Arbitrary phase profile for better equalization in dynamic gain equalizer

Description

FIELD OF THE INVENTION

The present invention relates to an apparatus for improving equalization within a Dynamic Gain Equalizer (DGE). More particularly, this invention relates to an arbitrary phase profile for better equalization in a DGE.

BACKGROUND OF THE INVENTION

Designers and inventors have sought to develop a light modulator which can operate alone or together with other modulators. Such modulators should provide high operating speeds (KHz frame rates), a high contrast ratio or modulation depth, have optical flatness, be compatible with VLSI processing techniques, be easy to handle and be relatively low in cost. Two such related systems are found in U.S. Pat. Nos. 5,311,360 and 5,841,579 which are hereby incorporated by reference.

According to the teachings of the '360 and '579 patents, a diffractive light modulator is formed of a multiple mirrored-ribbon structure. An example of such a diffractive light modulator 10 is shown in FIG. 1. The diffractive light modulator 10 comprises elongated elements 12 suspended by first and second posts, 14 and 16, above a substrate 20. Thc substrate 20 comprises a conductor 18. In operation, the diffractive light modulator 10 operates to produce modulated light selected from a reflection mode and a diffraction mode.

FIGS. 2 and 3 illustrate a cross-section of the diffractive light modulator 10 in a reflection mode and a diffraction mode, respectively. The elongated elements 12 comprise a conducting and reflecting surface 22 and a resilient material 24. The substrate 20 comprises conductor 18.

FIG. 2 depicts the diffractive light modulator 10 in the reflection mode. In the reflection mode, the conducting and reflecting surfaces 22 of the elongated elements 12 form a plane so that incident light I reflects from, the elongated elements 12 to produce reflected light R.

FIG. 3 depicts the diffractive light modulator 10 in the diffraction mode. In the diffraction mode, an electrical bias causes alternate ones of the elongated elements 12 to move toward the substrate 20. The electrical bias is applied between the reflecting and conducting surfaces 22 of the alternate ones of the elongated elements 12 and the conductor 18. The electrical bias results in a height difference between the alternate ones of the elongated elements 12 and non-biased ones of the elongated elements 12. A height difference of a quarter wavelength λ/4 of the incident light I produces maximum diffracted light including plus one and minus one diffraction orders, D₊₁ and D₋₁.

FIGS. 2 and 3 depict the diffractive light modulator 10 in the reflection and diffraction modes, respectively. For a deflection of the alternate ones of the elongated elements 12 of less than a quarter wavelength λ/4, the incident light I both reflects and diffracts producing the reflected light R and the diffracted light including the plus one and minus one diffraction orders, D₊₁ and D₋₁. In other words, by deflecting the alternate ones of the elongated elements 12 less the quarter wavelength λ/4, the diffractive light modulator 10 produces a variable reflectivity.

In WDM (wavelength division multiplex) optical communication, multiple component wavelengths of light each carry a communication signal. Each of the multiple component wavelengths of light form a WDM channel. A dynamic gain equalizer (DGE) can he used for WDM signal management. A variety of dynamic equalization techniques have been advanced, which seek to equalize component signals in a WDM system. Most rely on some spectral multiplexer/de-multiplexer component, followed by an electrically-controllable variable optical attenuator which can operate on the de-multiplexed channels (or possibly a band of channels). Diffractive light modulators arc often used as the variable optical attenuator within a DGE. Each channel is directed to a corresponding portion of the diffractive light modulator. To maximize space, each channel partially overlaps an adjacent channel as the channels impinge the diffractive light modulator. Overlapping channels is useful to minimize the number of required ribbons. If channels are not overlapped, then the optical path has to be increased, which leads to a larger optical package.

What is needed is a method and apparatus for improving the accuracy of a DGE that utilizes overlapping channels. What is also needed is a method and apparatus for reducing the computational power required for DGE that utilizes overlapping channels.

SUMMARY OF THE INVENTION

An embodiment of the present invention includes a device for selectively adjusting power levels of component signals of a wavelength division multiplexed signal including a first wavelength signal and a second wavelength signal. The device includes a light modulator comprising a plurality of elements. The plurality of elements are configured to form an arbitrary phase profile. The plurality of elements includes a first group of elements configured to receive the first wavelength signal and a second group of elements configured to receive the second wavelength signal. The first group of elements and the second group of elements include at least one common element. Each element is controllable such that each group of elements directs a selected portion of a corresponding received wavelength signal in a first mode. Each first mode is collected such that a power level of each wavelength signal is selectively adjusted.

The plurality of elements can be arranged in parallel and each element can include a light reflective planar surface with the light reflective planar surfaces lying in one or more parallel planes. The first group of elements are in series with the second group of elements, where the common elements are the elements of the first and second groups of elements that are closest to each other. The light modulator can also include a support structure coupled to each end of the plurality of elements to maintain a position of each element relative to each other and to enable movement of each of the plurality of elements in a direction normal to the one or more parallel planes of the plurality of elements. Each element can also include a first conductive element and the light modulator can also include a substrate coupled to the support structure. The substrate can also include a second conductive element such that in operation an electric bias applied between the first conductive element and the second conductive element enables individually controlled movement of each of the plurality of elements. The light reflective planar surface can include the first conductive element. The arbitrary phase profile determines the portion of the received wavelength signal that is selectively directed in the first mode. A remaining portion of the received wavelength signal can be randomly scattered away from the first mode.

The first mode can be a reflection mode in which the plurality of elements are configured to reflect the selected portion of the received wavelength signal as a plane mirror. The first mode can also be a diffraction mode in which the plurality of elements are configured to diffract the selected portion of the received wavelength signal. The remaining portion can be randomly scattered by diffraction. The light modulator can be a diffractive light modulator. The diffractive light modulator can be a grating light valve type device.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 illustrates an exemplary diffractive light modulator.

FIG. 2 illustrates a cross-section of the diffractive light modulator in a reflection mode.

FIG. 3 illustrates a cross-section of the diffractive light modulator in a diffraction mode.

FIG. 4 illustrates overlapping channels impinging the diffractive light modulator that imparts a square well profile.

FIG. 5 illustrates overlapping channels impinging the diffractive light modulator that imparts an arbitrary phase profile.

FIG. 6 illustrates an exemplary energy profile for a given channel.

FIG. 7 illustrates the diffractive light modulator imparting an arbitrary phase profile for two overlapping channels and the corresponding energy profiles of the two channels.

DETAILED DESCRIPTION OF THE EMBODIMENTS

Preferably, the present invention relates to methods of equalizing channels within a multi-channel environment in which adjacent channels partially overlap. One method of equalizing utilizes a diffractive light modulator operating such that alternating elongated elements are active. Such an alternating pattern creates a square well grating where the depth of each well is determined by the amount of deflection of the corresponding active element. The remaining, non-active elements preferably lie in a plane that forms a planar top layer of the square well grating. Preferably, the diffractive light modulator is a grating light valve type device and the elongated elements are ribbons of the grating light valve type device. The amount that each ribbon is deflected can be considered as a variable for purposes of mathematically determining the necessary attenuation for each channel. In such a usage, only the variables corresponding to active ribbon deflections (i.e. half of the total number of ribbons in this first method) are available for achieving equalization.

It should be born in mind that terms like “equalize” and “equalization” as used with respect to the present invention are to be broadly interpreted with respect to regulating the power levels of component light signals to any pre-determined level of relative power levels. Accordingly, the term “equalize” as used herein is not to be limited to any one particular curve or ratio, but simply constitutes a regulation or normalization of signal power against any pre-determined curve or ratio of power levels at different frequencies.

FIG. 4 illustrates a cross-sectional view of a first three channels impinging a grating light valve type device. The grating light valve type device in FIG. 4 operates according to the first method in which alternating ribbons are active. Preferably, each channel impinges six ribbons, three of which are active. It is understood that the number of ribbons corresponding to each channel can be more, or less, than six. Each channel partially overlaps an adjacent channel as the channels impinge the grating light valve type device. Overlapping channels is useful to minimize the number of required ribbons. If channels are not overlapped, then a longer optical path is required, which leads to a larger optical package. Preferably, the adjacent channels overlap such that they share two ribbons. As shown in FIG. 4, channel 1 impinges ribbons 1-6, channel 2 impinges ribbons 5-10, and channel 3 impinges ribbons 9-14. It is understood that more, or less, channels can impinge the grating light valve type device, each channel impinging a corresponding six ribbons. Channel 1 and channel 2 share ribbons 5 and 6, channel 2 and channel 3 share ribbons 9 and 10, and so on.

Ribbons 2, 4, 6, 8, 10, 12, 14, etc. are the alternating ribbons which are active. Each of the ribbons 2, 4, 6, 8, 10, 12, 14 can be individually moved by a distance d. Each active ribbon i can be moved a distance di. A phase profile of the grating light valve type device is that of a square well grating where di is varied. An incident light impinging the grating light valve type device can be attenuated by deflecting the active ribbons on which the incident light impinges. The amount of attenuation is determined by the distance that each active ribbon is moved, or deflected. Therefore, attenuation is a function of the deflection distance. Channel 1 impinges the active ribbons 2, 4, and 6. The ribbons 2, 4, and 6 can be deflected a distance d1, d2, and d3, respectively, as shown in FIG. 4. The distances d1, d2, and d3 shown in FIG. 4 are arbitrary and the actual values of d1, d2, and d3 are dependent on the amount that channel 1 is to be attenuated.

Each incoming channel can have a different intensity, I. In most cases, the intensity of each channel is different. For example, the intensity of channel 1, I(Ch 1), does not equal the intensity of channel 2, I(Ch 2). A portion of channel 1 and a portion of channel 2 are shared due to overlapping. As a result, a portion of channel 1 and a portion of channel 2 experience the same common attenuation resulting from the shared portion. This is illustrated in FIG. 4. Channels 1 and 2 share ribbons 5 and 6. Of the shared ribbons 5 and 6, ribbon 6 is active. Therefore, any attenuation attributed to the deflection of ribbon 6 applies equally to both channel 1 and channel 2. Since this shared portion impacts both channels, the shared portion can be best utilized for performing a macro attenuation of both channels. This leaves the remaining, non-shared portions of each channel to perform any fine-tuning, or micro attenuation, of the channel.

An output intensity for a given channel h is a factor of the distance di of each ribbon corresponding to channel h. In the first method, alternating ribbons are deflected and the remaining ribbons are fixed. In the preferred case where there are 6 ribbons per channel, 3 of the 6 ribbons are deflected, and the output intensity for channel h is a factor of the distance di for 3 ribbons. Essentially, there are 3 variables to equalize the output intensity of channel h.

In general, where channel h impinges n ribbons, the output intensity for channel h, OI(Ch h) is a function of di where i=1 to n/2. In other words, there are n/2 variables that can be used to determine the output intensity, OI(Ch h). The equalization process is controlled by n/2 variables.

Further, since channels are overlapping on the grating light valve type device, the number of variables that can be used to independently control the output intensity for a specific channel is further reduced. For example, in the 6 ribbon per channel case above, only 3 of the ribbons are movable, which constitutes 3 variables. However, 2 ribbons are shared with each adjacent channel, which means 4 of the 6 ribbons for each channel are shared (this is not the case for each of the end channels, channel 1 and channel n, because each of the end channels only has one adjacent channel). Of these 4 ribbons, 2 ribbons are movable, which constitutes 2 variables. This reduces the number of independent variables for each channel from 3 to 1.

The output intensity for each channel can be expressed as a system of equations. The system of equations includes the variables corresponding to the deflection distances of each of the movable ribbons for that channel. The output intensity for channel 1 can be expressed as:
Ch. 1: OI(Ch 1)=a·f(d1)+b·f(d2)+c·f(d3)  (1)
where f(d1) represents the output intensity corresponding to ribbons 1 and 2 and is a function of the distance d1, f(d2) represents the output intensity corresponding to ribbons 3 and 4 and is a function of the distance d2, and f(d3) represents the output intensity corresponding to ribbons 5 and 6 and is a function of the distance d3. Similarly, the output intensities for channels 2 and 3 can be expressed as:
Ch2: OI(Ch 2)=c′·f(d3)+d′·f(d4)+e′·f(d5)  (2)
Ch3: OI(Ch 3)=e″f(d5)+f″·f(d6)+g″·f(d7).  (3)
Additional channels can be similarly expressed, each equation including three variables in the case of the first method. As can be seen from the equations 1 and 2, the output intensities of channels 1 and 2 include a common element, f(d3). Channels 1 and 2 constrain each other because of the common element f(d3). As can be seen from equations 2 and 3, the output intensities of channels 2 and 3 include a common portion, f(d5). Channels 2 and 3 constrain each other because of the common element f(d5). The system of equations can be expanded to the general case where there are x total ribbons within the grating light valve type device. In this generalized case of the first method, there are x/2 total variables to be solved by the system of equations. Because these equations are interdependent, they must be solved simultaneously. They can not be solved independently.

The system of equation are solved iteratively by an equalization algorithm, so they are said to converge. With iterative problem solving, there is error. Therefore, for any number of iterations there is an associated error. Reducing the error is advantageous to better equalize the channels to a specific level. However, there is a cost to reducing the error. To reduce the error, more iterations are required. Each iteration requires computational power to perform. Therefore, to reduce the error requires additional computational power. Typically, a finite amount of computational power is allocated to solve each system of equations. At times, the allocated computational power is not sufficient to reduce the error below an acceptable threshold. In these cases, the system of equations is said to be unsolvable.

The present invention provides another, and preferred, method that increases the number of variables. This provides more degrees of freedom to solve the same system of equations. The number of variables is increased by utilizing a diffractive light modulator operating such that each elongated element is active. Preferably, the diffractive light modulator is a grating light valve type device and the elongated elements are ribbons of the grating light valve type device. By enabling all ribbons to be active, this creates an arbitrary phase profile as opposed to the square well profile of the first method. The arbitrary phase profile is dictated by the system of equations, but the system of equations now has more variables. Where channel h impinges n ribbons, there are n variables. This provides twice the degree of freedom as compared to the first method where only alternating ribbons are active. This allows for better convergence characteristics of the equalization algorithm, e.g. lower ripple and/or faster convergence time.

FIG. 5 illustrates a cross-sectional view of a first three channels impinging a grating light valve type device. The grating light valve type device in FIG. 5 operates according to the second method in which each ribbon is active. The channels in FIG. 5 impinge the grating light valve type device the same as the channels in FIG. 4. Preferably, each channel impinges six ribbons, all six of which are active. It is understood that the number of ribbons corresponding to each channel can be more, or less, than six.

Each of the ribbons can be individually moved by a distance d. Each active ribbon i can be moved a distance di. A phase profile of the grating light valve type device is that of an arbitrary phase profile where di is varied. Channel 1 impinges the ribbons 1-6. The ribbons 1-6 can be deflected a distance d1-d6, respectively, as shown in FIG. 5. The distances d1-d6 shown in FIG. 5 are arbitrary and the actual values of d1-d6 are dependent on the amount that channel 1 is to be attenuated.

A portion of channel 1 and a portion of channel 2 are shared due to overlapping. As a result, a portion of channel 1 and a portion of channel 2 experience the same common attenuation resulting from the shared portion. This is illustrated in FIG. 5. Channels 1 and 2 share ribbons 5 and 6. Therefore, any attenuation attributed to the deflection of ribbon 5 and 6 applies equally to both channel 1 and channel 2. In this second method, 2 active ribbons are shared by each pair of adjacent channels, as opposed to the single active ribbon that is shared in the first method. This improves the degree of freedom in applying the macro attenuation to the channels 1 and 2.

In the preferred case where there are 6 ribbons per channel, all 6 ribbons are deflected in the preferred method, and the output intensity for channel h is a factor of the distance di for 6 ribbons. Essentially, there arc 6 variables to equalize the output intensity of channel h.

In general, where channel h impinges n ribbons, the output intensity for channel h, OI(Ch h), is a function of di where i=1 to n. In other words, there are n variables that can be used to determine an output intensity, OI(Ch h). The equalization process is controlled by n variables in the second method as opposed to n/2 variables as in the first method.

The output intensity for each channel can again be expressed as a system of equations. The system of equations includes the variables corresponding to the deflection distances of each of the movable ribbons for that channel. The output intensity for channel 1 can be expressed as:
OI(CH 1)=a·f(d1)+b·f(d2)+c·f(d3)+d·f(d4)+e·f(d5)+f·f(d6)  (4)
where f(d1) represents the output intensity corresponding to ribbon 1 and is a function of the distance d1, f(d2) represents the output intensity corresponding to ribbon 2 and is a function of the distance d2, f(d3) represents the output intensity corresponding to ribbon 3 and is a function of the distance d3, f(d4) represents the output intensity corresponding to ribbon 4 and is a function of the distance d4, f(d5) represents the output intensity corresponding to ribbon 5 and is a function of the distance d5, and f(d6) represents the output intensity corresponding to ribbon 6 and is a function of the distance d6. Similarly, the output intensity for channel 2 can be expressed as:
OI(Ch 2)=e′·f(d5)+f′·f(d6)+g′·f(d7)+h′·f(d8)+i′·f(d9)+j′·f(d10).  (5)
Additional channels can be similarly expressed, each equation including six variables in the case of the second method. As can be seen from the equations 4 and 5, the output intensities of channels 1 and 2 include common elements f(d5) and f(d6). Channels 1 and 2 constrain each other because of the common elements f(d5) and f(d6). The system of equations can be expanded to the general case where there are x total ribbons within the grating light valve type device. In this generalized case of the first method, there are x total variables to be solved by the system of equations. Because these equations are interdependent, they must be solved simultaneously. They can not be solved independently.

Although each channel impinges 6 ribbons, the intensity of the incoming channel is not evenly distributed across each of the 6 ribbons. An exemplary energy distribution of an incoming channel is illustrated in FIG. 6. The energy distribution is typically a gaussian distribution including a maxima and trailing edges. As can be seen in FIG. 6, it is preferable that the energy of a single channel is not entirely directed onto the ribbons corresponding to that channel. The trailing edges of the energy distribution “leak” into the adjacent channel on either side. It is understood that the energy distribution illustrated in FIG. 6 can be a different function, or shape. It is also understood that the amount of the trailing edges that leaks onto adjacent channels can be more, or less, than that illustrated in FIG. 6.

FIG. 7 illustrates the cross-sectional view of the channels 1 and 2 impinging the grating light valve type device as in FIG. 5 where the channels 1 and 2 arc represented by their energy distributions, as in FIG. 6. Consider the energy peak for each of the channels 1 and 2 and the relation of each of the peaks to the corresponding ribbons. For channel 1, the deflection distance of ribbon 3, d3, and ribbon 4, d4, are more influential in determining the output intensity of channel 1, OI(Ch 1), than the deflection distance of ribbon 2, d2, and ribbon 5, d5. This is because the peak of the incoming channel energy is directed onto ribbons 3 and 4. Similarly, d2 and d5 are more influential in determining OI(Ch 1) than the deflection distance of ribbon 1, d1, and ribbon 6, d6. Therefore, in regards to channel 1, d5 is more strongly coupled to d4 than d6 is to d4. For channel 2, the peak of the energy distribution is directed onto ribbons 7 and 8. Therefore, the deflection distance of ribbon 7, d7, and ribbon 8, d8, are more influential in determining OI(Ch 2) than are the deflection distances of ribbon 6, d6, and ribbon 9,d9. Similarly, d6 and d9 are more influential in determining OI(Ch 2) than d5 and the deflection distance of the ribbon 10, d10. Therefore, in regards to channel 2, d6 is more strongly coupled to d7 than d5 is to d7. Recall that ribbons 5 and 6 are both shared by channels 1 and 2, and therefore d5 and d6 each impact OI(Ch 1) and OI(Ch 2). However, since d5 is more strongly coupled to d4 and d6 is more strongly coupled to d7, d5 can be set to better follow the requirements of channel 1 and d6 can be set to better follow the requirements of channel 2. This further increases the degree of freedom in solving the system of equations for OI(Ch 1) and OI(Ch 2).

It is understood that the orientation by which the channels 1 and 2 impinge the grating light valve type device can be different than that illustrated in FIG. 7. For example, the energy distributions for channel 1 and/or channel 2 can be shifted to the right or to the left. It is further understood that the shape of the energy distribution curve of channel 1 and channel 2 can be different than that illustrated in FIG. 7.

As with the first method, the system of equations associated with the second method are solved iteratively. It is expensive to compute the system of equations and therefore determine the value of each deflection distance di. Expensive in this case refers to computing power. By doubling the number of variables to n, more variations are available to hasten the convergence of the equations. The number of iterations needed is determined by the acceptable error tolerance. With a fixed number of variables, the more iterations that are performed means the more computational power is required. However, by increasing the number of variables, as in the second method, the impact of each iteration is greater. Although each iteration requires more computational power because of the increased number of variables, the increased impact of each iteration is greater than the increased computational power necessary to perform each iteration. As a result, there is a net improvement using the second method by increasing the number of variables. This improvement can be used to decrease the number of iterations required to reach the same error tolerance, which reduces the necessary computational power. Or, for the same number of iterations, the size of the error can be reduced. A combination of decreasing computational power and reducing the error can also by used. This improvement also enables some previously unsolvable systems of equations to be solved by reducing the error to within acceptable error tolerances.

In summary, for the same error, it takes fewer iterations. For the same number of iterations, the error is reduced. By deflecting each ribbon, the ability to achieve specified levels of equalization are improved and/or thc computational power requirements are reduced. The specific values of di are determined by the specific error tolerance required.

The preferred light modulator of the present invention preferably utilizes the second method to create an arbitrary phase profile that essentially forms a “rough surface.” The rough surface causes the incident light to scatter. Preferably, the light modulator is used within a DGE in which a normal incident light is diffracted and the zero order light is collected while the attenuated light is scattered. Alternatively, the light modulator can be configured to receive an incident light off-axis to normal. In this case, there is still zero order light, but it is reflected at an angle.

Although the methods and apparatus of the present invention are intended to be used with overlapping channels, the present invention can also be used as a variable scatterer to attenuate a channel that is not overlapping.

It will be readily apparent to one skilled in the art that other various modifications may be made to the embodiments without departing from the spirit and scope of the invention as defined by the appended claims.