01S3/1301—Stabilisation of laser output parameters, e.g. frequency or amplitude in optical amplifiers
- H01S3/1302—Stabilisation of laser output parameters, e.g. frequency or amplitude in optical amplifiers by all-optical means, e.g. gain-clamping
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B10/00—Transmission systems employing electromagnetic waves other than radio-waves, e.g. infrared, visible or ultraviolet light, or employing corpuscular radiation, e.g. quantum communication
- H04B10/25—Arrangements specific to fibre transmission
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B10/00—Transmission systems employing electromagnetic waves other than radio-waves, e.g. infrared, visible or ultraviolet light, or employing corpuscular radiation, e.g. quantum communication
- H04B10/29—Repeaters
- H04B10/291—Repeaters in which processing or amplification is carried out without conversion of the main signal from optical form
- H04B10/293—Signal power control
- H04B10/2933—Signal power control considering the whole optical path
- H04B10/2935—Signal power control considering the whole optical path with a cascade of amplifiers
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01S—DEVICES USING THE PROCESS OF LIGHT AMPLIFICATION BY STIMULATED EMISSION OF RADIATION [LASER] TO AMPLIFY OR GENERATE LIGHT; DEVICES USING STIMULATED EMISSION OF ELECTROMAGNETIC RADIATION IN WAVE RANGES OTHER THAN OPTICAL
- H01S2301/00—Functional characteristics
- H01S2301/04—Gain spectral shaping, flattening
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01S—DEVICES USING THE PROCESS OF LIGHT AMPLIFICATION BY STIMULATED EMISSION OF RADIATION [LASER] TO AMPLIFY OR GENERATE LIGHT; DEVICES USING STIMULATED EMISSION OF ELECTROMAGNETIC RADIATION IN WAVE RANGES OTHER THAN OPTICAL
- H01S2301/00—Functional characteristics
- H01S2301/06—Gain non-linearity, distortion; Compensation thereof
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01S—DEVICES USING THE PROCESS OF LIGHT AMPLIFICATION BY STIMULATED EMISSION OF RADIATION [LASER] TO AMPLIFY OR GENERATE LIGHT; DEVICES USING STIMULATED EMISSION OF ELECTROMAGNETIC RADIATION IN WAVE RANGES OTHER THAN OPTICAL
- H01S3/00—Lasers, i.e. devices using stimulated emission of electromagnetic radiation in the infrared, visible or ultraviolet wave range
- H01S3/05—Construction or shape of optical resonators; Accommodation of active medium therein; Shape of active medium
- H01S3/06—Construction or shape of active medium
- H01S3/063—Waveguide lasers, i.e. whereby the dimensions of the waveguide are of the order of the light wavelength
- H01S3/067—Fibre lasers
- H01S3/06754—Fibre amplifiers
Definitions
- the present invention is in the field of fiber optic communication systems, and, more particularly, is in the field of amplifiers for use in fiber optic communication systems.
- FIG. 2A the input power spectrum is plotted in FIG. 2A
- FIG. 2B The differences in the two spectra in FIGS. 2A and 2B illustrate the large disparity in resulting channel powers caused by EDFA gain variations.
- the present invention is directed to gain flattening with nonlinear Sagnac amplifiers.
- the use of nonlinear Sagnac amplifiers for gain flattening is a novel solution.
- the present invention is directed to a filter which has a loss that is a function of power. More particularly, the filter in accordance with the present invention attenuates a specific channel i based on the power of that channel.
- the filter in accordance with the present invention does not provide a loss that is a conventional broadband power-dependent loss. Rather, the filter in accordance with the present invention provides a narrowband power-dependent loss.
- the attenuation of the filter at ⁇ i is a function of the power around that wavelength (i.e., at ⁇ i ⁇ O), but is not a function of the power at a separate wavelength ⁇ i ⁇ n outside of the ⁇ i ⁇ O window.
- NSA nonlinear Sagnac amplifier
- One aspect of the present invention is an amplification system for reducing power differences in a plurality of output optical signals responsive to a plurality of input optical signals having a plurality of respective optical wavelengths and having a plurality of respective input powers.
- the amplification system comprises an interferometric loop.
- a coupler couples the plurality of input optical signals to the loop to cause respective first portions of the input optical signals to propagate in a first direction in the loop and to cause respective second portions of the input optical signals to propagate in a second direction in the loop.
- the coupler combines the first and second portions after the first and second portions propagate in the loop to produce a plurality of output optical signals.
- An amplifier is located at an asymmetric location with respect to the center of the loop.
- the amplifier has a gain spectrum which causes the amplifier to have a plurality of respective gains at the plurality of optical wavelengths.
- the asymmetric location of the amplifier with respect to the center of the loop causes differences in powers of the first signal portions and the second signal portions of the input optical signals while these portions are traveling through the interferometric loop.
- the differences in powers of the first and second signal portions cause respective phase shifts in the first and second signal portions to occur in the fiber loop due to the optical Kerr effect.
- the Kerr-induced phased shifts vary in response to differences in the respective input powers and the respective gains to cause a greater Kerr-induced attenuation of input optical signals having a greater gain-power product.
- the amplifier comprises an erbium-doped fiber amplifier.
- Certain preferred embodiments further include a wavelength division multiplexed coupler in the loop proximate to the amplifier.
- a pump source is coupled to the wavelength division multiplexed coupler to provide pump light for the amplifier via the wavelength division multiplexed coupler.
- Another aspect of the present invention is an amplification system for reducing output power differences in a plurality of output optical signals responsive to a plurality of input optical signals having a plurality of respective optical wavelengths and having a plurality of respective input powers.
- the amplification system comprises an interferometric loop which has first and second lengths of optical fiber separated by an optical amplifier. The first length of optical fiber is substantially longer than the second length of optical fiber.
- a coupler couples the optical signals into the interferometric loop to cause respective first and second portions of the optical signals to counterpropagate in first and second directions in the interferometric loop.
- the coupler combines the respective first and second portions of the optical signals after propagation through the interferometric loop to produce a plurality of respective output signals at the plurality of optical wavelengths.
- the plurality of output signals have a plurality of respective output powers.
- the amplifier has a gain characteristic which causes the amplifier to have a plurality of respective gains at the plurality of optical wavelengths.
- the first and second portions of the optical signals propagating in the first and second directions experience respective Kerr-induced phase shifts caused by self-phase modulation, by copropagating cross-phase modulation, and by counterpropagating cross-modulation.
- the location of the amplifier causes light propagating in the first direction to pass through the first length of optical fiber before propagating through the amplifier and the second length of optical fiber.
- the location of the amplifier also causes light propagating in the second direction to propagate through the second length of optical fiber and the amplifier before propagating through the first length of optical fiber.
- the location of the amplifier also causes the light propagating in the first direction to experience greater counterpropagating cross-phase modulation than the light propagating in the second direction.
- the location of the amplifier also causes the light propagating in the second direction to experience greater self-phase modulation and greater copropagating cross-modulation than light propagating in the first direction.
- the Kerr-induced phase shifts of the plurality of optical signals at the plurality of optical wavelengths are responsive to the respective amplifier gains at the plurality of optical wavelengths and are further responsive to the respective input powers of the plurality of optical signals, such that differences in the output powers caused both by differences in the input powers and by differences in the gains at the plurality of optical wavelengths are reduced.
- Another aspect of the present invention is an optical amplification system which comprises a serial chain of at least first and second amplification sections.
- the first amplification section is coupled to receive a plurality of input optical signals. Each input optical signal has a respective optical wavelength and a respective input optical power.
- Each amplification section includes a respective amplifier therein. The amplifiers have respective gain characteristics such that gains applied to the optical signals vary with wavelength.
- the first amplification section provides a first plurality of optical output signals.
- the second amplification section is coupled to receive the first plurality of optical output signals and to provide a second plurality of optical output signals. Each of the second plurality of optical output signals has a respective optical wavelength and has a respective output optical power.
- the amplification sections operate to reduce differences in the respective output optical powers of the second plurality of optical output signals caused by differences in the input optical powers and by differences in gains applied to the optical signals.
- Each amplification section comprises an interferometric loop.
- the amplifier of the amplification section is located asymmetrically in the loop.
- a coupler couples light to the interferometric loop to cause the light to propagate as first and second counterpropagating portions at each of the optical wavelengths.
- the coupler also combines the first and second counterpropagating portions at each of the wavelengths after the first and the counterpropagating portions of the light have propagated through the loop.
- the first and second counterpropagating portions at each of the wavelengths interfere to provide an output signal at each of the wavelengths.
- the output signal at each of the wavelengths has a power responsive to input power at the wavelength, responsive to amplifier gain at the wavelength, and responsive to Kerr-induced phase shift at the wavelength.
- the Kerr-induced phase shift is greater for optical wavelengths having greater gain-power products to at least partially reduce differences in output power caused by differences in gain-power products.
- each of the amplifiers comprises an erbium-doped fiber amplifier.
- the amplifiers have gains which vary with optical wavelength.
- the optical signals at the plurality of wavelengths have varying powers.
- the amplification sections operate to cause output powers at each of the wavelengths at the output of the serial chain to converge toward nominally the same output power within a selectable range of output power.
- Another aspect of the present invention is a method of amplifying a plurality of input optical signals.
- the input optical signals have respective optical wavelengths within a range of optical wavelengths and have respective optical powers within a range of input optical powers.
- the method produces a corresponding plurality of output optical signals having respective output powers within a selected range of output optical powers.
- the method comprises passing the input optical signals through a first nonlinear Sagnac amplifier to produce a plurality of intermediate optical signals.
- the plurality of intermediate optical signals are responsive to the plurality of input optical powers, to amplifier gain and to Kerr-induced phase shift to have a respective plurality of intermediate optical powers within an intermediate range of optical powers.
- the intermediate range of optical powers is a smaller range than the range of input optical powers.
- the method further comprises passing the intermediate optical signals through at least a second nonlinear Sagnac amplifier to produce the plurality of output optical signals.
- the plurality of output optical signals are responsive to the plurality of intermediate optical powers, to amplifier gain and to Kerr-induced phase shift to have a plurality of output optical powers within the selected range of output optical powers.
- the selected range of output optical powers is smaller than the range of intermediate optical powers and is smaller than the range of input optical powers.
- Another aspect of the present invention is an optical system that comprises an input that receives optical input signals having wavelengths within a range of wavelengths.
- An output outputs optical signals responsive to the optical input signals.
- At least a first amplifier is interposed between the input and the output.
- the first amplifier has a first gain spectrum.
- the first gain spectrum varies with wavelength over the range of wavelengths to cause a first optical signal produced by the first amplifier at a first wavelength within the range of wavelengths to have a first optical power and to cause a second optical signal produced by the first amplifier at a second wavelength within the range of wavelengths to have a second optical power.
- the first optical power and the second optical power differ by a power difference.
- the system further comprises at least a second amplifier interposed between the first amplifier and the output.
- the second amplifier comprises a nonlinear Sagnac amplifier.
- the second amplifier operates to at least partially compensate for differences in gain of the first amplifier over the range of wavelengths to reduce the power difference between the first optical signal and the second optical signal.
- Another aspect of the present invention is an optical system that comprises an input that receives optical input signals having wavelengths within a range of wavelengths and that comprises an output that outputs optical output signals responsive to said optical input signals.
- a first plurality of amplifiers of a first type are interposed between the input and the output.
- the amplifiers of the first type have a first gain spectrum that varies over the range of wavelengths to cause a first optical signal at a first wavelength within the range of wavelengths to be amplified with a first gain by the amplifiers of the first type and to cause a second optical signal at a second wavelength within the range of wavelengths to be amplified with a second gain by the amplifiers of the first type.
- the second gain differs from the first gain.
- a second plurality of nonlinear Sagnac amplifiers are interposed at selected locations in the optical system.
- the nonlinear Sagnac amplifiers operate to at least partially compensate for differences in the first gain and the second gain of the amplifiers of the first type to reduce power differences between the first optical signal and the second optical signal.
- FIG. 1A illustrates a linear amplifier chain comprising a plurality of multiple loss-gain sections, wherein the loss sections comprise fiber sections and the gain sections comprise erbium doped fiber amplifiers (EDFAs);
- EDFAs erbium doped fiber amplifiers
- FIG. 1B illustrates the gain variation of an exemplary erbium doped fiber amplifier as a function of wavelength
- FIG. 2A illustrates the input power spectrum of a wavelength division multiplexed (WDM) system with an input signal having a channel spacing of 2 nanometers;
- WDM wavelength division multiplexed
- FIG. 2B illustrates the output power spectrum resulting from the input signal spectrum of FIG. 2A being applied to five gain-loss sections wherein the gain variation as a function of wavelength is shown in FIG. 1B;
- FIG. 3 illustrates an embodiment of a nonlinear Sagnac amplifier in accordance with the present invention
- FIG. 3A illustrates a preferred embodiment of an optical amplifier of FIG. 3.
- FIG. 6 illustrates a graph of the asymptotic behavior of P i (n) as a function of the amplifier gain, showing that as the gain increases the power in the channel oscillates and then becomes chaotic;
- FIG. 7B illustrates the resulting spread in the channel spectrum for a pseudo-random bit pattern applied to a PSK system, wherein the amplitude is normalized in linear units;
- FIG. 8 illustrates P cr , the critical power in milliwatts above which stimulated Brillouin scattering become significant, as a function of the bit rate, wherein the bit rate is shown in megabits per second on a logarithmic scale;
- FIG. 9 illustrates a plot of the asymptotic values of G i P i (n) X/2 as a function of gain
- FIG. 10 illustrates the gain spectra for 16 wavelength division multiplexed channels for a total input power of 1.4 milliwatts (curve a), 0.56 milliwatts (curve b) and 0.22 milliwatts (curve b)
- FIG. 12B illustrates the channel power spectrum for a chain of standard amplifiers after 20 segments
- FIG. 14B illustrates an embodiment of an EDFA for use in the loop of a nonlinear Sagnac amplifier in which the EDFA comprises two amplifiers in series for amplifying two optical bands;
- FIG. 15 illustrates a configuration having two nonlinear Sagnac amplifiers in parallel to provide amplification over two different optical bands
- the NSA 300 comprises an input optical fiber 312 (or other suitable waveguide) which is coupled to a first port (A) of an optical circulator 314 .
- the optical circulator 314 has a second port (B) and a third port (C).
- the optical circulator 314 operates in a known manner to cause light provided as an input to one port to propagate in a single direction (e.g., clockwise in FIG. 3) to the next port where substantially all the light is output from the circulator.
- light provided as an input to port A of the circulator 314 is output from port B of the circulator 314 .
- Light provided as an input to port B of the circulator 314 is output from port C of the circulator 314 .
- Port C of the circulator 314 is connected to an output fiber 320 which provides an output port 1.
- Port B of the circulator 314 is connected via a fiber section 322 to a first port 332 on the input/output side of a conventional 2 ⁇ 2 optical coupler 330 .
- the coupler 330 has a second port 334 on the input/output side of the coupler 330 .
- the coupler 330 has a third port 336 and a fourth port 338 on the loop side of the coupler 330 .
- the second port 334 of the 2 ⁇ 2 optical coupler 330 is connected to an output fiber 340 which provides an output port 2.
- the WDM coupler 380 operates to couple pump light from the pump source 386 to the erbium-doped fiber 370 via the first port 381 .
- substantially no light at the signal wavelengths is coupled either to the second port 382 or to the fourth port 384 .
- substantially all light at the signal wavelengths remains in the interferometric loop.
- An input signal to be amplified is applied to the input fiber 312 .
- the signals are modulated either in phase or in frequency. Examples of such systems include phase-shift keying (PSK) and frequency-shift keying (FSK).
- PSK phase-shift keying
- FSK frequency-shift keying
- the signals are modulated in amplitude; however, in such an implementation, limitations may apply due to pulse distortion and channel cross-talk.
- the first portion of the input signal propagates from the third port 336 of the 2 ⁇ 2 optical coupler 330 , through the loop 354 in a clockwise (CW) direction, to the first port 362 of the optical amplifier 360 .
- the light is amplified in the optical amplifier 360 , and the amplified first portion propagates from the second end 364 of the optical amplifier 360 to the fourth port 338 of the 2 ⁇ 2 optical coupler 330 via the length of optical fiber 366 .
- the second portion of the input signal propagates from the fourth port 338 of the 2 ⁇ 2 optical coupler 330 , propagates to the second end 364 of the optical amplifier 360 and is amplified therein.
- the amplified second portion then propagates from the first end 362 of the optical amplifier 360 and propagates through the loop 354 in the counterclockwise (CCW) direction to the third port 336 of the 2 ⁇ 2 optical coupler 330 .
- the amplified first portion of light entering the fourth port 338 of the 2 ⁇ 2 optical coupler 330 and the amplified second portion of light entering the third port 336 of the 2 ⁇ 2 optical coupler 330 are interferometrically combined therein.
- the combined signal is output from the first port 332 or from the second port 334 or from both ports of the 2 ⁇ 2 optical coupler 330 .
- Light output from the second port 334 propagates to the output port 2 via the output fiber 340 .
- Light output from the first port 332 propagates to port B of the circulator 314 , is coupled to port C of the circulator 314 , and propagates to the output port 1 via the fiber 320 .
- the power in the output port 1 and the power in the output port 2 thus depend on the phase relationship between the clockwise propagating and counterclockwise propagating signals.
- the Kerr-induced phase shift, d ⁇ i , of a channel i of wavelength ⁇ i results from a third-order nonlinearity in the fiber, and its effect is to impart a nonlinear phase shift on a beam given by: dI i ⁇ n 2 ⁇ ⁇ i A eff ⁇ ( P self + 2 ⁇ P other ⁇ dl . ( 1 )
- N channels are launched into port 312 of the NSA shown in FIG. 3.
- the length of the loop 354 is L
- the gain of the amplifier at ⁇ i is G i
- P i is the input signal power at ⁇ i in the port 312 .
- Equation 4 the numerical value of the nonlinear coefficient n 2 depends on several parameters, in particular the fiber material and the nature and concentration of the dopants it may contain. For example, for a nearly pure silica fiber core, the value of n 2 is typically around 2.4 ⁇ 10 ⁇ 16 cm 2 /W in the 1.3-1.6 ⁇ m spectral region. In silica-based fibers heavily doped with germanium, this value can be enhanced by a factor of a few.
- the Kerr-induced index change can be calculated by averaging first the relative phase of each signal, and second the spatial orientation of the fiber birefringent axes, as described in the Clausen, et al., reference cited above.
- Equation 4 it can be shown that Equation 4 still applies, provided that n 2 is replaced by a smaller effective nonlinear coefficient n 2,eff .
- n 2eff 5.47 ⁇ 10 ⁇ 20 cm 2 /W.
- n 2eff 5.47 ⁇ 10 ⁇ 20 cm 2 /W.
- the Kerr-induced phase difference is a function of only the power in channel i, which is exactly the narrowband behavior needed.
- the output of a Sagnac interferometer is a function of the phase difference between the interfering signals.
- the output of the Sagnac amplifier 300 is a function of this Kerr-induced phase difference as well as any phase bias between the two counterpropagating light signals.
- the Kerr-induced phase difference is converted into a wavelength-dependent attenuation via the operation of the Sagnac interferometer 310 .
- the transfer function of the NSA 300 which relates the input power of channel i to the output power of channel i on a particular port, is given by: P i out ⁇ ⁇ - ⁇ ⁇ ⁇ L ⁇ G i ⁇ P i ⁇ 1 + cos ⁇ ( ⁇ ⁇ ⁇ ⁇ i Kerr + ⁇ bias ) 2 . ( 5 )
- ⁇ bias is the bias phase shift of the Sagnac interferometer 310 other than the Kerr effect, which, when the light is initially input to the port 312 , is naturally 0 for output port 1 and ⁇ for output port 2.
- the transfer function becomes: P i out ⁇ ⁇ - ⁇ ⁇ ⁇ L ⁇ G i ⁇ P i ⁇ 1 + cos ⁇ ( n 2 ⁇ ⁇ i 2 ⁇ A eff ⁇ ( G i - 1 ) ⁇ P i ⁇ L eff + ⁇ bias ) 2 . ( 6 )
- Equation 6 it can be seen that P i out is nonlinearly related to P i .
- the phase bias ⁇ bias and/or the length L eff are selected properly, as the gain-power product G i P i increases, the effect of the Kerr-induced phase difference is to multiply the gain-power product by a smaller and smaller ratio, so that increases in the gain-power product do not result in proportionate increases in the output power.
- the output power P i out in Equation 6 depends on the power P i in channel i, but it does not depend on the powers P i carried by any of the other channels j ⁇ i.
- a plurality of nonlinear Sagnac amplifiers (NSAs) 300 in accordance with FIG. 3 are coupled together to form a chain 400 of amplifiers which achieves gain flattening.
- the exemplary NSA chain 400 of FIG. 4 is formed by coupling an input signal to be amplified to a first fiber section 410 ( 1 ) and then to the input fiber 312 ( 1 ) of a first NSA 300 ( 1 ).
- the output port 1 (fiber 320 ( 1 )) of the first NSA 300 ( 1 ) is connected to the input fiber 312 ( 2 ) of a second NSA 300 ( 2 ) via a second fiber section 410 ( 2 ).
- the output fiber 320 ( 2 ) of the second NSA 300 ( 2 ) is connected to the input fiber 312 ( 3 ) of a third NSA 300 ( 3 ) via a third fiber section 410 ( 3 ), and so on, until a desired number of NSAs 300 are connected together.
- the fiber sections 410 ( i ) represent lengths of fiber such as, for example, tens of kilometers (e.g., 70 kilometers) of fiber between amplifiers in a long distance communication system.
- the fiber sections can also represent losses, such as splitter losses, or the like, or a combination of component losses and fiber losses.
- the amplifiers within each NSA 300 can be pumped by a common pump source, if the NSAs 300 ( i ) are located close to each other, or the amplifiers can be pumped with individual pump sources.
- L eff is the effective length of the Sagnac loop 354 in each NSA 300 ( i ) of FIG. 4.
- a total loss from the output of one NSA 300 ( n ) to the input of the next NSA 300 ( n+ 1) is approximately the same for each NSA 300 and has a value of X.
- X has a value of 0.01 ( ⁇ 20 dB).
- P i (n) is defined as the power of channel i at the output of the loss-gain section n of the NSA chain 400 .
- the following values are also used in the following discussion:
- Equation 7 the transfer function of a single NSA is based upon Equation 6, and further includes the effect of the loss X.
- the evolution of the NSA output powers P i (n) as n increases is plotted in FIGS. 5 A- 5 D, as discussed below.
- Equation 8 TF (n) denotes the application of the function TF n times.
- P i (n) TF(P i (n) )
- FIG. 5A For a gain, G i of 26 dB.
- An initial power, P i (0) is chosen (in this case 3 milliwatts).
- the asymptotic value of P i (n) is thus independent of the initial value for a large range of P i (0) .
- FIG. 5B plots the value of P i (n) as a function of n for five different initial values of P i (0) . Note that all values of P i (n) converge to the same asymptotic value of approximately 5 milliwatts.
- the gain, G i has been increased from 26 dB to 28 dB.
- the power, P i (n) in FIG. 5C oscillates between an upper value (about 5 milliwatts) and a lower value (about 3.5 milliwatts).
- the oscillation can be seen more clearly in FIG. 5D, which graphically shows the evolution of P i (n) as n increases. Note that for the example of FIGS. 5C and 5D, P i (n) exhibits a periodic behavior, which in FIG. 5D has a period of 2.
- FIG. 6 plots the asymptotic value (or set of values) of P i (n) as a function of gain.
- P i (n) When the gain reaches a value of about 27.6 dB, the asymptotic behavior of P i (n) becomes periodic. P i (n) first oscillates between two values. The difference between the two values becomes greater as the gain increases. As illustrated in FIG. 6, at approximately 28.46 dB of gain, P i (n) begins oscillating between four values. At approximately 28.63 dB of gain, P i (n) begins oscillating between eight values. For a very small range of gains, P i (n) oscillates between 16 values, then 32 values (not shown), and then 64 values (not shown). For slightly higher gains, the asymptotic behavior of P i (n) becomes chaotic, as approximately represented by the stippling and white bands in FIG.
- the irregular white bands are caused by narrow ranges of gain wherein the asymptotic behavior resumes periodic behavior.
- the present invention operates with a range of gains less than the gain where chaotic behavior begins, and it is not necessary to further discuss the chaotic behaviors shown in FIG. 6. Furthermore, the invention preferably operates at gains less than the gains wherein periodic behavior begins.
- an NSA chain tends to force the power of a channel into a steady-state value due to the self-regulating transmission resulting from the Kerr-induced phase shift.
- the channel at ⁇ 2 eventually overwhelms the channel at ⁇ 1
- P i (n) the power in the channel at ⁇ 1
- the foregoing description characterizes how the NSA chain 400 behaves.
- the following discussion describes the simulation of the NSA chain 400 to find optimal configurations.
- the relatively complicated behavior of the NSA chain 400 is first reduced to a single figure of merit parameter.
- the figure of merit is defined as an allowable gain variability, ⁇ G.
- the allowable gain variability parameter describes a range over which the gain can vary while still forcing P i (n) to converge to an acceptable asymptotic behavior.
- Equation 9 places an upper limit on the amount of power exiting the optical amplifier 360 in the Sagnac loop fiber 356 in the counterclockwise direction.
- the factor of two in Equation 10 results from the 50-percent coupler 330 which splits the light into clockwise and counterclockwise beams.
- the upper power limit results from various undesirable nonlinear processes in the loop fiber, such as Brillouin scattering or four-wave mixing (FWM).
- the upper limit may also be set by limitations in the EDFA power output.
- each channel employs phase-shift keying (PSK), in which 0 and 1 are encoded on a continuous wave (cw) laser by applying a 0 or ⁇ phase shift, respectively, as illustrated in FIG. 7A.
- PSK phase-shift keying
- each bit has a duration of T. This phase modulation spreads the linewidth of the channel.
- ⁇ Q B is the intrinsic Brillouin gain linewidth, which is between 10 and 100 MHz, depending on the fiber doping.
- the value of the Brillouin gain parameter, g B , for silica fiber is 4.6 ⁇ 10 ⁇ 11 meters/watt.
- the critical power, P cr for Brillouin scattering is plotted in FIG. 8 as a function of the channel bit rate. As shown in FIG. 8, P cr is well under a milliwatt for an unmodulated channel. However, P cr is greater than 10 milliwatts for a 2.5 gigabit per second channel.
- the value of P cr is calculated for the case of PSK, it should be understood that the present invention is compatible with other types of signal modulations, as discussed above. The example of PSK is used only to provide a specific value for P cr .
- the methodology for evaluating various configurations of NSA chains compares the resulting allowable gain variation ⁇ G.
- ⁇ G the asymptotic values of G i P i (n) X/2 is plotted as a function of gain in FIG. 9. From the plots, the ranges of gain for which Equation 9 is satisfied is determined. In FIG. 9, the nonlinear coefficient n 2 is varied while the other parameters are held constant. Note that higher values of n 2 force the asymptotic value of G i P i (n) X/2 lower, enabling a larger ⁇ G. If P max is set to 10 milliwatts, as shown by the horizontal line in FIG.
- n 2 5.47 ⁇ 10 ⁇ 20 m 2 /W (a value currently achieved in commercially available fibers) produces a ⁇ G of 1.65 dB.
- a moderate decrease in n 2 by a factor of 2 decreases ⁇ G to 0.4 dB.
- an increase in n 2 by a factor of 2 increases ⁇ G to 3.6 dB (if the present invention is not operated in the oscillatory regime above approximately 27.6 dB).
- Plots of the asymptotic values of G i P i (n) X/2 versus gain can be generated as a function of other parameters such as L or ⁇ bias to further analyze and optimize the NSA chain.
- FIG. 10 illustrates the gain spectra for 16 wavelength division multiplexed channels for a total input power of 1.4 milliwatts (curve a), 0.56 milliwatts (curve b) and 0.22 milliwatts (curve b).
- the total input power is distributed equally amongst the 16 wavelengths, and is input into the EDFA from both sides as would be the case if the EDFA were inserted into a Sagnac loop.
- the parameters of the NSA chain are the same as those discussed above.
- the gain experienced by each of the N input signals that travel through each individual amplifier in each NSA of the chain is first computed. Because the gain in an optical amplifier depends in a generally complex manner on the input signal powers through gain saturation, and because the input signal power spectrum is different for each amplifier, the gain experienced by each one of the N input signals at each amplifier in the chain should be computed to be completely accurate. Although all the gains can be calculated using existing EDFA simulation programs, the calculations are quite time consuming. Instead of calculating all the gains in this manner, the approximation described in D. M. Baney, et al., “WDM EDFA gain characteristics with a reduced set if saturating channels”, IEEE Photonics Technology Letters, Vol.
- Equation 12 G i is the (saturated) gain at wavelength ⁇ i , G s is the (saturated) gain at wavelength ⁇ s , and G i and G s are linear gains (as opposed to gains expressed in dB).
- the saturating tone wavelength ⁇ s can be selected over a broad range of wavelengths without affecting the accuracy of the method. With this method, it is clear that the gain spectrum of the EDFA only needs to be computed for a range of saturating tone powers (instead of having to compute it for any arbitrary combination of N signal powers).
- the unsaturated gain values are inserted in Equation 12 to calculate an approximate value of P s .
- This approximate value of P s is then used in the EDFA simulation program to calculate a second set of (saturated) gains G i and G s . The process is repeated until the gains converge to a stable value. In practice, convergence of the gains is obtained at the second step.
- the EDFA simulated in FIG. 10 is a 7-meter section of erbium-doped fiber bidirectionally pumped by 110-milliwatt, 1,480-nanometer light (220 milliwatts total pump).
- the EDFA exhibits gain spectra as shown in FIG. 10. In the region from 1,544 nanometers to 1,559 nanometers where the 16 WDM signals are located (1-nanometer channel spacing), the typical gain variation was on the order of 1-2 dB, depending on the input signal power.
- FIGS. 11A and 11B show the evolution of the power in all 16 channels as a function of the amplifier number for an NSA chain (FIG. 11A) and a standard amplifier chain (FIG. 11B).
- the loss between sections (X) was increased from 20 dB to 24 dB to account for the 4-dB loss of the nonlinear fiber, which is set to zero in the simulator.
- the EDFA is identical to that used in the NSA chain.
- the NSA chain forces all powers into an asymptotic value such that all channels maintain a power between 3.7 milliwatts and 5.2 milliwatts.
- the convergence occurs after the first few NSAs.
- the channels with the highest gains quickly dominate the other channels. The stabilizing effect of the NSA is therefore clearly seen by comparing FIG. 11A and FIG. 11B.
- FIGS. 12A and 12B The channel power spectrum is shown in FIGS. 12A and 12B for the NSA chain and the standard chain after segment 20 , respectively.
- all channel powers are roughly equal after segment 20 (see FIG. 12A).
- At the output end of the NSA chain see FIG. 12A
- all of the 16 output signals are useful, (i.e., they carry sufficient power to be detected with a high signal-to-noise ratio).
- the output end of the standard amplifier chain see FIG. 12B
- only about 6 of the 16 output signals are useful. All other signals carry too little power and cannot be detected with a sufficiently high signal-to-noise ratio to be useful.
- This comparison illustrates that the present invention produces output signals with approximately equal output powers over a large bandwidth after a large number of gain-loss sections.
- AGC automatic gain control
- It achieves automatic gain control in the sense that it automatically maintains the powers of a large number of signals at the output of a chain of gain-loss sections within a small range of power values, by correcting internally for unpredictable variations in the input signal powers, the pump powers, and the number of signals.
- the overwhelming majority of optical gain flattening schemes developed to date do not exhibit true automatic gain control. Instead, the other schemes rely on additional external feedback schemes (which add complexity and cost to the system) to correct for variations in these quantities.
- the Kerr-induced phase modulation from a second beam was twice that due to self-phase modulation, a feature which enables the NSA chain 400 to flatten the gain.
- the conditions for self-phase modulation and cross-phase modulation are addressed in the context of fiber-optic gyroscopes in B. Crosignani, “Kerr effect and chromatic dispersion in fiber-optic gyroscopes,” Journal of Lightwave Technology, Vol. LT-3, No. 4, pp. 914-918, 1985, wherein it is shown that self-phase modulation and cross-phase modulation are connected with the dispersion in the fiber.
- the information from the Crosignani article is used below to define the conditions for self-phase modulation and cross-phase modulation.
- a characteristic bandwidth, ⁇ c is defined as: ⁇ ⁇ ⁇ v ′ ⁇ 1 2 ⁇ ⁇ ⁇
- the Kerr-induced phase shift of one frequency component at ⁇ 1 on another frequency component at ⁇ 2 can be considered self-phase modulation if
- cross-phase modulation occurs when ⁇ 1 - ⁇ 2
- Equation 14 B is the bit rate 1/T for random PSK, where T is the bit duration.
- the channel spacing has a minimum value which is a function of the loop parameters.
- ⁇ c an optimal value by modifying the group-velocity dispersion, A, of the loop fiber (see Equation 13).
- a typical value for ⁇ ′ can be calculated by taking L eff to be 5 kilometers and A ⁇ 1 to be 3.28 ⁇ 10 ⁇ 28 sec 2 /meter (appropriate for a dispersion shifted fiber near the zero-dispersion wavelength). The two values yield a value of 124 GHz for ⁇ c. This value can be modified easily through the parameter A.
- the nonlinear Sagnac amplifier in accordance with the present invention produces an output signal spectrum that is flat, notwithstanding significant differences in the power of input optical signals at different wavelengths. Furthermore, unlike known gain-flattening schemes, the flatness of the output signal spectrum generally improves with increasing number of gain-loss elements. The flatness of the output signal spectrum is quite good over a large bandwidth. For example, in a chain of nonlinear Sagnac amplifiers with EDFAs that are flat to within approximately 1-2 dB, the flatness after a large number of gain-loss sections is better than 1.5 dB over a very large bandwidth.
- passive filters typically flatten the gain of an individual EDFA to within a few tenths of a dB, which is not sufficient to guarantee a flat gain after traveling through a large number of such EDFAs.
- the output signal spectrum flatness is typically 1.5 dB or worse.
- the NSA chain achieved this same level of spectrum flatness with amplifiers which are flattened to only 1-2 dB.
- the nonlinear Sagnac interferometer is more beneficial as the size of the chain (i.e., the number of amplifiers in the chain) increases. This is particularly advantageous for chains comprising a very large number of amplifiers, such as, for example, a transoceanic communication bus.
- the chain of nonlinear Sagnac amplifiers converges to a flat gain after a number of gain-loss sections.
- the number of amplifiers needed to obtain flat gain e.g., an asymptotic gain value or a percentage (e.g., 90 percent) of the asymptotic gain value
- the ability to control this rate of convergence is significant because in some systems, a fast convergence is needed (e.g., convergence to flat gain after four amplifiers) and in other systems, a slower convergence is acceptable (e.g., convergence to flat gain after twenty amplifiers).
- the system of the present invention is particular advantageous because the output signal spectrum remains very flat even when one or more of the input signals are dropped.
- the present invention keeps the flatness of the output signal spectrum from changing at all or from changing more than a very small amount.
- the gain flatness does not change or changes very little if the power in one or more of the input signals changes (i.e., increases or decreases).
- FIG. 13 shows the power evolution of a 16 channel WDM system in which 8 channels are dropped after the fourteenth and before the fifteenth NSA. The remaining eight channels now see a different gain and they therefore converge to different asymptotic values. The power deviation, however, is quite small, and all 8 remaining channels arrive at the end of the NSA with powers between 5 milliwatts and 5.5 milliwatts.
- the output signal spectrum flatness is also robust to changes in the power that pumps one or more of the EDFAs in the chain of nonlinear Sagnac amplifiers. Changes in any or all of the pump powers cause changes in the optical gain and thus cause changes in the power carried by all the output signals at the output of the chain. However, because of the converging effect of the chain of nonlinear Sagnac amplifiers, the flatness of the output signal spectrum will be nominally unchanged (assuming the pump power changes are reasonably modest in magnitude). This robustness in the flatness of the output signal spectrum is not provided by filter-flattened EDFAs.
- the flattening properties of a nonlinear Sagnac amplifier are more tolerant to manufacturing variations in the parameters of components than most other gain-flattened amplifiers, particularly when compared to gain-flattened EDFAs relying on passive filters in which each EDFA must be flattened with a particular filter matched to the particular gain curve of the EDFA.
- passive filters When passive filters are used, a given filter will flatten the gain of a particular EDFA operated under certain conditions, but the filter will perform poorly, if at all, if it is used with the same EDFA pumped with a different pump power or operated with a different number of signals or operated with different signal powers or a combination of differences.
- the filter may also perform poorly when the same filter is used with two nominally identical EDFAs made with erbium-doped fibers that are slightly different in design, in erbium concentrations, in co-dopant concentrations, in length, or in a combination of differences.
- changes in the filter transmission spectrum e.g., through changes in the external temperature
- a nonlinear Sagnac amplifier constructed in accordance with the present invention with a particular loop fiber and loop length can flatten the gain of EDFAs having a large range of characteristics, and the flatness of the output signal spectrum of the chain is robust to fluctuations in the parameters of the nonlinear Sagnac amplifiers.
- the figures presented herein illustrate the performance of NSAs with loop lengths of 10 km. It should be understood that the choice of loop length impacts the system performance. For example, there is a tradeoff between length and mode confinement. In particular, the stronger the mode confinement, the shorter the length that is required. There is also a tradeoff between the loop length and the fiber nonlinearity. In particular, the stronger the fiber nonlinearity, the shorter the length that is required. Similarly, there is also a tradeoff between the loop length and the signal power. In particular, the stronger the signal power, the shorter the length that is required.
- NSAs with loop lengths either substantially shorter or substantially longer than 10 km can be designed to operate with nominally the same characteristics as described in this invention.
- the general principles of the nonlinear Sagnac amplifier described herein are not restricted to trivalent erbium (Er 3+ ) as the active dopant.
- the nonlinear Sagnac amplifier can also be used to flatten the gain of any optical amplifier based upon a fiber, a waveguide or a bulk optical component.
- the optical amplifier may use any active ion or active laser medium, and may operate over any portion of the spectral range.
- the nonlinear Sagnac amplifier may be used with Raman amplifiers, which may be used, for example, for amplification applications not suitably covered by other optical amplifiers.
- a Raman amplifier or other amplifier used in connection with the present invention should be bidirectional so that it can amplify the counterpropagating signals in the Sagnac loop.
- the Raman amplifier can be operated in a resonant mode or it may be pumped in two different directions.
- a single EDFA may have a limited amplification range.
- a first amplifier AMP1 may amplify over one band of wavelengths and may absorb over a second band of wavelengths.
- a second amplifier AMP2 may absorb over the first band and may amplify over the second band.
- the two amplifiers AMP1 and AMP2 may be positioned in the loop in parallel as a composite amplifier by using two WDM couplers.
- the nonlinear Sagnac amplifier of FIG. 3 is utilized with an amplifier 360 which is made of the parallel arrangement of the two amplifiers AMP1 and AMP2 shown in FIG. 14A, wherein AMP1 and AMP2 are two individual amplifiers configured, for example, like the amplifier of FIG. 3A.
- a first amplifier AMP1 amplifies a first band and passes a second band and a second amplifier AMP2 passes the first band and amplifies the second band
- the two amplifiers may be placed in series in the loop to provide amplification over both bands.
- This composite amplifier is illustrated in FIG. 14B.
- the individual amplifiers AMP1 and AMP2 can both be configured, for example, like the amplifier of FIG. 3A.
- the nonlinear Sagnac amplifier then consists of the nonlinear Sagnac amplifier of FIG. 3 comprising an amplifier 360 which is made of the composite amplifier of FIG. 14B.
- NSA1 and NSA2 represent two nonlinear Sagnac amplifiers that provide gain over different (and perhaps adjacent) portions of the spectrum (for example 1300 nanometers to 1350 nanometers for NSA1 and 1500 nanometers to 1650 nanometers for NSA2).
- the first WDM coupler acts to separate the input signals and direct them to the appropriate nonlinear Sagnac amplifier, namely to direct signals with wavelengths between 1300 nanometers and 1350 nanometers (in this example) to NSA1 and to direct signals with wavelengths between 1500 nanometers and 1650 nanometers to NSA2.
- the second WDM coupler recombines the signals exiting NSA1 and NSA2 onto the communication fiber (represented by the loss box in FIG. 15).
- a gain-loss string 1600 comprises both nonlinear Sagnac amplifiers (NSAs) 1610 and standard optical amplifiers (AMPs) 1620 separated by loss sections 1630 .
- NSAs nonlinear Sagnac amplifiers
- AMPs standard optical amplifiers
- the NSAs 1610 and the AMPs 1620 are positioned alternatively (i.e., every other amplifier).
- the percentage of nonlinear Sagnac amplifiers in the chain illustrated in FIG. 16 is thus 50% (i.e., one nonlinear Sagnac amplifier 1610 for every standard amplifier 1620 ).
- the percentage of nonlinear Sagnac amplifiers 1610 in the chain can be smaller (for example, every third amplifier in the chain can be a nonlinear Sagnac amplifier 1610 ) or larger (for example, every third amplifier in the chain can be a standard amplifier 1620 ) depending on the degree to which the standard amplifiers 1620 are individually flattened. For example, if the standard amplifiers 1620 are not flattened well, only one standard amplifier 1620 may be used for every few nonlinear Sagnac amplifiers 1610 . If the standard amplifiers 1620 are moderately well flattened, every other amplifier may be a standard amplifier 1620 .
- This hybrid chain 1600 is advantageous because (1) it maintains a high gain flatness even after a very large number of gain-loss sections, and (2) it makes the amplifier-loss chain robust to changes (intentional or unintentional) in the input conditions (i.e., the number of signals, the signal power, or the pump power).
- the hybrid chain 1600 is beneficial because by substituting relatively inexpensive standard amplifiers 1620 for the more expensive nonlinear Sagnac amplifiers (which have the fiber loops and optical circulators), a significant cost reduction can be obtained.
- the present invention can also be used with individual amplifiers having a gain which exhibits either no or little wavelength dependence.
- the invention works even better (i.e., it provides a flatter gain after fewer amplifiers) if the gain spectra of the individual amplifiers are flatter.
- the present invention can also be used with individual amplifiers with a gain spectrum that is nominally flat (e.g., to within 0.3 dB over an 80-nanometer bandwidth), which can be accomplished by a number of techniques well known in the art.
- the invention performs the function of maintaining or even improving the gain flatness after a large number (e.g., tens) of amplifiers.
- the amplifier string yields an increasing deterioration of the gain flatness as the number of gain-loss section is increased.
- the present invention when using the present invention with individual amplifiers, which each has a gain spectrum that is nominally flat (e.g., to within 0.3 dB over an 80-nanometer bandwidth), the present invention provides a nominally flat gain spectrum that is robust against changes in the signal power, against changes in the number of signals, and, to some degree, against changes in the pump power.
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Electromagnetism (AREA)
- Nonlinear Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Optics & Photonics (AREA)
- General Physics & Mathematics (AREA)
- Power Engineering (AREA)
- Plasma & Fusion (AREA)
- Lasers (AREA)
- Optical Modulation, Optical Deflection, Nonlinear Optics, Optical Demodulation, Optical Logic Elements (AREA)
- Optical Communication System (AREA)
Abstract
An apparatus and a method provide gain flattening in communications systems wherein a large number of optical signals at different wavelengths must be amplified while maintaining signal power within an acceptable range. Because of differences in gain of typical optical amplifiers as a function of wavelength and input power, the signals at different wavelengths are not amplified by the same amounts. Thus, when amplified multiple times, certain signals tend to become severely attenuated to the point of being no longer useable. The present gain flattening apparatus and method cause signals having higher gain-power products to be attenuated by a greater amount in response to Kerr-induced phase shifting such that after multiple stages of amplification, all the signal powers converge toward a small range of acceptable output powers. The apparatus provides amplification, multiple times, of a series of signals with a plurality of wavelengths covering a very wide spectral range, while maintaining the power of all the signals within a small range. The spread of this signal power range is robust against changes in the signal power, against changes in the number of signals, and, to some degree, against changes in the amplifier's pump power. The apparatus design is also robust against manufacturing changes in the parameters of the apparatus' components. The apparatus and invention are preferably implemented as multiple nonlinear Sagnac amplifiers having erbium-doped fiber amplifiers positioned asymmetrically in an interferometer loop.
Description
- This application is a continuation of U.S. patent application Ser. No. 09/537,639, filed on Mar. 29, 2000, which claims the benefit of priority under 35 U.S.C. §119(e) from U.S. Provisional Patent Application No. 60/146,744 filed on Aug. 2, 1999, which are incorporated herein by reference.
- 1. Field of the Invention
- The present invention is in the field of fiber optic communication systems, and, more particularly, is in the field of amplifiers for use in fiber optic communication systems.
- 2. Description of the Related Art
- In order to satisfy the increasing demand for bandwidth, fiber optic communication systems are moving towards wavelength-division multiplexing (WDM), in which many channels at separate wavelengths are carried on the same fiber. Integral to most current fiber communication systems is the incorporation of erbium-doped fiber amplifiers (EDFAs). When EDFAs are integrated into WDM systems, the gain flatness of the amplifiers becomes critical. For example, a
fiber communication line 100 comprising a serial chain of loss sections 110(i) and gain sections 120(i) is shown in FIG. 1A. In FIG. 1A, the loss sections 110(i) are represented as lengths of fiber, and the gain sections 120(i) are represented by EDFAs. Each loss section 110(i) and an associated gain section 120(i) is referred to herein as a “loss-gain section.” - In a WDM system, multiple independent channels are input at discrete wavelengths. The gain of EDFAs is a function of the wavelength, with a typical unfiltered gain variation appearing in FIG. 1B. After propagation through the first loss-gain section110(1), 120(1), the channels around 1532 nanometers have higher powers than the other channels due to the EDFA gain variation. Propagating through multiple loss-gain sections causes this disparity in channel powers to grow, and eventually causes the power of some channels out of the last loss-gain section 110(n), 120(n) to drop to unacceptable levels.
- To illustrate the foregoing process, the input power spectrum is plotted in FIG. 2A, and the output power spectrum after 5 loss-gain sections (i.e., n=5 in FIG. 1A) is plotted in FIG. 2B. The differences in the two spectra in FIGS. 2A and 2B illustrate the large disparity in resulting channel powers caused by EDFA gain variations.
- Many prior solutions to this problem have involved adding a filter to the EDFA to produce a gain-filter section which is flatter than the gain of the EDFA alone. However, fabricating a filter with a correct shape, which is independent of EDFA parameters (e.g., signal and pump powers) and which is stable with time and temperature, is not trivial.
- The present invention is directed to gain flattening with nonlinear Sagnac amplifiers. The use of nonlinear Sagnac amplifiers for gain flattening is a novel solution. Instead of using a filter which has a loss which varies as a function of wavelength, the present invention is directed to a filter which has a loss that is a function of power. More particularly, the filter in accordance with the present invention attenuates a specific channel i based on the power of that channel. The filter in accordance with the present invention does not provide a loss that is a conventional broadband power-dependent loss. Rather, the filter in accordance with the present invention provides a narrowband power-dependent loss. In other words, the attenuation of the filter at λi, the wavelength of channel i, is a function of the power around that wavelength (i.e., at λi±δO), but is not a function of the power at a separate wavelength λi±n outside of the λi±δO window. Such a filter is achieved by replacing a standard linear amplifier, as depicted in FIG. 1A, with a nonlinear Sagnac amplifier (NSA), which will be described in detail below.
- One aspect of the present invention is an amplification system for reducing power differences in a plurality of output optical signals responsive to a plurality of input optical signals having a plurality of respective optical wavelengths and having a plurality of respective input powers. The amplification system comprises an interferometric loop. A coupler couples the plurality of input optical signals to the loop to cause respective first portions of the input optical signals to propagate in a first direction in the loop and to cause respective second portions of the input optical signals to propagate in a second direction in the loop. The coupler combines the first and second portions after the first and second portions propagate in the loop to produce a plurality of output optical signals. An amplifier is located at an asymmetric location with respect to the center of the loop. The amplifier has a gain spectrum which causes the amplifier to have a plurality of respective gains at the plurality of optical wavelengths. The asymmetric location of the amplifier with respect to the center of the loop causes differences in powers of the first signal portions and the second signal portions of the input optical signals while these portions are traveling through the interferometric loop. The differences in powers of the first and second signal portions cause respective phase shifts in the first and second signal portions to occur in the fiber loop due to the optical Kerr effect. The Kerr-induced phased shifts vary in response to differences in the respective input powers and the respective gains to cause a greater Kerr-induced attenuation of input optical signals having a greater gain-power product. Preferably, the amplifier comprises an erbium-doped fiber amplifier. Certain preferred embodiments further include a wavelength division multiplexed coupler in the loop proximate to the amplifier. A pump source is coupled to the wavelength division multiplexed coupler to provide pump light for the amplifier via the wavelength division multiplexed coupler.
- Another aspect of the present invention is an amplification system for reducing output power differences in a plurality of output optical signals responsive to a plurality of input optical signals having a plurality of respective optical wavelengths and having a plurality of respective input powers. The amplification system comprises an interferometric loop which has first and second lengths of optical fiber separated by an optical amplifier. The first length of optical fiber is substantially longer than the second length of optical fiber. A coupler couples the optical signals into the interferometric loop to cause respective first and second portions of the optical signals to counterpropagate in first and second directions in the interferometric loop. The coupler combines the respective first and second portions of the optical signals after propagation through the interferometric loop to produce a plurality of respective output signals at the plurality of optical wavelengths. The plurality of output signals have a plurality of respective output powers. The amplifier has a gain characteristic which causes the amplifier to have a plurality of respective gains at the plurality of optical wavelengths. The first and second portions of the optical signals propagating in the first and second directions experience respective Kerr-induced phase shifts caused by self-phase modulation, by copropagating cross-phase modulation, and by counterpropagating cross-modulation. The location of the amplifier causes light propagating in the first direction to pass through the first length of optical fiber before propagating through the amplifier and the second length of optical fiber. The location of the amplifier also causes light propagating in the second direction to propagate through the second length of optical fiber and the amplifier before propagating through the first length of optical fiber. The location of the amplifier also causes the light propagating in the first direction to experience greater counterpropagating cross-phase modulation than the light propagating in the second direction. The location of the amplifier also causes the light propagating in the second direction to experience greater self-phase modulation and greater copropagating cross-modulation than light propagating in the first direction. The Kerr-induced phase shifts of the plurality of optical signals at the plurality of optical wavelengths are responsive to the respective amplifier gains at the plurality of optical wavelengths and are further responsive to the respective input powers of the plurality of optical signals, such that differences in the output powers caused both by differences in the input powers and by differences in the gains at the plurality of optical wavelengths are reduced.
- Another aspect of the present invention is an optical amplification system which comprises a serial chain of at least first and second amplification sections. The first amplification section is coupled to receive a plurality of input optical signals. Each input optical signal has a respective optical wavelength and a respective input optical power. Each amplification section includes a respective amplifier therein. The amplifiers have respective gain characteristics such that gains applied to the optical signals vary with wavelength. The first amplification section provides a first plurality of optical output signals. The second amplification section is coupled to receive the first plurality of optical output signals and to provide a second plurality of optical output signals. Each of the second plurality of optical output signals has a respective optical wavelength and has a respective output optical power. The amplification sections operate to reduce differences in the respective output optical powers of the second plurality of optical output signals caused by differences in the input optical powers and by differences in gains applied to the optical signals. Each amplification section comprises an interferometric loop. The amplifier of the amplification section is located asymmetrically in the loop. A coupler couples light to the interferometric loop to cause the light to propagate as first and second counterpropagating portions at each of the optical wavelengths. The coupler also combines the first and second counterpropagating portions at each of the wavelengths after the first and the counterpropagating portions of the light have propagated through the loop. The first and second counterpropagating portions at each of the wavelengths interfere to provide an output signal at each of the wavelengths. The output signal at each of the wavelengths has a power responsive to input power at the wavelength, responsive to amplifier gain at the wavelength, and responsive to Kerr-induced phase shift at the wavelength. The Kerr-induced phase shift is greater for optical wavelengths having greater gain-power products to at least partially reduce differences in output power caused by differences in gain-power products. Preferably, each of the amplifiers comprises an erbium-doped fiber amplifier. Also preferably, the amplifiers have gains which vary with optical wavelength. The optical signals at the plurality of wavelengths have varying powers. The amplification sections operate to cause output powers at each of the wavelengths at the output of the serial chain to converge toward nominally the same output power within a selectable range of output power.
- Another aspect of the present invention is a method of amplifying a plurality of input optical signals. The input optical signals have respective optical wavelengths within a range of optical wavelengths and have respective optical powers within a range of input optical powers. The method produces a corresponding plurality of output optical signals having respective output powers within a selected range of output optical powers. The method comprises passing the input optical signals through a first nonlinear Sagnac amplifier to produce a plurality of intermediate optical signals. The plurality of intermediate optical signals are responsive to the plurality of input optical powers, to amplifier gain and to Kerr-induced phase shift to have a respective plurality of intermediate optical powers within an intermediate range of optical powers. The intermediate range of optical powers is a smaller range than the range of input optical powers. The method further comprises passing the intermediate optical signals through at least a second nonlinear Sagnac amplifier to produce the plurality of output optical signals. The plurality of output optical signals are responsive to the plurality of intermediate optical powers, to amplifier gain and to Kerr-induced phase shift to have a plurality of output optical powers within the selected range of output optical powers. The selected range of output optical powers is smaller than the range of intermediate optical powers and is smaller than the range of input optical powers.
- Another aspect of the present invention is an optical system that comprises an input that receives optical input signals having wavelengths within a range of wavelengths. An output outputs optical signals responsive to the optical input signals. At least a first amplifier is interposed between the input and the output. The first amplifier has a first gain spectrum. The first gain spectrum varies with wavelength over the range of wavelengths to cause a first optical signal produced by the first amplifier at a first wavelength within the range of wavelengths to have a first optical power and to cause a second optical signal produced by the first amplifier at a second wavelength within the range of wavelengths to have a second optical power. The first optical power and the second optical power differ by a power difference. The system further comprises at least a second amplifier interposed between the first amplifier and the output. The second amplifier comprises a nonlinear Sagnac amplifier. The second amplifier operates to at least partially compensate for differences in gain of the first amplifier over the range of wavelengths to reduce the power difference between the first optical signal and the second optical signal.
- Another aspect of the present invention is an optical system that comprises an input that receives optical input signals having wavelengths within a range of wavelengths and that comprises an output that outputs optical output signals responsive to said optical input signals. A first plurality of amplifiers of a first type are interposed between the input and the output. The amplifiers of the first type have a first gain spectrum that varies over the range of wavelengths to cause a first optical signal at a first wavelength within the range of wavelengths to be amplified with a first gain by the amplifiers of the first type and to cause a second optical signal at a second wavelength within the range of wavelengths to be amplified with a second gain by the amplifiers of the first type. The second gain differs from the first gain. A second plurality of nonlinear Sagnac amplifiers are interposed at selected locations in the optical system. The nonlinear Sagnac amplifiers operate to at least partially compensate for differences in the first gain and the second gain of the amplifiers of the first type to reduce power differences between the first optical signal and the second optical signal.
- The present invention will be described in detail below on connection with the attached drawings in which:
- FIG. 1A illustrates a linear amplifier chain comprising a plurality of multiple loss-gain sections, wherein the loss sections comprise fiber sections and the gain sections comprise erbium doped fiber amplifiers (EDFAs);
- FIG. 1B illustrates the gain variation of an exemplary erbium doped fiber amplifier as a function of wavelength;
- FIG. 2A illustrates the input power spectrum of a wavelength division multiplexed (WDM) system with an input signal having a channel spacing of 2 nanometers;
- FIG. 2B illustrates the output power spectrum resulting from the input signal spectrum of FIG. 2A being applied to five gain-loss sections wherein the gain variation as a function of wavelength is shown in FIG. 1B;
- FIG. 3 illustrates an embodiment of a nonlinear Sagnac amplifier in accordance with the present invention;
- FIG. 3A illustrates a preferred embodiment of an optical amplifier of FIG. 3.
- FIG. 4 illustrates a chain of nonlinear Sagnac amplifiers in accordance with the present invention;
- FIG. 5A illustrates a transfer function of an exemplary nonlinear Sagnac amplifier in accordance with the present invention, wherein the gain is 21.8 dB, wherein Pi (n) is the power of channel i at the output of the gain-loss section n and Pi (n+1) is the power of channel i at the output of the gain-loss section n+1, and wherein the output power converges to a steady state value;
- FIG. 5B illustrates the convergence of the output power Pi (n) of the gain-loss sections as a function of n for various initial values of Pi (0) for the transfer function of FIG. 5A;
- FIG. 5C illustrates a transfer function of an exemplary nonlinear Sagnac amplifier in accordance with the present invention, wherein the gain is 24 dB, wherein the increased nonlinearity of the transfer function with increased gain is shown, and wherein Pi (n) does not converge to a steady state value;
- FIG. 5D illustrates the lack of convergence of the output power Pi (n) of the gain-loss sections as a function of n for various initial values of Pi (0) for the transfer function of FIG. 5C;
- FIG. 6 illustrates a graph of the asymptotic behavior of Pi (n) as a function of the amplifier gain, showing that as the gain increases the power in the channel oscillates and then becomes chaotic;
- FIG. 7A illustrates an example of phase shift keying (PSK);
- FIG. 7B illustrates the resulting spread in the channel spectrum for a pseudo-random bit pattern applied to a PSK system, wherein the amplitude is normalized in linear units;
- FIG. 8 illustrates Pcr, the critical power in milliwatts above which stimulated Brillouin scattering become significant, as a function of the bit rate, wherein the bit rate is shown in megabits per second on a logarithmic scale;
- FIG. 9 illustrates a plot of the asymptotic values of GiPi (n)X/2 as a function of gain;
- FIG. 10 illustrates the gain spectra for 16 wavelength division multiplexed channels for a total input power of 1.4 milliwatts (curve a), 0.56 milliwatts (curve b) and 0.22 milliwatts (curve b)
- FIGS. 11A and 11B illustrate the evolution of the power in the 16 WDM channels of FIG. 10 for a chain of nonlinear Sagnac amplifiers (FIG. 11A) and for a chain of standard amplifiers (FIG. 11B);
- FIG. 12A illustrates the channel power spectrum for a chain of nonlinear Sagnac amplifiers after 20 segments;
- FIG. 12B illustrates the channel power spectrum for a chain of standard amplifiers after 20 segments;
- FIG. 13 illustrates the power evolution of a 16-channel WDM system in which 8 channels are dropped after the fourteenth and before the fifteenth nonlinear Sagnac amplifier;
- FIG. 14A illustrates an embodiment of an EDFA for use in the loop of a nonlinear Sagnac amplifier in which the EDFA comprises two amplifiers in parallel for amplifying two optical bands;
- FIG. 14B illustrates an embodiment of an EDFA for use in the loop of a nonlinear Sagnac amplifier in which the EDFA comprises two amplifiers in series for amplifying two optical bands;
- FIG. 15 illustrates a configuration having two nonlinear Sagnac amplifiers in parallel to provide amplification over two different optical bands; and
- FIG. 16 illustrates an alternative embodiment to the embodiment of FIG. 4, wherein selected ones of the nonlinear Sagnac amplifiers in a chain are replaced with standard amplifiers.
- FIG. 3 illustrates an embodiment of a nonlinear Sagnac amplifier (NSA)300 in accordance with the present invention. The
NSA 300 comprises aSagnac interferometer 310 which is modified in accordance with the following description. - The
NSA 300 comprises an input optical fiber 312 (or other suitable waveguide) which is coupled to a first port (A) of anoptical circulator 314. Theoptical circulator 314 has a second port (B) and a third port (C). Theoptical circulator 314 operates in a known manner to cause light provided as an input to one port to propagate in a single direction (e.g., clockwise in FIG. 3) to the next port where substantially all the light is output from the circulator. In particular, in FIG. 3, light provided as an input to port A of thecirculator 314 is output from port B of thecirculator 314. Light provided as an input to port B of thecirculator 314 is output from port C of thecirculator 314. No light is provided as an input to port C of thecirculator 314. Light does not propagate in the counterclockwise direction in thecirculator 314. An exemplaryoptical circulator 314 is a polarization-independent fiber optic circulator, such as a Model PIFC23A211000 circulator available from E-Tek Dynamics, Inc., of San Jose, Calif. (Although the preferred embodiment of the present invention includes theoptical circulator 314, thecirculator 314 can be replaced by a standard 3-dB coupler (not shown) to provide a less expensive component for inputting signals to and returning signals from theSagnac interferometer 310; however, it will be appreciated that such a coupler will induce splitting losses, which losses may be acceptable in some applications.) - Port C of the
circulator 314 is connected to anoutput fiber 320 which provides anoutput port 1. Port B of thecirculator 314 is connected via afiber section 322 to afirst port 332 on the input/output side of a conventional 2×2optical coupler 330. Thecoupler 330 has asecond port 334 on the input/output side of thecoupler 330. Thecoupler 330 has athird port 336 and afourth port 338 on the loop side of thecoupler 330. Thesecond port 334 of the 2×2optical coupler 330 is connected to anoutput fiber 340 which provides anoutput port 2. - The
third port 336 of thecoupler 330 is coupled to afirst end 352 of anoptical fiber loop 354. The length of theoptical fiber loop 354 is selected to provide a third-order nonlinear effect, as discussed below. Theoptical fiber loop 354 has asecond end 356 which is coupled to afirst port 362 of an optical amplifier (AMP) 360. Theoptical amplifier 360 hassecond port 364 which is coupled to thefourth port 338 of the 2×2coupler 330 via alength 366 of optical fiber. Theoptical amplifier 360 is a bidirectional amplifier which amplifies light entering thefirst port 362 and provides the amplified light as an output on thesecond port 364. Similarly, theoptical amplifier 360 amplifies light entering thesecond port 364 and provides the amplified light as an output on thefirst port 362. - In particularly preferred embodiments, the
optical amplifier 360 of FIG. 3 comprises an erbium doped fiber amplifier (EDFA). A preferred embodiment of theoptical amplifier 360 implemented as an EDFA is illustrated in FIG. 3A. As illustrated in FIG. 3A, theoptical amplifier 360 comprises a length ofoptical fiber 370 which has afirst end 372 and asecond end 374. Thefiber 370 is doped with erbium. In alternative embodiments, thefiber 370 may be doped with another rare-earth element. - The
second end 374 of thefiber 370 is connected to thesecond port 364 of theoptical amplifier 360. Thefirst end 372 of thefiber 370 is connected to a wavelength division multiplexing (WDM)coupler 380. TheWDM coupler 380 has afirst port 381, asecond port 382, athird port 383 and afourth port 384. Thefirst end 372 of theoptical fiber 370 is connected to thefirst port 381. Thethird port 383 is opposite thefirst port 381, and thethird port 383 is connected to thefirst port 362 of theoptical amplifier 360. - The
fourth port 384 of theWDM coupler 380 is also opposite thefirst port 381 and is connected to apump source 386. In the embodiment described herein, thepump source 386 advantageously comprises a light source operating at a wavelength of approximately 980 nanometers or 1,480 nanometers. In particularly preferred embodiments, thepump source 386 comprises a laser diode, or the like. - The
second port 382 of theWDM coupler 380 is on the same side of theWDM coupler 380 as thefirst port 381. Thesecond port 382 can be advantageously non-reflectively terminated by aterminator 388. - The
WDM coupler 380 is configured to propagate light entering the coupler at the wavelength in the range of approximately 1,500 nanometers to approximately 1,650 nanometers or higher from thefirst port 381 to thethird port 383 and from thethird port 383 to thefirst port 381. Very little light in that range of wavelengths is coupled to thefourth port 384 or to thesecond port 382. TheWDM coupler 380 is further configured to couple light entering thefourth port 384 at the pump wavelength (either around 980 nanometers or around 1,480 nanometers) to thefirst port 381. Because of the configuration of theWDM coupler 380, little, if any, pump light is coupled to thesecond port 382. Because of thenon-reflective terminator 388, any such light will not be reflected back into thesecond port 382. - It can be seen that the
WDM coupler 380 operates to couple pump light from thepump source 386 to the erbium-dopedfiber 370 via thefirst port 381. In contrast, substantially no light at the signal wavelengths is coupled either to thesecond port 382 or to thefourth port 384. Thus, substantially all light at the signal wavelengths remains in the interferometric loop. - As shown in FIG. 3, the 2×2
optical coupler 330 and the fiber elements interconnecting thethird port 336 and thefourth port 338 form theSagnac interferometer 310, the operation of which is described below. - An input signal to be amplified is applied to the
input fiber 312. The input signal comprises light having a plurality of wavelengths λi (i=1, 2, . . . n). In a first preferred implementation of the present invention in communication systems, the signals are modulated either in phase or in frequency. Examples of such systems include phase-shift keying (PSK) and frequency-shift keying (FSK). In a second preferred implementation of the present invention in communication systems, the signals are modulated in amplitude; however, in such an implementation, limitations may apply due to pulse distortion and channel cross-talk. - The input signal propagates to port A of the
circulator 314. Thecirculator 314 circulates the input signal to port B, and the input signal propagates via thefiber 322 to the first input/output port 332 of the 2×2optical coupler 330. The 2×2optical coupler 330 operates in a conventional manner to divide the input light into two substantially equal portions, wherein a first portion is coupled to thethird port 336 and a second portion is coupled to thefourth port 338. - The first portion of the input signal propagates from the
third port 336 of the 2×2optical coupler 330, through theloop 354 in a clockwise (CW) direction, to thefirst port 362 of theoptical amplifier 360. The light is amplified in theoptical amplifier 360, and the amplified first portion propagates from thesecond end 364 of theoptical amplifier 360 to thefourth port 338 of the 2×2optical coupler 330 via the length ofoptical fiber 366. - The second portion of the input signal propagates from the
fourth port 338 of the 2×2optical coupler 330, propagates to thesecond end 364 of theoptical amplifier 360 and is amplified therein. The amplified second portion then propagates from thefirst end 362 of theoptical amplifier 360 and propagates through theloop 354 in the counterclockwise (CCW) direction to thethird port 336 of the 2×2optical coupler 330. - The amplified first portion of light entering the
fourth port 338 of the 2×2optical coupler 330 and the amplified second portion of light entering thethird port 336 of the 2×2optical coupler 330 are interferometrically combined therein. Depending on the phase relationship between the two signal portions, the combined signal is output from thefirst port 332 or from thesecond port 334 or from both ports of the 2×2optical coupler 330. Light output from thesecond port 334 propagates to theoutput port 2 via theoutput fiber 340. Light output from thefirst port 332 propagates to port B of thecirculator 314, is coupled to port C of thecirculator 314, and propagates to theoutput port 1 via thefiber 320. The power in theoutput port 1 and the power in theoutput port 2 thus depend on the phase relationship between the clockwise propagating and counterclockwise propagating signals. - As described above, the
optical amplifier 360 is positioned asymmetrically in theinterferometer 310 such that theoptical amplifier 360 is optically closer to thefourth port 338 of the 2×2optical coupler 330 than it is to thethird port 336 of the 2×2optical coupler 330. Because of the asymmetrical positioning of theoptical amplifier 360, light entering theinterferometer 310 via thefourth port 338 of the 2×2optical coupler 330 is amplified by theoptical amplifier 360 before propagating (in the counterclockwise direction) through theloop 354 and exiting via thethird port 336. In contrast, light entering theinterferometer 310 via thethird port 336 propagates (in the clockwise direction) through theloop 354 before being amplified by theoptical amplifier 360 and exiting via thefourth port 338. Thus, for the configuration shown in FIG. 3, the light propagating in the counterclockwise direction within theloop 354 experiences the gain of theoptical amplifier 360 before propagating in theloop 354, and the light propagating in the clockwise direction within theloop 354 experiences the gain of theoptical amplifier 360 after propagating in theloop 354. As discussed in more detail below, this difference in when the two counterpropagating light signals experience the gain of theoptical amplifier 360 is advantageously used to achieve gain flattening. - As will be shown below, the desired filter performance is obtained by using the Kerr-induced phase shift generated in the
Sagnac loop 354 to vary the gain in accordance with the power in each wavelength of the optical signal applied to theloop 354. -
- In
Equation 1, Pself is the power in channel i, Pother is any other power in the fiber, dl is the interaction length, and Aeff is the effective mode area. (A criteria for defining what the power is in channel i is presented below.) The parameter n2 is the nonlinear coefficient, and the parameter κi is the k vector for the signal at the wavelength λi (i.e., κi=2π/λi). Note that the Kerr-induced cross-phase modulation (phase modulation due to other channels) is a factor of two greater than the self-phase modulation (i.e., the phase modulation due to itself). It is assumed that all copropagating light within a given channel produces self-phase modulation on that channel and that all remaining light (including all other copropagating channels along with all counterpropagating channels) produces cross-phase modulation. - To calculate the effect of the Kerr-induced phase shift, it is assumed that N channels are launched into
port 312 of the NSA shown in FIG. 3. The length of theloop 354 is L, the gain of the amplifier at λi is Gi, and Pi is the input signal power at λi in theport 312. -
- In
Equation 2, thefactor 2 in the denominator is the result of splitting each input signal power into two portions by thecoupler 330. The first term within the parentheses is the self-phase modulation. The second term within the parentheses is the cross-phase modulation due to other clockwise traveling channels. The third term within the parentheses is the cross-phase modulation due to all the counterclockwise traveling channels. Note that the third term has a factor of Gj because each CCW traveling channel has experienced the gain Gj of theamplifier 360 before traveling theSagnac loop 354. The parameter Leff is the effective loop length and is defined by Leff=(1/α)[1−e−αL], where α is the fiber attenuation (assumed to be wavelength independent for simplicity). -
- In
Equation 3, the gain Gi is applied to the first two terms because the first two terms represent counterclockwise propagating signals, which experience the gain Gi before propagating through theloop 354. Therefore, fromEquation 2 andEquation 3, the Kerr-induced phase difference for channel i is given as: - In
Equation 4, the numerical value of the nonlinear coefficient n2 depends on several parameters, in particular the fiber material and the nature and concentration of the dopants it may contain. For example, for a nearly pure silica fiber core, the value of n2 is typically around 2.4×10−16 cm2/W in the 1.3-1.6 μm spectral region. In silica-based fibers heavily doped with germanium, this value can be enhanced by a factor of a few. -
Equation 4 is applicable, in particular, when all the signals involved have the same state of polarization. However, if all the signals involved do not have the same state of polarization,Equation 4 is still applicable, but the value of n2 that must be used will depend on the relative polarization of the interacting signals. If the fiber loop of the present invention is made with a standard, low-birefringence fiber, the state of polarization of the signal in channel i will evolve along the fiber length. The first reason for this evolution is that the orientations of the axes of birefringence of the fiber vary along the length of the fiber. The second reason is that the orthogonal slow and fast polarization components of a particular channel i propagate at slightly different velocities and therefore experience slightly different phase shifts. Furthermore, because each channel has a different wavelength, and perhaps a slightly different input state of polarization, the state of polarization of each channel will evolve according to its own distribution. Therefore, in general, at any given point along the fiber the state of polarization of a given channel i, the state of polarization of all other copropagating channels j≠i, and the state of polarization of all counterpropagating channels j, will all be different. Because the index change due to the Kerr effect depends on the state of polarization of these various signals, in a long fiber loop (typically longer than a few tens of meters), the value of n2 needs to be revisited. - The dependence of the Kerr-induced index change on polarization is rather complex (see, for example, C. B. Clausen, et al., “Polarization sensitivity of the nonlinear amplifying loop mirror,”Optics Letters, Vol. 21, No. 19, October 1996, pp. 1535-1537). In a long enough optical fiber (again typically a few tens meters or longer), the state of polarization of each of the N signals will go through all possible states, while the fiber birefringent axes will vary through all possible spatial orientations. In this case, the Kerr-induced index change can be calculated by averaging first the relative phase of each signal, and second the spatial orientation of the fiber birefringent axes, as described in the Clausen, et al., reference cited above. Applying the methodology described in the Clausen, et al., reference to the loop configuration with imbalanced counterpropagating signal powers specific to the present invention, it can be shown that
Equation 4 still applies, provided that n2 is replaced by a smaller effective nonlinear coefficient n2,eff. - Given the two aforementioned dependencies of n2, namely the dependence on signal polarization and the dependence on fiber composition, in the numerical simulations presented herein, unless otherwise specified we used a value of n2eff=5.47×10−20 cm2/W. Although this value is well within the range of reported effective nonlinear coefficients, it is important to realize that it will vary, sometime substantially, from fiber to fiber. However, it is equally important to realize that the specific value of n2eff used in these simulations is not critical to either the principle of the present invention or its practical applicability. The present invention can be utilized with fibers exhibiting either considerably smaller or orders of magnitude larger effective nonlinearities, provided the length of the fiber is adjusted accordingly. The present invention can also be utilized with optical waveguides and/or bulk materials exhibiting either considerably smaller or orders of magnitude larger effective third-order nonlinearities.
- In
Equation 4, the Kerr-induced phase difference is a function of only the power in channel i, which is exactly the narrowband behavior needed. The output of a Sagnac interferometer is a function of the phase difference between the interfering signals. Thus, the output of theSagnac amplifier 300 is a function of this Kerr-induced phase difference as well as any phase bias between the two counterpropagating light signals. As discussed below, the Kerr-induced phase difference is converted into a wavelength-dependent attenuation via the operation of theSagnac interferometer 310. -
- In
Equation 5, φbias is the bias phase shift of theSagnac interferometer 310 other than the Kerr effect, which, when the light is initially input to theport 312, is naturally 0 foroutput port 1 and π foroutput port 2. Using the Kerr-induced phase difference calculated above (Equation 4), the transfer function becomes: - In
Equation 6, it can be seen that Pi out is nonlinearly related to Pi. In particular, if the phase bias φbias and/or the length Leff are selected properly, as the gain-power product GiPi increases, the effect of the Kerr-induced phase difference is to multiply the gain-power product by a smaller and smaller ratio, so that increases in the gain-power product do not result in proportionate increases in the output power. Furthermore, the output power Pi out inEquation 6 depends on the power Pi in channel i, but it does not depend on the powers Pi carried by any of the other channels j≠i. Thus, this effect, namely the fact that an increase in the gain-product does not result in a proportionate increase in the output power, is not affected by the power in the other channels. In other words, each channel acts on itself independently of the other channels. Thus,Equation 6 shows that thenonlinear Sagnac amplifier 300 achieves a narrowband power-dependent transfer function (gain). - In FIG. 4, a plurality of nonlinear Sagnac amplifiers (NSAs)300 in accordance with FIG. 3 are coupled together to form a chain 400 of amplifiers which achieves gain flattening. The exemplary NSA chain 400 of FIG. 4 is formed by coupling an input signal to be amplified to a first fiber section 410(1) and then to the input fiber 312(1) of a first NSA 300(1). The output port 1 (fiber 320(1)) of the first NSA 300(1) is connected to the input fiber 312(2) of a second NSA 300(2) via a second fiber section 410(2). The output fiber 320(2) of the second NSA 300(2) is connected to the input fiber 312(3) of a third NSA 300(3) via a third fiber section 410(3), and so on, until a desired number of
NSAs 300 are connected together. It should be appreciated that the fiber sections 410(i) represent lengths of fiber such as, for example, tens of kilometers (e.g., 70 kilometers) of fiber between amplifiers in a long distance communication system. The fiber sections can also represent losses, such as splitter losses, or the like, or a combination of component losses and fiber losses. - In FIG. 4, the amplifiers within each
NSA 300 can be pumped by a common pump source, if the NSAs 300(i) are located close to each other, or the amplifiers can be pumped with individual pump sources. - In the following discussion, Leff is the effective length of the
Sagnac loop 354 in each NSA 300(i) of FIG. 4. For the purpose of the following discussion, it is assumed that a total loss from the output of one NSA 300(n) to the input of the next NSA 300(n+1) (including fiber attenuation and circulator insertion loss) is approximately the same for eachNSA 300 and has a value of X. For example, for the following discussion X has a value of 0.01 (−20 dB). As shown in FIG. 4, Pi (n) is defined as the power of channel i at the output of the loss-gain section n of the NSA chain 400. The following values are also used in the following discussion: - n2,eff=5.47×10−20 m2/W; Aeff=12 μm2; α=−0.4 dB/km; λ=1.55 μm; φbias=0; and Leff=10 km.
- Note that for the following discussion, Leff and the gains are assumed to be the same for each NSA(i) in FIG. 4. It should be understood that the values do not have to be the same in order for the present invention to operate.
-
- In
Equation 7, the transfer function of a single NSA is based uponEquation 6, and further includes the effect of the loss X. In order to evaluate the NSA chain performance, the evolution of the NSA output powers Pi (n) as n increases is plotted in FIGS. 5A-5D, as discussed below. - The successive values of Pi (n) can be found by continuously applying the transfer function (Equation 7) to the initial power, Pi (0), i.e.,:
- P i (1) =TF(P i (0)
- P i (2) =TF P i (1) =TF(TF(P i (0)))
-
- P i (n) TF (n) P i (0). (8)
- In
Equation 8, TF(n) denotes the application of the function TF n times. A conceptual understanding of how Pi (n) evolves can be gained by performing this iterative process graphically. (See, for example, M. J. Feigenbaum, “Quantitative universality for a class of nonlinear transformations,” Journal of Statistical Physics, Vol. 19, No. 1, pp. 25-52, 1978.) To this end, two curves are plotted in FIG. 5A. The first curve (curved line) represents the function Pi (n+1)=TF(Pi (n)), namelyEquation 7. The second curve (straight line) represents the function Pi (n+1)=Pi (n), namely a line of slope unity going through the origin. - The foregoing is illustrated in FIG. 5A for a gain, Gi of 26 dB. An initial power, Pi (0), is chosen (in this
case 3 milliwatts). A vertical line is then drawn from x=P1 (0) to y=Pi (1)=TF(Pi (0)) on the transfer function (point A). To find Pi (2), a horizontal line is then drawn from point A on the transfer function to the Pi (n+1)=Pi (n) line (point B). Then, a vertical line is drawn from point B to the transfer function (point C). This value is Pi (3). This process can be repeated to find any Pi (n), as illustrated for points D, E, F, G and H in FIG. 5A. For the transfer function plotted in FIG. 5A, Pi (n) converges to an asymptotic value of approximately 5 milliwatts (the intersection of the transfer function and the y=x line at point H). The asymptotic value of Pi (n) is thus independent of the initial value for a large range of Pi (0). This is shown in FIG. 5B, which plots the value of Pi (n) as a function of n for five different initial values of Pi (0). Note that all values of Pi (n) converge to the same asymptotic value of approximately 5 milliwatts. - The convergence of Pi (n) to a single steady-state value is not the only possible behavior. As the gain is increased, the nonlinearity of the transfer function increases, and the magnitude of the slope of the transfer function can become greater than 1 at the point at which the transfer function crosses the Pi (n+1)=Pi (n) line. When the magnitude of the slope of the transfer function is greater than 1 where the two functions cross, i.e., when the slope of the transfer function is either greater than +1 or smaller than −1, the asymptotic behavior is no longer single-valued but is instead oscillatory. This is illustrated in FIG. 5C, wherein the transfer function is plotted with the same set of parameters as in FIG. 5A except that the gain, Gi, has been increased from 26 dB to 28 dB. Rather than converging, the power, Pi (n), in FIG. 5C oscillates between an upper value (about 5 milliwatts) and a lower value (about 3.5 milliwatts). The oscillation can be seen more clearly in FIG. 5D, which graphically shows the evolution of Pi (n) as n increases. Note that for the example of FIGS. 5C and 5D, Pi (n) exhibits a periodic behavior, which in FIG. 5D has a period of 2.
- As the nonlinearity of the transfer function TF further increases with increased gain Gi, the period shown in FIG. 5D doubles to 4, then to 8, then to 16, etc. With further increases in gain, the asymptotic behavior becomes chaotic. The foregoing is illustrated in FIG. 6, which plots the asymptotic value (or set of values) of Pi (n) as a function of gain. When the gain is lower than the loss in the
segment fiber 410 and theloop fiber 354, the power Pi (n) converges to zero as expected. Once the gain becomes greater than the loss, Pi (n) converges to a non-zero value. When the gain reaches a value of about 27.6 dB, the asymptotic behavior of Pi (n) becomes periodic. Pi (n) first oscillates between two values. The difference between the two values becomes greater as the gain increases. As illustrated in FIG. 6, at approximately 28.46 dB of gain, Pi (n) begins oscillating between four values. At approximately 28.63 dB of gain, Pi (n) begins oscillating between eight values. For a very small range of gains, Pi (n) oscillates between 16 values, then 32 values (not shown), and then 64 values (not shown). For slightly higher gains, the asymptotic behavior of Pi (n) becomes chaotic, as approximately represented by the stippling and white bands in FIG. 6, such that no periodicity of the behavior can be determined. The irregular white bands are caused by narrow ranges of gain wherein the asymptotic behavior resumes periodic behavior. The present invention operates with a range of gains less than the gain where chaotic behavior begins, and it is not necessary to further discuss the chaotic behaviors shown in FIG. 6. Furthermore, the invention preferably operates at gains less than the gains wherein periodic behavior begins. - It can be seen that an NSA chain tends to force the power of a channel into a steady-state value due to the self-regulating transmission resulting from the Kerr-induced phase shift. For example, when two channels having wavelengths λ1 and λ2 and having corresponding gains of 25 and 27 dB, respectively, in an EDFA are applied to the input of a linear chain of EDFAs, such as illustrated in FIG. 1A, the channel at λ2 eventually overwhelms the channel at λ1, and Pi (n) (the power in the channel at λ1) converges to 0. For the NSA chain 400, FIG. 6A shows that P1 (n) converges to approximately 5.0 milliwatts and P2 (n) converges to approximately 5.2 milliwatts. These values are sufficiently close to ensure that both channels arrive at the receiver with adequate power. The self-regulation of the NSA chain 400 prevents single channels from dominating, and thereby enables a larger number of channels to arrive at the receiver with adequate power to be detected. It has been observed from simulations that the convergence of the power to its steady state slows down as the gain approaches the region where periodic and chaotic behavior occurs. As shown from FIGS. 5B and 5D, the convergence at lower gains occurs after only a few iterations.
- The foregoing description characterizes how the NSA chain400 behaves. The following discussion describes the simulation of the NSA chain 400 to find optimal configurations. In order to evaluate various configurations, the relatively complicated behavior of the NSA chain 400 is first reduced to a single figure of merit parameter. In particular, for the purpose of the following discussion, the figure of merit is defined as an allowable gain variability, ΔG. The allowable gain variability parameter describes a range over which the gain can vary while still forcing Pi (n) to converge to an acceptable asymptotic behavior.
-
- The restriction in Equation 9 places an upper limit on the amount of power exiting the
optical amplifier 360 in theSagnac loop fiber 356 in the counterclockwise direction. The factor of two inEquation 10 results from the 50-percent coupler 330 which splits the light into clockwise and counterclockwise beams. The upper power limit results from various undesirable nonlinear processes in the loop fiber, such as Brillouin scattering or four-wave mixing (FWM). The upper limit may also be set by limitations in the EDFA power output. -
- In the approximation of
Equation 10, the parameter gB is the Brillouin gain coefficient and is a function of the linewidth of each channel. - For the following analysis, it is assumed that each channel employs phase-shift keying (PSK), in which 0 and 1 are encoded on a continuous wave (cw) laser by applying a 0 or π phase shift, respectively, as illustrated in FIG. 7A. In FIG. 7A, each bit has a duration of T. This phase modulation spreads the linewidth of the channel. (See, for example, Y. Aoki, et al., “Input power limits of single-mode optical fibers due to stimulated Brillouin scattering in optical communication systems,”Journal of Lightwave Technology, Vol. 6, No. 5, pp. 710-719, 1988.) For a pseudo-random bit pattern, the resulting frequency spectrum of a PSK channel with bit rate B=1/T is shown in FIG. 7B. The full width at half maximum (FWHM) linewidth of the PSK signal is approximately equal to B.
-
- In the approximation of Equation 11, ΔQB is the intrinsic Brillouin gain linewidth, which is between 10 and 100 MHz, depending on the fiber doping. The value of the Brillouin gain parameter, gB, for silica fiber is 4.6×10−11 meters/watt. With these parameters and with Aeff=9.0 μm2 and Leff=10 kilometers, the critical power, Pcr, for Brillouin scattering is plotted in FIG. 8 as a function of the channel bit rate. As shown in FIG. 8, Pcr is well under a milliwatt for an unmodulated channel. However, Pcr is greater than 10 milliwatts for a 2.5 gigabit per second channel. Although the value of Pcr is calculated for the case of PSK, it should be understood that the present invention is compatible with other types of signal modulations, as discussed above. The example of PSK is used only to provide a specific value for Pcr.
- As discussed in more detail below, the methodology for evaluating various configurations of NSA chains compares the resulting allowable gain variation ΔG. To find ΔG, the asymptotic values of GiPi (n)X/2 is plotted as a function of gain in FIG. 9. From the plots, the ranges of gain for which Equation 9 is satisfied is determined. In FIG. 9, the nonlinear coefficient n2 is varied while the other parameters are held constant. Note that higher values of n2 force the asymptotic value of GiPi (n)X/2 lower, enabling a larger ΔG. If Pmax is set to 10 milliwatts, as shown by the horizontal line in FIG. 9, the value n2=5.47×10−20 m2/W (a value currently achieved in commercially available fibers) produces a ΔG of 1.65 dB. A moderate decrease in n2 by a factor of 2 decreases ΔG to 0.4 dB. Likewise, an increase in n2 by a factor of 2 increases ΔG to 3.6 dB (if the present invention is not operated in the oscillatory regime above approximately 27.6 dB).
- Plots of the asymptotic values of GiPi (n)X/2 versus gain can be generated as a function of other parameters such as L or φbias to further analyze and optimize the NSA chain.
- FIG. 10 illustrates the gain spectra for 16 wavelength division multiplexed channels for a total input power of 1.4 milliwatts (curve a), 0.56 milliwatts (curve b) and 0.22 milliwatts (curve b). The total input power is distributed equally amongst the 16 wavelengths, and is input into the EDFA from both sides as would be the case if the EDFA were inserted into a Sagnac loop. The parameters of the NSA chain are the same as those discussed above.
- To simulate the behavior of a chain of NSAs, the gain experienced by each of the N input signals that travel through each individual amplifier in each NSA of the chain is first computed. Because the gain in an optical amplifier depends in a generally complex manner on the input signal powers through gain saturation, and because the input signal power spectrum is different for each amplifier, the gain experienced by each one of the N input signals at each amplifier in the chain should be computed to be completely accurate. Although all the gains can be calculated using existing EDFA simulation programs, the calculations are quite time consuming. Instead of calculating all the gains in this manner, the approximation described in D. M. Baney, et al., “WDM EDFA gain characteristics with a reduced set if saturating channels”,IEEE Photonics Technology Letters, Vol. 8, No. 12, 1996, pp. 1615-1617, is used. The Baney, et al., reference teaches that the gain spectrum of an amplifier can be accurately calculated by replacing N input signals to the amplifier by a single input signal (called a saturating tone) of suitably selected wavelength and power. Specifically, according to the Baney, et al., reference, the gain spectrum of an amplifier operated with N input signals of individual power Pi is very nearly equal to the gain spectrum of the same EDFA (i.e., the EDFA pumped under the same conditions) saturated by a saturating tone of wavelength λs and power Ps given by:
- In
Equation 12, Gi is the (saturated) gain at wavelength λi, Gs is the (saturated) gain at wavelength λs, and Gi and Gs are linear gains (as opposed to gains expressed in dB). When inhomogeneous broadening in the EDFA is negligible, as is assumed here, the saturating tone wavelength λs can be selected over a broad range of wavelengths without affecting the accuracy of the method. With this method, it is clear that the gain spectrum of the EDFA only needs to be computed for a range of saturating tone powers (instead of having to compute it for any arbitrary combination of N signal powers). -
- In a second step, the unsaturated gain values are inserted in
Equation 12 to calculate an approximate value of Ps. This approximate value of Ps is then used in the EDFA simulation program to calculate a second set of (saturated) gains Gi and Gs. The process is repeated until the gains converge to a stable value. In practice, convergence of the gains is obtained at the second step. - In the following discussion, the common situation of an EDFA with negligible inhomogeneous broadening is simulated. This simulation enables the selection (again, without loss of accuracy) of an arbitrary saturating tone wavelength λs of 1,551 nanometers. To simulate the EDFA in the Sagnac loop, in which each signal of wavelength λi travels in two directions, the saturating signal is input into both input ports of the EDFA with proper respective powers.
- The EDFA simulated in FIG. 10 is a 7-meter section of erbium-doped fiber bidirectionally pumped by 110-milliwatt, 1,480-nanometer light (220 milliwatts total pump). The EDFA exhibits gain spectra as shown in FIG. 10. In the region from 1,544 nanometers to 1,559 nanometers where the 16 WDM signals are located (1-nanometer channel spacing), the typical gain variation was on the order of 1-2 dB, depending on the input signal power.
- FIGS. 11A and 11B show the evolution of the power in all 16 channels as a function of the amplifier number for an NSA chain (FIG. 11A) and a standard amplifier chain (FIG. 11B). In the standard amplifier chain, the loss between sections (X) was increased from 20 dB to 24 dB to account for the 4-dB loss of the nonlinear fiber, which is set to zero in the simulator. The EDFA is identical to that used in the NSA chain. As shown, the NSA chain forces all powers into an asymptotic value such that all channels maintain a power between 3.7 milliwatts and 5.2 milliwatts. The convergence occurs after the first few NSAs. In the standard amplifier chain, the channels with the highest gains quickly dominate the other channels. The stabilizing effect of the NSA is therefore clearly seen by comparing FIG. 11A and FIG. 11B.
- The channel power spectrum is shown in FIGS. 12A and 12B for the NSA chain and the standard chain after
segment 20, respectively. In the NSA chain, all channel powers are roughly equal after segment 20 (see FIG. 12A). In contrast, aftersegment 20 of the standard chain, there is a large disparity between the powers around 1,557 nanometers and the powers around 1,544 nanometers (see FIG. 12B). At the output end of the NSA chain (see FIG. 12A), all of the 16 output signals are useful, (i.e., they carry sufficient power to be detected with a high signal-to-noise ratio). In contrast, at the output end of the standard amplifier chain (see FIG. 12B), only about 6 of the 16 output signals are useful. All other signals carry too little power and cannot be detected with a sufficiently high signal-to-noise ratio to be useful. This comparison illustrates that the present invention produces output signals with approximately equal output powers over a large bandwidth after a large number of gain-loss sections. - The foregoing discussion illustrates the results of specific simulations which indicate the gain flattening achieved by the NSA chain400 and also presents a method for evaluating the performance of the NSA chain 400, namely by using the parameter ΔG to find the maximum allowable gain variation.
- It will be appreciated by one skilled in the art that the present invention achieves true (i.e., without external feedback) automatic gain control (AGC). It achieves automatic gain control in the sense that it automatically maintains the powers of a large number of signals at the output of a chain of gain-loss sections within a small range of power values, by correcting internally for unpredictable variations in the input signal powers, the pump powers, and the number of signals. In contrast, the overwhelming majority of optical gain flattening schemes developed to date do not exhibit true automatic gain control. Instead, the other schemes rely on additional external feedback schemes (which add complexity and cost to the system) to correct for variations in these quantities.
- As discussed above, the Kerr-induced phase modulation from a second beam (cross-phase modulation) was twice that due to self-phase modulation, a feature which enables the NSA chain400 to flatten the gain. The conditions for self-phase modulation and cross-phase modulation are addressed in the context of fiber-optic gyroscopes in B. Crosignani, “Kerr effect and chromatic dispersion in fiber-optic gyroscopes,” Journal of Lightwave Technology, Vol. LT-3, No. 4, pp. 914-918, 1985, wherein it is shown that self-phase modulation and cross-phase modulation are connected with the dispersion in the fiber. The information from the Crosignani article is used below to define the conditions for self-phase modulation and cross-phase modulation.
-
- In Equation 13, Leff is the effective fiber length, as previously defined, and A is the fiber group-velocity dispersion (i.e., 1/A=d2β/dω2, where β is the propagation constant and ω is the frequency). The Kerr-induced phase shift of one frequency component at ν1 on another frequency component at ν2 can be considered self-phase modulation if |ν1-ν2|<<Δνc Likewise, cross-phase modulation occurs when ν1-ν2|>>Δνc Equation 13 enables the computation of the value of Δν′ for a particular Sagnac loop.
- The foregoing results impose limits on the channel bandwidth (i.e., the bit rate) and the channel separation of the nonlinear Sagnac amplifier. All the light within each channel is required to self-phase modulate all other co-propagating light in the same channel. Thus, the bandwidth of each channel, B, should satisfy the following relationship:
- B Δν′ (14)
- In Equation 14, B is the
bit rate 1/T for random PSK, where T is the bit duration. Thus, there is a relationship between the maximum bit rate of each channel and the parameters of the nonlinear Sagnac amplifier. - All the light within one channel is also required to cross-phase modulate all light in other channels. This implies that the channel spacing, Δνchannel, should satisfy the following relationship:
- Δνchannel>>Δν′ (15)
- and thus, the channel spacing has a minimum value which is a function of the loop parameters.
- Because the length, Leff, of the loop is chosen to optimize the gain flattening performance, as discussed above, it is preferable to set Δνc to an optimal value by modifying the group-velocity dispersion, A, of the loop fiber (see Equation 13). A typical value for Δν′ can be calculated by taking Leff to be 5 kilometers and A−1 to be 3.28×10−28 sec2/meter (appropriate for a dispersion shifted fiber near the zero-dispersion wavelength). The two values yield a value of 124 GHz for Δνc. This value can be modified easily through the parameter A.
- From the foregoing description, it can be seen that the nonlinear Sagnac amplifier in accordance with the present invention produces an output signal spectrum that is flat, notwithstanding significant differences in the power of input optical signals at different wavelengths. Furthermore, unlike known gain-flattening schemes, the flatness of the output signal spectrum generally improves with increasing number of gain-loss elements. The flatness of the output signal spectrum is quite good over a large bandwidth. For example, in a chain of nonlinear Sagnac amplifiers with EDFAs that are flat to within approximately 1-2 dB, the flatness after a large number of gain-loss sections is better than 1.5 dB over a very large bandwidth. In contrast, passive filters typically flatten the gain of an individual EDFA to within a few tenths of a dB, which is not sufficient to guarantee a flat gain after traveling through a large number of such EDFAs. For example, after traveling through 24 gain-loss sections comprising EDFAs flattened to 0.3 dB, the output signal spectrum flatness is typically 1.5 dB or worse. The NSA chain achieved this same level of spectrum flatness with amplifiers which are flattened to only 1-2 dB. In view of the foregoing, it can be seen that the nonlinear Sagnac interferometer is more beneficial as the size of the chain (i.e., the number of amplifiers in the chain) increases. This is particularly advantageous for chains comprising a very large number of amplifiers, such as, for example, a transoceanic communication bus.
- As discussed above, the chain of nonlinear Sagnac amplifiers converges to a flat gain after a number of gain-loss sections. The number of amplifiers needed to obtain flat gain (e.g., an asymptotic gain value or a percentage (e.g., 90 percent) of the asymptotic gain value) is a system characteristic which can be adjusted by adjusting the parameters of the nonlinear Sagnac amplifiers in the chain. The ability to control this rate of convergence is significant because in some systems, a fast convergence is needed (e.g., convergence to flat gain after four amplifiers) and in other systems, a slower convergence is acceptable (e.g., convergence to flat gain after twenty amplifiers).
- Unlike EDFAs flattened with passive amplifiers, the system of the present invention is particular advantageous because the output signal spectrum remains very flat even when one or more of the input signals are dropped. When the power in one input signal is dropped, all of the output signals corresponding to the remaining (non-dropped) input signals will carry a higher power. However, the present invention keeps the flatness of the output signal spectrum from changing at all or from changing more than a very small amount. Similarly, the gain flatness does not change or changes very little if the power in one or more of the input signals changes (i.e., increases or decreases). To illustrate this, FIG. 13 shows the power evolution of a 16 channel WDM system in which 8 channels are dropped after the fourteenth and before the fifteenth NSA. The remaining eight channels now see a different gain and they therefore converge to different asymptotic values. The power deviation, however, is quite small, and all 8 remaining channels arrive at the end of the NSA with powers between 5 milliwatts and 5.5 milliwatts.
- The output signal spectrum flatness is also robust to changes in the power that pumps one or more of the EDFAs in the chain of nonlinear Sagnac amplifiers. Changes in any or all of the pump powers cause changes in the optical gain and thus cause changes in the power carried by all the output signals at the output of the chain. However, because of the converging effect of the chain of nonlinear Sagnac amplifiers, the flatness of the output signal spectrum will be nominally unchanged (assuming the pump power changes are reasonably modest in magnitude). This robustness in the flatness of the output signal spectrum is not provided by filter-flattened EDFAs.
- The flattening properties of a nonlinear Sagnac amplifier are more tolerant to manufacturing variations in the parameters of components than most other gain-flattened amplifiers, particularly when compared to gain-flattened EDFAs relying on passive filters in which each EDFA must be flattened with a particular filter matched to the particular gain curve of the EDFA. When passive filters are used, a given filter will flatten the gain of a particular EDFA operated under certain conditions, but the filter will perform poorly, if at all, if it is used with the same EDFA pumped with a different pump power or operated with a different number of signals or operated with different signal powers or a combination of differences. The filter may also perform poorly when the same filter is used with two nominally identical EDFAs made with erbium-doped fibers that are slightly different in design, in erbium concentrations, in co-dopant concentrations, in length, or in a combination of differences. Similarly, changes in the filter transmission spectrum (e.g., through changes in the external temperature) will alter the flatness of the chain gain. In contrast to conventional amplifiers having gain flattened by passive filters, a nonlinear Sagnac amplifier constructed in accordance with the present invention with a particular loop fiber and loop length can flatten the gain of EDFAs having a large range of characteristics, and the flatness of the output signal spectrum of the chain is robust to fluctuations in the parameters of the nonlinear Sagnac amplifiers.
- The figures presented herein illustrate the performance of NSAs with loop lengths of 10 km. It should be understood that the choice of loop length impacts the system performance. For example, there is a tradeoff between length and mode confinement. In particular, the stronger the mode confinement, the shorter the length that is required. There is also a tradeoff between the loop length and the fiber nonlinearity. In particular, the stronger the fiber nonlinearity, the shorter the length that is required. Similarly, there is also a tradeoff between the loop length and the signal power. In particular, the stronger the signal power, the shorter the length that is required. NSAs with loop lengths either substantially shorter or substantially longer than 10 km can be designed to operate with nominally the same characteristics as described in this invention.
- Although described above in connection with erbium-doped fiber amplifiers, the general principles of the nonlinear Sagnac amplifier described herein are not restricted to trivalent erbium (Er3+) as the active dopant. The nonlinear Sagnac amplifier can also be used to flatten the gain of any optical amplifier based upon a fiber, a waveguide or a bulk optical component. The optical amplifier may use any active ion or active laser medium, and may operate over any portion of the spectral range. In particular, the nonlinear Sagnac amplifier may be used with Raman amplifiers, which may be used, for example, for amplification applications not suitably covered by other optical amplifiers. It should be noted that a Raman amplifier or other amplifier used in connection with the present invention should be bidirectional so that it can amplify the counterpropagating signals in the Sagnac loop. With respect to a Raman amplifier embodiment, the Raman amplifier can be operated in a resonant mode or it may be pumped in two different directions.
- Note further that the foregoing embodiments have been described in connection with a single amplifier in each loop. One skilled in the art will appreciate that a single EDFA may have a limited amplification range. For example, a first amplifier AMP1 may amplify over one band of wavelengths and may absorb over a second band of wavelengths. A second amplifier AMP2 may absorb over the first band and may amplify over the second band. To provide amplification over both bands, the two amplifiers AMP1 and AMP2 may be positioned in the loop in parallel as a composite amplifier by using two WDM couplers. In other words, the nonlinear Sagnac amplifier of FIG. 3 is utilized with an
amplifier 360 which is made of the parallel arrangement of the two amplifiers AMP1 and AMP2 shown in FIG. 14A, wherein AMP1 and AMP2 are two individual amplifiers configured, for example, like the amplifier of FIG. 3A. - Similarly, if a first amplifier AMP1 amplifies a first band and passes a second band and a second amplifier AMP2 passes the first band and amplifies the second band, the two amplifiers may be placed in series in the loop to provide amplification over both bands. This composite amplifier is illustrated in FIG. 14B. The individual amplifiers AMP1 and AMP2 can both be configured, for example, like the amplifier of FIG. 3A. Again, the nonlinear Sagnac amplifier then consists of the nonlinear Sagnac amplifier of FIG. 3 comprising an
amplifier 360 which is made of the composite amplifier of FIG. 14B. - It should be understood that combinations of amplifiers in series and in parallel can be used to cover multiple bands of wavelengths.
- In the event that amplification is required over a very wide spectral range, it may be necessary to provide separate loops for different portions of the spectral range so that the parameters of the loops can be optimized for the different portions of the spectral range. This configuration is shown schematically in FIG. 15, in which NSA1 and NSA2 represent two nonlinear Sagnac amplifiers that provide gain over different (and perhaps adjacent) portions of the spectrum (for example 1300 nanometers to 1350 nanometers for NSA1 and 1500 nanometers to 1650 nanometers for NSA2). The first WDM coupler acts to separate the input signals and direct them to the appropriate nonlinear Sagnac amplifier, namely to direct signals with wavelengths between 1300 nanometers and 1350 nanometers (in this example) to NSA1 and to direct signals with wavelengths between 1500 nanometers and 1650 nanometers to NSA2. The second WDM coupler recombines the signals exiting NSA1 and NSA2 onto the communication fiber (represented by the loss box in FIG. 15).
- An alternative embodiment to the embodiment of FIG. 4 is illustrated in FIG. 16. In the embodiment of FIG. 16, a gain-
loss string 1600 comprises both nonlinear Sagnac amplifiers (NSAs) 1610 and standard optical amplifiers (AMPs) 1620 separated byloss sections 1630. In particular, in the embodiment illustrated in FIG. 16, theNSAs 1610 and theAMPs 1620 are positioned alternatively (i.e., every other amplifier). The percentage of nonlinear Sagnac amplifiers in the chain illustrated in FIG. 16 is thus 50% (i.e., onenonlinear Sagnac amplifier 1610 for every standard amplifier 1620). The percentage ofnonlinear Sagnac amplifiers 1610 in the chain can be smaller (for example, every third amplifier in the chain can be a nonlinear Sagnac amplifier 1610) or larger (for example, every third amplifier in the chain can be a standard amplifier 1620) depending on the degree to which thestandard amplifiers 1620 are individually flattened. For example, if thestandard amplifiers 1620 are not flattened well, only onestandard amplifier 1620 may be used for every fewnonlinear Sagnac amplifiers 1610. If thestandard amplifiers 1620 are moderately well flattened, every other amplifier may be astandard amplifier 1620. If thestandard amplifiers 1620 are very flat, only onenonlinear Sagnac amplifier 1610 needs to be placed in thestring 1600 every so often (for example, every third amplifier is annonlinear Sagnac amplifier 1610 and the rest of the amplifiers are standard amplifiers 1620). Thishybrid chain 1600 is advantageous because (1) it maintains a high gain flatness even after a very large number of gain-loss sections, and (2) it makes the amplifier-loss chain robust to changes (intentional or unintentional) in the input conditions (i.e., the number of signals, the signal power, or the pump power). Thehybrid chain 1600 is beneficial because by substituting relatively inexpensivestandard amplifiers 1620 for the more expensive nonlinear Sagnac amplifiers (which have the fiber loops and optical circulators), a significant cost reduction can be obtained. - Although described herein in connection with the flattening of the gain of amplifiers having gains which vary with wavelength, the present invention can also be used with individual amplifiers having a gain which exhibits either no or little wavelength dependence. The invention works even better (i.e., it provides a flatter gain after fewer amplifiers) if the gain spectra of the individual amplifiers are flatter.
- The present invention can also be used with individual amplifiers with a gain spectrum that is nominally flat (e.g., to within 0.3 dB over an 80-nanometer bandwidth), which can be accomplished by a number of techniques well known in the art. In this case the invention performs the function of maintaining or even improving the gain flatness after a large number (e.g., tens) of amplifiers. In contrast, if the same individual amplifiers are strung together in the traditional linear amplifier chain, as described above in connection with FIG. 1A, the amplifier string yields an increasing deterioration of the gain flatness as the number of gain-loss section is increased.
- Furthermore, when using the present invention with individual amplifiers, which each has a gain spectrum that is nominally flat (e.g., to within 0.3 dB over an 80-nanometer bandwidth), the present invention provides a nominally flat gain spectrum that is robust against changes in the signal power, against changes in the number of signals, and, to some degree, against changes in the pump power.
- Although described above in connection with particular embodiments of the present invention, it should be understood the descriptions of the embodiments are illustrative of the invention and are not intended to be limiting. Various modifications and applications may occur to those skilled in the art without departing from the true spirit and scope of the invention as defined in the appended claims.
Claims (3)
1. A method of amplifying a plurality of input optical signals, said input optical signals having respective optical wavelengths within a range of optical wavelengths and having respective optical powers within a range of input optical powers, said method producing a corresponding plurality of output optical signals having respective output powers within a selected range of output optical powers, said method comprising the steps of
passing said input optical signals through a first nonlinear Sagnac amplifier to produce a plurality of intermediate optical signals, said plurality of intermediate optical signals being responsive to said plurality of input optical powers, to amplifier gain and to Kerr-induced phase shift to have a respective plurality of intermediate optical powers within an intermediate range of optical powers, said intermediate range of optical powers being a smaller range than said range of input optical powers; and
passing said intermediate optical signals through at least a second nonlinear Sagnac amplifier to produce said plurality of output optical signals, said plurality of output optical signals being responsive to said plurality of intermediate optical powers, to amplifier gain and to Kerr-induced phase shift to have a plurality of output optical powers within said selected range of output optical powers, said selected range of output optical powers being smaller than said range of intermediate optical powers.
2. An optical system comprising:
an input that receives optical input signals having wavelengths within a range of wavelengths;
an output that outputs optical output signals responsive to said optical input signals;
at least a first amplifier interposed between said input and said output, said first amplifier having a first gain spectrum, said first gain spectrum varying over said range of wavelengths to cause a first optical signal produced by said first amplifier at a first wavelength within said range of wavelengths to have a first optical power and to cause a second optical signal produced by said first amplifier at a second wavelength within said range of wavelengths to have a second optical power, said first optical power and second optical power differing by a power difference; and
at least a second amplifier interposed between said first amplifier and said output, said second amplifier comprising a nonlinear Sagnac amplifier, said second amplifier operating to at least partially compensate for differences in gain of said first amplifier over said range of wavelengths to reduce said power difference between said first optical signal and said second optical signal.
3. An optical system comprising:
an input that receives optical input signals having wavelengths within a range of wavelengths;
an output that outputs optical output signals responsive to said optical input signals;
a first plurality of amplifiers of a first type interposed between said input and said output, said amplifiers of said first type having a first gain spectrum, said first gain spectrum varying over said range of wavelengths to cause a first optical signal at a first wavelength within said range of wavelengths to be amplified with a first gain by said amplifiers of said first type and to cause a second optical signal at a second wavelength within said range of wavelengths to be amplified with a second gain by said amplifiers of said first type, said second gain differing from said first gain; and
a second plurality of nonlinear Sagnac amplifiers interposed at selected locations in said optical system, said nonlinear Sagnac amplifiers operating to at least partially compensate for differences in said first gain and said second gain of said amplifiers of said first type to reduce power differences between said first optical signal and said second optical signal.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US10/060,868 US6466364B1 (en) | 1999-08-02 | 2002-01-29 | Gain flattening with nonlinear sagnac amplifiers |
Applications Claiming Priority (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US14674499P | 1999-08-02 | 1999-08-02 | |
US09/537,639 US6377391B1 (en) | 1999-08-02 | 2000-03-29 | Gain flattening with nonlinear Sagnac amplifiers |
US10/060,868 US6466364B1 (en) | 1999-08-02 | 2002-01-29 | Gain flattening with nonlinear sagnac amplifiers |
Related Parent Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US09/537,639 Continuation US6377391B1 (en) | 1999-08-02 | 2000-03-29 | Gain flattening with nonlinear Sagnac amplifiers |
Publications (2)
Publication Number | Publication Date |
---|---|
US20020145795A1 true US20020145795A1 (en) | 2002-10-10 |
US6466364B1 US6466364B1 (en) | 2002-10-15 |
Family
ID=26844261
Family Applications (2)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US09/537,639 Expired - Lifetime US6377391B1 (en) | 1999-08-02 | 2000-03-29 | Gain flattening with nonlinear Sagnac amplifiers |
US10/060,868 Expired - Lifetime US6466364B1 (en) | 1999-08-02 | 2002-01-29 | Gain flattening with nonlinear sagnac amplifiers |
Family Applications Before (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US09/537,639 Expired - Lifetime US6377391B1 (en) | 1999-08-02 | 2000-03-29 | Gain flattening with nonlinear Sagnac amplifiers |
Country Status (9)
Country | Link |
---|---|
US (2) | US6377391B1 (en) |
EP (1) | EP1201011B1 (en) |
JP (1) | JP4514382B2 (en) |
KR (1) | KR100680682B1 (en) |
CN (1) | CN1158567C (en) |
AU (1) | AU1324301A (en) |
CA (1) | CA2380060A1 (en) |
DE (1) | DE60034341T2 (en) |
WO (1) | WO2001009991A2 (en) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2005078391A1 (en) * | 2004-01-21 | 2005-08-25 | Northrop Grumman Corporation | Schemes for computing performance parameters of fiber optic gyroscopes |
US20090015843A1 (en) * | 2007-07-11 | 2009-01-15 | Demers Joseph R | Fiber optic gyroscope |
US20090103931A1 (en) * | 2007-10-19 | 2009-04-23 | Grigoryan Vladimir S | Systems and methods for the coherent non-differential detection of optical communication signals |
US20090169208A1 (en) * | 2007-10-19 | 2009-07-02 | Grigoryan Vladimir S | Systems and methods for the polarization insensitive coherent detection and the polarization division multiplexed transmission of optical communication signals |
US20130050708A1 (en) * | 2010-05-07 | 2013-02-28 | Ralph A. Bergh | Method and apparatus for multiplexing multiple sagnac interferometers with single input for source light |
US8773665B1 (en) | 2011-04-01 | 2014-07-08 | Emcore Corporation | Compact fiber optic gyroscope |
US8798405B2 (en) | 2011-08-04 | 2014-08-05 | Emcore Corporation | Method of making a fiber optic gyroscope |
US8823946B1 (en) | 2011-06-13 | 2014-09-02 | Emcore Corporation | Multi-axis fiber optic gyroscope with single light source |
Families Citing this family (19)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6377391B1 (en) * | 1999-08-02 | 2002-04-23 | The Board Of Trustees Of The Leland Stanford Junior University | Gain flattening with nonlinear Sagnac amplifiers |
WO2001078264A2 (en) * | 2000-02-14 | 2001-10-18 | Xtera Communications, Inc. | Nonlinear optical loop mirror |
US6490380B2 (en) * | 2000-03-24 | 2002-12-03 | Oprel Technologeis Inc. | Optical amplifier with loop mirror filter noise reducer, and loop mirror filter per se |
US6496615B2 (en) * | 2000-05-22 | 2002-12-17 | Sumitomo Electric Industries, Ltd. | WDM transmission system |
CA2373912A1 (en) | 2001-03-01 | 2002-09-01 | Takeshi Hirasawa | Optical transmission system |
US6563629B2 (en) * | 2001-05-18 | 2003-05-13 | Redc Optical Networks Ltd. | Method and apparatus for full C-band amplifier with high dynamic gain range |
US6804057B1 (en) | 2002-02-06 | 2004-10-12 | Novera Optics, Inc. | Various methods and apparatuses for a tunable chromatic dispersion compensator |
US6788845B1 (en) | 2002-05-31 | 2004-09-07 | Novera Optics, Inc. | Methods and apparatuses to provide a tunable chromatic dispersion compensator |
CN101105422B (en) * | 2007-07-30 | 2010-04-21 | 浙江大学 | Depolarizer performance accurate estimation method |
US8068232B2 (en) * | 2008-04-01 | 2011-11-29 | The Board Of Trustees Of The Leland Stanford Junior University | Unidirectional crow gyroscope |
CN101340050B (en) * | 2008-08-14 | 2010-06-02 | 南京大学 | Rational harmonic mode locking optical fiber laser with pulse amplitude uniformized |
US9019482B2 (en) | 2009-06-05 | 2015-04-28 | The Board Of Trustees Of The Leland Stanford Junior University | Optical device with fiber Bragg grating and narrowband optical source |
US9025157B2 (en) | 2010-09-08 | 2015-05-05 | The Board Of Trustees Of The Leland Stanford Junior University | System and method for measuring perturbations using a slow-light fiber Bragg grating sensor |
US8797540B2 (en) | 2010-09-08 | 2014-08-05 | The Board Of Trustees Of The Leland Stanford Junior University | Slow-light fiber Bragg grating sensor |
CN103443701B (en) * | 2011-03-04 | 2017-04-12 | 古河电气工业株式会社 | Optical parameter amplifier, optical amplifying system, wavelength converter, optical amplification method, and optical communication system |
US10563986B2 (en) | 2016-09-22 | 2020-02-18 | LGS Innovations LLC | Cladding-pumped waveguide optical gyroscope |
CN106949954B (en) * | 2017-03-15 | 2019-03-15 | 武汉理工大学 | A kind of fiber-optic vibration signal supervisory instrument and method |
WO2019182502A1 (en) * | 2018-03-19 | 2019-09-26 | Telefonaktiebolaget L M (Publ) | System and method of signaling spectrum flatness configuration |
CN113542939B (en) * | 2021-07-12 | 2022-04-22 | 苏州大学 | Multi-cycle upgrading scheduling method based on ultra-low loss optical fiber |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
GB2265751B (en) * | 1992-03-23 | 1995-12-20 | Univ Southampton | Optical amplifier with automatic self adjusting gain spectrum |
JPH0611743A (en) * | 1992-06-29 | 1994-01-21 | Nippon Telegr & Teleph Corp <Ntt> | Light pulse amplifier device |
US6374006B1 (en) * | 1998-03-20 | 2002-04-16 | Xtera Communications, Inc. | Chirped period gratings for raman amplification in circulator loop cavities |
US6377391B1 (en) * | 1999-08-02 | 2002-04-23 | The Board Of Trustees Of The Leland Stanford Junior University | Gain flattening with nonlinear Sagnac amplifiers |
-
2000
- 2000-03-29 US US09/537,639 patent/US6377391B1/en not_active Expired - Lifetime
- 2000-07-12 CA CA002380060A patent/CA2380060A1/en not_active Abandoned
- 2000-07-12 JP JP2001514512A patent/JP4514382B2/en not_active Expired - Fee Related
- 2000-07-12 EP EP00975152A patent/EP1201011B1/en not_active Expired - Lifetime
- 2000-07-12 WO PCT/US2000/018953 patent/WO2001009991A2/en active IP Right Grant
- 2000-07-12 DE DE60034341T patent/DE60034341T2/en not_active Expired - Lifetime
- 2000-07-12 CN CNB008109729A patent/CN1158567C/en not_active Expired - Fee Related
- 2000-07-12 KR KR1020027001452A patent/KR100680682B1/en active IP Right Grant
- 2000-07-12 AU AU13243/01A patent/AU1324301A/en not_active Abandoned
-
2002
- 2002-01-29 US US10/060,868 patent/US6466364B1/en not_active Expired - Lifetime
Cited By (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2005078391A1 (en) * | 2004-01-21 | 2005-08-25 | Northrop Grumman Corporation | Schemes for computing performance parameters of fiber optic gyroscopes |
US20090015843A1 (en) * | 2007-07-11 | 2009-01-15 | Demers Joseph R | Fiber optic gyroscope |
US7746476B2 (en) * | 2007-07-11 | 2010-06-29 | Emcore Corporation | Fiber optic gyroscope |
US20090103931A1 (en) * | 2007-10-19 | 2009-04-23 | Grigoryan Vladimir S | Systems and methods for the coherent non-differential detection of optical communication signals |
US20090169208A1 (en) * | 2007-10-19 | 2009-07-02 | Grigoryan Vladimir S | Systems and methods for the polarization insensitive coherent detection and the polarization division multiplexed transmission of optical communication signals |
US7853157B2 (en) * | 2007-10-19 | 2010-12-14 | Ciena Corporation | Systems and methods for the polarization insensitive coherent detection and the polarization division multiplexed transmission of optical communication signals |
US7853156B2 (en) * | 2007-10-19 | 2010-12-14 | Ciena Corporation | Systems and methods for the coherent non-differential detection of optical communication signals |
US20130050708A1 (en) * | 2010-05-07 | 2013-02-28 | Ralph A. Bergh | Method and apparatus for multiplexing multiple sagnac interferometers with single input for source light |
US9518827B2 (en) * | 2010-05-07 | 2016-12-13 | Ralph A. Bergh | Method and apparatus for multiplexing multiple Sagnac interferometers with single input for source light |
US8773665B1 (en) | 2011-04-01 | 2014-07-08 | Emcore Corporation | Compact fiber optic gyroscope |
US8823946B1 (en) | 2011-06-13 | 2014-09-02 | Emcore Corporation | Multi-axis fiber optic gyroscope with single light source |
US8798405B2 (en) | 2011-08-04 | 2014-08-05 | Emcore Corporation | Method of making a fiber optic gyroscope |
Also Published As
Publication number | Publication date |
---|---|
KR100680682B1 (en) | 2007-02-09 |
JP2003506733A (en) | 2003-02-18 |
CN1158567C (en) | 2004-07-21 |
US6466364B1 (en) | 2002-10-15 |
AU1324301A (en) | 2001-02-19 |
DE60034341D1 (en) | 2007-05-24 |
JP4514382B2 (en) | 2010-07-28 |
WO2001009991A3 (en) | 2001-08-16 |
EP1201011B1 (en) | 2007-04-11 |
US6377391B1 (en) | 2002-04-23 |
EP1201011A2 (en) | 2002-05-02 |
DE60034341T2 (en) | 2008-01-03 |
KR20020027515A (en) | 2002-04-13 |
CN1365456A (en) | 2002-08-21 |
CA2380060A1 (en) | 2001-02-08 |
WO2001009991A2 (en) | 2001-02-08 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US6377391B1 (en) | Gain flattening with nonlinear Sagnac amplifiers | |
CN1351428B (en) | Distributive optical amplifier, optical communication station, system and cables | |
US7548368B2 (en) | Raman amplifier, optical repeater, and raman amplification method | |
US6876489B2 (en) | All band amplifier | |
US5600481A (en) | Optical fiber amplifier and optical transmission system using the same | |
US20030035202A1 (en) | Active gain equalization | |
US6943936B2 (en) | Co-propagating Raman amplifiers | |
JP4425740B2 (en) | Optical amplifier | |
EP1830442A2 (en) | Method and device for generation of phase conjugate light and wavelength conversion, and system having the device | |
US7190861B2 (en) | Monolithic semiconductor light source with spectral controllability | |
WO1999048176A1 (en) | Chirped period gratings for raman amplification in circulator loop cavities | |
KR100319748B1 (en) | Wideband multichannel fiber lasers with output power equalization | |
US6924926B2 (en) | Laser diode pump sources | |
US20040196882A1 (en) | Optical amplifier | |
US20020122244A1 (en) | Temperature-stabilized optical amplifier and method for temperature-stabilizing an optical amplifier | |
Costa et al. | Modified split-step Fourier method for the numerical simulation of soliton amplification in erbium-doped fibers with forward-propagating noise | |
EP1206016A1 (en) | Temperature-stabilized optical amplifier and method for temperature-stabilizing an optical amplifier | |
Jiang et al. | Multi-wavelength sub-100 Hz linewidth Brillouin fiber laser with cascaded SBS polarization parity-time symmetry | |
Lee et al. | All-optical gain-clamped erbium-doped fiber amplifier for wavelength-division multiplexed networking using stimulated Brillouin scattering | |
Aghababaeian et al. | EFFECTS OF STIMULATED SCATTERING ON THE GAIN OF ERBIUM DOPED FIBER AMPLIFIERS |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
STCF | Information on status: patent grant |
Free format text: PATENTED CASE |
|
FPAY | Fee payment |
Year of fee payment: 4 |
|
FEPP | Fee payment procedure |
Free format text: PAYOR NUMBER ASSIGNED (ORIGINAL EVENT CODE: ASPN); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY |
|
FPAY | Fee payment |
Year of fee payment: 8 |
|
FPAY | Fee payment |
Year of fee payment: 12 |