WO2002037624A2

WO2002037624A2 - Wavelength-selective demultiplexer

Info

Publication number: WO2002037624A2
Application number: PCT/US2001/023980
Authority: WO
Inventors: Zelda Gills; Gayle R. Link; Calvin J. Martin
Original assignee: Luxcore Networks, Inc.
Priority date: 2000-11-02
Filing date: 2001-07-31
Publication date: 2002-05-10
Also published as: AU2001280919A1; WO2002037624A3

Abstract

A wavelength demultiplexer (100) includes a beam splitter (105) with an input port (110) receiving a multiplexed signal, and a plurality of output ports for outputting demultiplexed signals. The output ports (120, 140, 160) contain wavelength-selective elements (130, 150, 170), such as tunable band pass filters, and, optionally, shutters. To compensate for the power losses within the demultiplexer, active optical fiber (175) amplifies the multiplexed signal. The demultiplexer may be used in WDM optical transmission systems as a wavelength-selective switch.

Description

WAVELENGTH-SELECTIVE DEMULTIPLEXER

BACKGROUND OF THE INVENTION

1. Field of the Invention

The present invention relates to optical signal processing and, more particularly, to wavelength-selective demultiplexing and switching in optical networks.

2. Background

The explosive growth of telecommunications is, to a large degree, both a cause and an effect of the proliferation of fiber optic communication systems. Because of its many advantages, optical fiber has now been used in telecommunications for approximately three decades. The advantages include low signal attenuation, immunity to electromagnetic interference (EMI), low crosstalk, fast propagation speed, physical flexibility, small size and low weight — all at a reasonable cost.

In a typical optical network, light modulated with a data signal is coupled to a fiber at a source node, transmitted by a fiber to a destination node, possibly through several intermediate nodes, received at the destination node, demodulated and converted into an electrical data signal. In the present context, "light" includes infrared light; in fact, two of the more commonly used bands are centered around 1550 nanometers and 1310 nanometers, both lying in the infrared region of the electromagnetic spectrum. Because of the continuing growth of telecommunication services, service providers need to accommodate ever-higher bandwidths requirements. At this time, bandwidth available on a single wavelength channel (i.e., on a single transmission frequency) is increasing from 10 Gbits/s (OC-192) to 40 Gbits/s (OC- 768). These rates, however, are just small fractions of the total bandwidth potentially available from an optical fiber, which is of the order of 20 Terahertz. As the need for more bandwidth exerts its relentless pressure, wavelength division multiplexed (WDM) systems have evolved to provide more carrying capacity from a single fiber. In WDM systems, separate data channels are transmitted through the same fiber on different wavelengths. Such expansion of capacity from existing fiber networks may reduce the need to install more fiber. Moreover, the use of WDM overcomes bandwidth limitations of the existing electronic end-point equipment because each of the bandwidth channels can be processed separately.

As more and more distinct channels are squeezed into a single fiber, narrowband wavelength division multiplexed (NWDM) systems become dense wavelength division multiplexed (DWDM) systems. But be it few or many, the distinct wavelength channels need to be multiplexed for transmission, and demultiplexed when received. These functions are performed by multiplexers and demultiplexers, respectively. Wavelength demultiplexers thus are the network elements that separate the various wavelength channels carried by a fiber in a WDM network.

Wavelength division multiplexed systems also provide much needed flexibility in selecting protocol and network topology. Both topology and protocol selection are severely restricted in telecommunication systems where data of multiple channels are embedded in the same stream. An example of such transmission scheme is synchronous optical network/synchronous digital hierarchy (SONET/SDH), a three-layer transport network architecture. In a SONET/SDH network, individual data flows, e.g., tributaries, are mapped into payloads and transported across the network's spans in envelopes, in a synchronous time division multiplexed (TDM) manner. The data flows of a SONET/SDH network must therefore be extracted from the payloads before they can be switched individually.

Because each multiplexed wavelength channel is independent from other channels, its data format and bit rate can be independent from formats and rates of other channels propagating in the same fiber. For example, one fiber can carry κ_1} κ₂, and κ₃ wavelength channels, where Ki is a 2.5 Gbit/s SONET OC-48 channel, κ₂ is a 10 Gbit/s SONET OC-192 channel, and κ₃ is a proprietary format channel. Unlike multiplexed data flows carried by the same wavelength, each wavelength channel can be optically routed or switched. In other words, each wavelength channel can be switched independently. Independent switching avoids the need for opto-electric (O-E) conversion of all the data carried by the fiber, electronic processing of the data, and subsequent electro-optic (E-O) conversion for further transmission. The conversions and electronic processing typically require arrays of photodetectors and transponders. Photodetectors optically detect signals, and translate them into electrical signals that can be demultiplexed and switched electronically. Transponders can then be employed to receive the separate wavelength channels and translate them to different wavelengths for subsequent multiplexing and transmission through appropriate fibers.

The use of photodetector and transponder arrays is expensive. Even more important is that photodetectors and transponders are usually wavelength-specific components, requiring a priori knowledge of the wavelengths. Switching flexibility is therefore lost. And redundancy, often needed for reliability expected from modern providers of telecommunication services, becomes a rather costly one-to-one redundancy.

To make full use of the above-described advantages offered by WDM, many optical networks implement all-optical wavelength-based routing (or wavelength routing) architectures. In such networks, switches, e.g., optical cross- connects (OXCs), need the ability to separate the multiple wavelength channels carried by one fiber, and to route the channels through different fibers.

The ability to separate channels based on wavelength is thus important to various elements in WDM networks, including demultiplexers and optical cross connects. (These and similar network elements that perform channel separation and re-direction will be referred to as switches.) To take full advantage of the routing flexibility afforded by WDM, it is important to be able to configure these network elements dynamically, essentially in real time, and without manual intervention. Note also that demultiplexing and switching are usually performed on received channels, i.e., channels that have already traveled some distance and, consequently, have been attenuated. It is, therefore, preferable to separate the channels with as little additional attenuation as possible.

Several methods of channel separation are known. One method spatially separates different wavelengths in arrayed waveguide gratings (AWGs), also known as diffraction gratings. A diffraction grating is a periodic arrangement of elements that reflect or transmit electromagnetic radiation. The elements are spaced apart by a distance comparable to the wavelengths of the radiation. For example, a pattern of transparent slits in an opaque screen, or a pattern of reflecting grooves on a substrate could be used to build a grating. When coherent radiation incident on a grating diffracts, the resulting multiple fields interfere constructively or destructively in different locations, creating a wavelength-dependent pattern of peaks and nulls. Different wavelength signals can be spatially separated in a grating because the peaks of the different wavelengths occur in different locations. A diffraction grating of a high diffraction order (i.e., having many elements) can achieve high wavelength resolution. For a theoretical treatment of diffraction gratings see SIMON RAMO ET AL., FIELDS AND WAVES IN COMMUNICATION ELECTRONICS 673-76 (John Wiley & Sons 1965).

Diffraction gratings for wavelength demultiplexing can be expensive to manufacture and difficult to tune, even within a narrow range. Additionally, the various ports are not independently tunable. Therefore, a diffraction grating-based switch will generally perform only static routing, with specific wavelength channels being transmitted to specific physical outputs. Diffraction gratings are also rather lossy.

Another method of separating wavelength channels employs optical interleavers. An interleaver is essentially a l-by-2 comb filter with periodically spaced band pass and band reject sections. The general principle underlying interleaver technology is an interferometric overlap of two beams. The overlap causes periodic interference, allowing alternating wavelength channels to pass from an input of the device to one output, and allowing complementary alternating channels to pass from the input to a second output. Controlling the fringe pattern sets the desired channels of the device.

A significant problem with interleavers is that a single interleaver can separate wavelengths channels into only two sets of channels, with a relatively wide channel spacing. Thus, multiple interleaver stages are required to separate individual channels. This increases, cost, size, and signal loss. Like diffraction gratings discussed above, interleavers are difficult to tune or reconfigure, requiring a priori knowledge of wavelength-port assignments. A major benefit of the WDM scheme — dynamic routing — is therefore lost.

Another channel separation method is described in U.S. Patent No. 6,005,992 to Augustsson et al. (the " '992 patent" hereinafter), which patent is hereby incorporated by reference as if fully set forth herein. A wavelength- selective switch according to the '992 patent uses multimode interference (MMI) waveguides, Bragg gratings, Mach-Zehnder waveguides, and controllable phase shifters. Briefly, the switch works as follows. A multi-wavelength signal introduced into an MMI waveguide produces a number of images equal to the number of wavelengths sought to be separated. Each image is then passed through a different series of Bragg gratings and controllable phase shifters. Bragg gratings reflect specific wavelengths, allowing all other wavelengths to pass through. The phase of each reflected signal depends on the phase shifters through which the reflected signal has passed. When the reflected signals return to the MMI waveguide, they exit out of the different ports of the MMI waveguide, depending on their phase relationships. Thus, because the different wavelengths pass through different phase shifters, incurring different phase shifts, they exit out of different ports.

The specification of the '992 patent expounds that invention in more detail. Here, we simply note that a number of problems with that scheme exist, and briefly explain two of them: signal attenuation and scalability.

First, the switch of the '992 patent is inherently lossy because of the optical power split (1/N) at the MMI waveguide. In addition, the signals are subjected to the attenuation of several phase shifters and Bragg gratings.

Second, the switch scales rather poorly. As the '992 patent acknowledges, to separate N channels, the switch has N x (N-l) Bragg gratings and N² phase shifters. Clearly, this second-order dependence of the numbers of gratings and phase shifters on the number of channels quickly become unwieldy as the number of channels increases beyond eight. Moreover, the power losses in the switch increase with N because of the power split and also because of the increased number of components that each signals has to pass through.

Yet another wavelength-selective switch is described in U.S. Patent No. 5,446,809 to Fritz et al. (the " '809 patent"). The switch of the '809 patent relies on straightforward power splitting and filtering by Bragg gratings with piezo-electric tuners. Because the switch of the '809 patent is architecturally somewhat similar to the switch of the '992 patent described above, it is not surprising that the '809 patent switch has some of the same disadvantages as the switch of the '992 patent: (1) the number of Bragg gratings grows as a square of the number of channels, and (2) the losses incurred depend on the optical power split and on the number of components, increasing with the increase in each of these factors.

Another type of a wavelength-selective switch is a fiber-based switch with impressed tunable gratings. Such a switch is architecturally similar to the switch described in the '809 patent, using multiple cascaded 1 x N and N x 1 couplers and isolators or circulators. Scalability is therefore also a major problem with this switch. Furthermore, because output ports and input ports of this switch are connected by cascading couplers, the switch requires elaborate active real time control of filters to block unwanted wavelength transmissions. Other problems with this scheme include wavelength contention, back reflectance, and optical power losses.

A need therefore exists for a wavelength-selective switch that scales well, has relatively low power losses, and is dynamically reconfigurable.

SUMMARY OF THE INVENTION In accordance with the broad principles of this invention, a beam splitter- based wavelength demultiplexer has an input for receiving multiplexed signals and a plurality of outputs for outputting the signals. At least some of the outputs are equipped with wavelength-selective elements, such as filters. The demultiplexer also includes an active fiber filler coupled to the input of the demultiplexer, so that the active fiber filler amplifies the multiplexed signals when they pass through the filler.

BRIEF DESCRIPTION OF THE DRAWINGS

The present invention will be explained, by way of examples only, with reference to the following description, appended claims, and accompanying figures where:

Figure 1 illustrates an optical wavelength channel demultiplexer;

Figure 2 illustrates an output port of the optical wavelength channel demultiplexer of Figure 1;

Figure 3 illustrates an optical wavelength channel demultiplexer with an integral optical pump for exciting active fiber within the demultiplexer; and

Figure 4 illustrates a processor-controlled 1 x N wavelength-selective switch operating in conjunction with an optical channel supervisor and a processor. DETAILED DESCRIPTION Figure 1 is a diagram showing an embodiment of an all-optical wavelength selective demultiplexer 100. The demultiplexer includes a beam splitter 105 with an input port 110 and output ports 120, 140, and 160. Active fiber filler 175 occupies part or all of the internal portion of the beam splitter 105. Each of the output ports 120, 140, and 160 feeds into one of wavelength-selective elements 130, 150, and 170.

A cutout of a representative output port 120 and its wavelength-selective element 130 are shown in Figure 2. The wavelength-selective element 130 includes a band pass filter 132 and a shutter 134. The shutter 134 is essentially an ON/OFF switch, as its name implies. It either allows incident light to pass through, or blocks the light. Note that the specific locations of the filter 132 and the shutter 134 within the wavelength-selective element 130 are not critical, and that the order in which the filter and the shutter appear may be reversed. The shutter 134 can be an electro-absorptive modulator using a shutter effect to block or transmit the light selectively. An electro-absorptive modulator becomes either transparent or absorptive (opaque) in response to a voltage applied across it. For real time ON/OFF switching, a lithium niobate modulator can be used as a shutter. Lithium niobate modulators, often used to modulate laser transmitters in 2.5 Gbit/s and even faster WDM systems, have response time of the order of 100 picoseconds, generally fast enough for dynamic wavelength routing. A gallium arsenide indium phosphate modulator can also be used as the shutter.

The shutter can be a micro-electromechanical system (MEMS). A MEMS- based shutter is described in U.S. Patent No. 6,173,105 to Aksyuk, et al. MEMS rotary structures that may be used in building a shutter are described in U.S. Patent No. 6,137,206 to Hill. Each of these patent documents is hereby incorporated by reference as if fully set forth herein. A shutter can also be a digital mirror device (DMD), a micro-mechanical spatial light modulator array of small mirrors (or a single mirror) supported above silicon addressing circuitry by small hinges attached to a support post. Each mirror can be made to rotate about its axis by, for example, electrostatic, electromagnetic, piezoelectric, or thermo-mechanical actuation.

The above discussion of optical shutters is, of course, far from exhaustive. The filter 132 is a band pass filter, i.e., a line filter that transmits wavelengths within a specific range, rejecting some or all other wavelengths. In some applications, other filters — e.g., band reject, low pass, or high pass — may suffice. The filter may be a Fabry-Perot resonator (an etalon), i.e., an optical resonator formed by mirrors. For increased flexibility of the demultiplexer, the filter 132 can be a tunable filter. Fabry-Perot resonators can be tuned with low voltage piezoelectric actuators varying the gap between a resonator's mirrors by positioning one or more of the mirrors.

A Fabry-Perot filter can also be tuned by inserting a liquid crystal layer between the opposed mirrors of the filter, and then applying an electric field across the layer. The electric field changes the refraction index (permittivity and/or permeability and the wavelength of interest) of the liquid crystal material, thus changing the resonant frequency of the cavity. Tunable Fabry-Perot liquid crystal filters are described in, for example, U.S. Patents with numbers 5,068,749 and 5,111,321, both to Patel, and U.S. Patent No. 6,154,591 to Kershaw. These patents are also incorporated by reference as if fully set forth herein.

Another type of optical filter is a tunable acousto-optical filter. Acousto- optical filters operate based on elasto-optical effect, which is the phenomenon of changing refraction index of a material in response to physical stresses. In acousto-optical filters, radio frequency waves are used to generate surface acoustic waves in an electro-optic medium, such as LiNbO crystal. The compressions and rarefications of the surface acoustic waves create a temporary grating within the crystal. The temporary diffraction grating works like its more permanent counterpart discussed above; that is, it spatially separates the various wavelengths of a diffracted optical signal. The temporary grating is tuned by controlling the radio frequency emitter.

United States Patent No. 6,157,025 to Katagiri teaches a disc-shaped transparent substrate with an optical filter layer deposited on the substrate. The filter layer is such that the center wavelength of the band pass region varies with the angular dimension of the filter. By rotating the filter in relation to a light beam incident upon it, different angular portions of the disc perform the filtering function. Therefore, different wavelengths can be selected by rotating the disc.

More generally, a tunable filter can be realized in an arrangement that allows physical movement of a filter element in some dimension in relation to an optical path of a beam of light being filtered. If the wavelength of the band pass region of the filter element is a variable of the dimension, the filter can be tuned by controlling an actuator that moves the filter element in the dimension of interest. The actuator may include a servomechanism, a position encoder, and a controller. The servomechanism moves the filter element, whose position the encoder senses and transmits to the controller. The controller receives the position data from the encoder and directs the servomechanism to position the filter element in accordance with an input signal. See U.S. Patent No. 6,111,997 to Jeong for examples of such tunable filters. Yet another example of a tunable optical filter is found in U.S. Patent No.

6,058,226 to Starodubov. Starodubov teaches an optical fiber including a core covered by a cladding. A grating within the core couples light at some resonant wavelength either into the cladding or into a coating surrounding the fiber adjacent to the grating. The resonant wavelength depends on the refractive index of the coating. The coating is made of a material whose refractive index is a function of an externally controllable stimulus, such as an electric or a magnetic field.

A tunable optical filter somewhat similar to that taught by Starodubov is disclosed in U.S. Re-examined Patent No. RE. 36,710 to Baets et al. Baets's filter is also based on a tunable optical grating embedded in a multi-waveguide structure. Another type of tunable optical filter uses an optical splitter to divide a beam into several components. The several components are passed through different phase shifters, and then recombined. The combined components interfere constructively or destructively, depending on their relative phases, which, of course, depend on the specific wavelengths carried by the beam. Controlling the phase shifters tunes such interferometric filter to different wavelengths.

Another type of optical filter uses a dielectric multi-layered filter element. Varying the optical lengths of the layers varies the passband of the filter. A simple method of varying the optical lengths of the layers is to change the angle of incidence of a beam upon the filter element. This can be done by, for example, rotating the filter element. See U.S. Patent No. 5,481,402 to Cheng et al. for a polarization-independent tunable filter based on this principle.

Other tunable optical filters exist, including those based on polarization interference effects. But the precise type of filter or filters is not critical to the operation of the present invention. The active fiber filler 175 is essentially an optical amplifier used by the demultiplexer to compensate for some of the losses incurred in the filters of the output ports 120, 140, and 160, and for the losses inherent in the beam splitting arrangement used. Clearly, the demultiplexer can overcompensate, providing a net amplification effect. The demultiplexer can also undercompensate for the losses.

Typical active fiber is fiber doped with rare earth element ions. The doped fiber becomes fluorescent, meaning that it can absorb excitation energy at one wavelength and emit the absorbed energy at a different wavelength. For optical amplification, active fiber is excited or "pumped" by a source of light ("optical pump"), e.g., a diode laser, at a wavelength other than any of the wavelengths of the channels being demultiplexed, elevating the energy states of the fiber's constituent particles. The particles then emit light triggered by the propagating signals at the signals' wavelengths, thus amplifying the signals. Fluorescent dopants often used in active fiber of non-coherent optical systems operating in the 1310 run and 1550 nm bands are erbium and praseodymium.

Active fiber, as most amplifiers, produces spontaneous wideband emissions, i.e., noise. Noise in communication systems is, of course, undesirable. One way to lower an amplifier's noise figure is to pass the amplified signal through a narrow band pass filter. The passband of the filter needs to be at least as broad as the signal, so that the signal is not attenuated. Thus, the filtering approach does not work well for wideband signals.

Although each of the wavelength channels is relatively narrowband, WDM systems may carry several such channels, spaced apart by some buffer bandwidth, e.g., 200 GHz. As was discussed in the BACKGROUND section of this document, the number of channels can be quite high, resulting in a relatively broad bandwidth of the combined optical signal. Therefore, the benefits of the filtering approach may be limited in amplifiers of WDM systems.

The demultiplexer 100, however, provides amplification with a relatively small penalty to the noise figure. This is a result of the per-channel filtering of the amplified signal. Because each filter need pass only one channel (or a subset of channels), the filter can be made relatively narrowband. The noise contribution of active fiber will therefore decrease.

Recall that active fiber needs an optical pump to provide energy for amplifying the signal. The pump can be upstream from the demultiplexer, with the pumping light entering the demultiplexer through the input port 110, together with the wavelength channels. The optical pump can also be part of the demultiplexer assembly. This latter arrangement is illustrated in Figure 3. In that figure, the body of the demultiplexer 100 has an opening 180. Optical pump 190 is coupled to the opening 180 to inject pumping light into the active fiber 175.

The active fiber filler, or another type of optical amplifier, can also be disposed upstream from the demultiplexer 100. In another embodiment, the active fiber filler can be disposed partially inside the demultiplexer and partially outside the demultiplexer. The demultiplexer can include a processor, e.g., a general purpose digital computer, coupled to the tunable filters and/or shutters. The processor can reconfigure the demultiplexer by selecting particular wavelength channels to be output from specific outputs. The processor may also control the amplification of the active fiber filler by varying the intensity of the excitation provided by the optical pump.

Figure 4 illustrates an exemplary application of a demultiplexer 410 used in a processor-controlled 1 x N wavelength-selective switch. Optical channel supervisor 420 is coupled to the aggregate optical stream to register the specific wavelength channels entering the switch. The supervisor 420 may also monitor the signal strength of the aggregate stream or of specific wavelength channels of the stream. Processor 430 communicates with the supervisor 420 and the demultiplexer 410. Upon detection of particular wavelength channels by the supervisor 420, the processor 430 may automatically reconfigure the demultiplexer 410 by tuning the wavelength-selective elements of the demultiplexer 410 to appropriate wavelengths for transmission, blocking, etc. Manual configuration in response to operator input control 440 can also be provided. If the supervisor 420 monitors signal strength and the demultiplexer 410 includes an optical pump, then the processor 430 may also regulate amplification of the signals within the demultiplexer 410 in response to the signal strength data provided by the supervisor 420.

We have described the invention and some of its features in considerable detail for illustration purposes. Neither the specific embodiments of the invention as a whole nor those of its features limit the general principles underlying the invention. In particular, the invention is not limited to specific regions of the light spectrum mentioned in this document, or to use in WDM optical transmission systems. The specific shutters, filters, beam splitters, and active fiber fillers described may be used in some embodiments, but not in others, without departure from the spirit and scope of the invention as set forth. Different geometries of the beam splitter and of the active fiber filler, which may occupy all or part of the splitter, also fall within the intended scope of the invention. Furthermore, the use of shutters on the outputs is optional. Many additional modifications are intended in the foregoing disclosure, and it will be appreciated by those of ordinary skill in the art that in some instances some features of the invention will be employed in the absence of a corresponding use of other features. The illustrative examples therefore do not define the metes and bounds of the invention, which function has been reserved for the following claims and their equivalents.

Claims

We claim:

1. A wavelength demultiplexer comprising: a beam splitter comprising an input for receiving multiplexed signals and a plurality of output paths for outputting the signals, the plurality of output paths comprising a first output path, the first output path comprising a first wavelength filter; and an active fiber filler, the active fiber filler being coupled to the input so that the multiplexed signals pass through the active fiber filler.

2. A wavelength demultiplexer according to claim 1, wherein the plurality of output paths further comprises a second output path, the second output path comprises a second wavelength filter.

3. A wavelength demultiplexer according to claim 2, wherein: the active fiber filler is at least partially disposed inside the beam splitter; and the first and the second wavelength filters are pass band filters.

4. A wavelength demultiplexer according to claim 3, wherein the portion of the active fiber filler disposed inside the beam splitter is excited through the input of the beam splitter.

5. A wavelength demultiplexer according to claim 3, wherein: the first output path further comprises a first shutter; and the second output path further comprises a second shutter.

6. A wavelength demultiplexer according to claim 5, wherein the beam splitter further comprises portions defining an opening, and the wavelength demultiplexer further comprises an optical pump coupled to the opening for exciting the active fiber filler.

7. A wavelength demultiplexer according to claim 6, wherein the active fiber filler comprises erbium doped fiber.

8. A wavelength demultiplexer according to claim 6, wherein the active fiber filler comprises praseodymium doped fiber.

9. A wavelength demultiplexer according to claim 4, wherein the first filter comprises a first Fabry-Perot resonator and the second filter comprises a second Fabry-Perot resonator.

10. A wavelength demultiplexer according to claim 4, wherein: the first filter comprises a first piezo-electrically tunable Fabry-Perot resonator; and the second filter comprises a second piezo-electrically tunable Fabry-Perot resonator.

11. A wavelength demultiplexer according to claim 4, wherein: the first filter comprises a first Fabry-Perot resonator, the first Fabry-Perot resonator comprises a first liquid crystal element for tuning the first Fabry-Perot resonator; and the second filter comprises a second Fabry-Perot resonator, the second

Fabry-Perot resonator comprises a second liquid crystal element for tuning the second Fabry-Perot resonator.

12. A wavelength demultiplexer according to claim 4, wherein the first and the second filters are tunable acousto-optical filters.

13. A wavelength demultiplexer according to claim 4, wherein: the first filter comprises a first filter element with a dimension-dependent filter function, the first filter element being movable relative to the first output path to vary the filter function of the first filter, and a first servomechanism for moving the first filter element; and the second filter comprises a second filter element with a dimension- dependent filter function, the second filter element being movable relative to the second output path to vary the filter function of the second filter, and a second servomechanism for moving the second filter element.

14. A wavelength demultiplexer according to claim 4, wherein: the first filter is a tunable filter comprising a first fiber element with a first tunable grating embedded in the first fiber element; and the second filter is a tunable filter comprising a second fiber element with a second tunable grating embedded in the second fiber element.

15. A wavelength demultiplexer according to claim 4, wherein: the first filter is a tunable filter comprising a first pair of waveguide elements with a first tunable grating embedded in the first pair of waveguide elements; and the second filter is a tunable filter comprising a second pair of fiber elements with a second tunable grating in the second pair of fiber elements.

16. A wavelength demultiplexer according to claim 4, wherein: the first filter is a tunable interferometric filter comprising a first filter beam splitter, a first filter beam combiner, and a first tunable phase shifter; and the second filter is a tunable interferometric filter comprising a second filter beam splitter, a second filter beam combiner, and a second tunable phase shifter.

17. A wavelength demultiplexer according to claim 4, wherein: the first filter is a tunable filter comprising a first multi-layered filter element with incidence angle-dependent filter function, and a first mechanism for varying a first angle of incidence of a signal outputted from the first output path upon the first multi-layered filter element; and the second filter is a tunable filter comprising a second multi-layered filter element with incidence angle-dependent filter function, and a second mechanism for varying a second angle of incidence of a signal outputted from the second output path upon the second multi-layered filter element.

18. A wavelength demultiplexer according to claim 5, wherein the first and the second filters are tunable filters.

19. A wavelength demultiplexer according to claim 5, wherein the first and the second filters are tunable filters, the demultiplexer further comprises a processor coupled to the first filter, the second filter, the first shutter, and the second shutter, the processor being capable of tuning the first and the second filters and controlling the first and the second shutters.

20. A wavelength demultiplexer comprising: means for receiving a signal comprising a plurality of WDM channels and for splitting the received signal into a plurality of beams; a plurality of means for filtering, one means for filtering per beam, each means for filtering performing filtering on its respective beam; and means for amplifying the received signal.

21. A wavelength demultiplexer according to claim 20, wherein the means for amplifying is disposed at least partially inside the means for receiving.

22. A wavelength demultiplexer according to claim 21, wherein said each means for filtering is a means for band pass filtering.

23. A wavelength demultiplexer according to claim 22, wherein said each means for band pass filtering is a tunable means for band pass filtering.

24. A wavelength demultiplexer according to claim 23, further comprising a plurality of shutter means, one shutter means per beam, each shutter means capable of controllably blocking and passing its respective beam.

25. A wavelength demultiplexer according to claim 24, further comprising processor means coupled to the plurality of means for filtering and the plurality of shutter means, the processor means being for configuring the demultiplexer by tuning the plurality of means for filtering and controlling the plurality of shutter means.

26. A wavelength demultiplexer according to claim 24, further comprising an optical pump means for exciting the means for amplifying.

27. A wavelength demultiplexer according to claim 24, further comprising processor means coupled to the plurality of means for filtering, the plurality of shutter means, and the optical pump means, the processor means being for configuring the demultiplexer by tuning the plurality of means for filtering an controlling the plurality of shutter means, and for varying the amplification provided by the means for amplifying.

28. A method of demultiplexing an aggregate optical signal comprising a plurality of wavelength channels, the method comprising the following steps: feeding the aggregate signal into an input of a beam splitter, the beam splitter comprising a plurality of outputs; amplifying the aggregate signal at least partially within the beam splitter; and filtering signals at each of the outputs of the plurality of outputs.

29. A method according to claim 28, wherein the step of filtering comprises a step of band pass filtering.

30. A method according to claim 29, further comprising the steps of: determining, for said each output, whether the filtered signal of said each output needs to be blocked from exiting said each output; switching OFF a first output whose filtered signal has been determined as needing to be blocked from exiting; and switching ON a second output whose filtered signal has not been determined as needing to be blocked from exiting.

31. A method according to claim 29, further comprising the step of varying the amplification of the aggregate signal within the beam splitter.

32. A method according to claim 29, further comprising the step of: obtaining a measure of power of the aggregate signal; and setting the amplification of the aggregate signal within the beam splitter in response to the measure of power of the aggregate signal.

33. A method according to claim 29, further comprising the steps of: monitoring the aggregate signal to determine wavelengths of the plurality of wavelength channels; and assigning a first wavelength channel determined to have a first wavelength to a first output; wherein the step of band pass filtering comprises the step of filtering the first output to pass through the first wavelength.