Reduced crosstalk in photonic switches

Abstract

Embodiments of the present disclosure relate to a reduced crosstalk photonic switch, which describes a method for propagating light using an optical switch. The method includes: using one or more waveguides of the optical switch to receive a first light and a second light, the first light being at a first optical power, the second light being at a second optical power, the second optical power being lower than the first optical power due to attenuation, the optical switch having a physical shape that compensates for the attenuation; using the physical shape of the optical switch to form a compensating light; and transmitting the compensating light to one of a plurality of ports in the optical switch.

Description

[0001] Related Applications

[0002] This application is a divisional application of the invention patent application with application number 201811638109.2, application date December 29, 2018, and invention name “Reduced Crosstalk Photonic Switch”. Technical Field

[0003] The present disclosure relates generally to switches for use in switching networks, and more particularly to crosstalk-reducing optical switches. Background Art

[0004] Some communication networks, such as those used in data centers, use switching networks to route communications between devices. In some networks, the switches are Mach-Zehnder optical switches that use phase modulators and couplers, adjusting the state of the phase modulators to select which switch input and output ports are connected, and routing the light carrying the signal to the terminal device (e.g., client device, server). The Mach-Zehnder optical switch can separate the incident light into components that are routed along multiple paths. The Mach-Zehnder optical switch then uses one or more phase modulators to create constructive and destructive interference in the output couplers at the potential output paths in order to select where the light is routed. Crosstalk between different light beams passing through the switching network may be generated due to the physical properties of the couplers and phase modulators, and may increase as the light passes through successive switch nodes. In addition, current schemes for reducing crosstalk, such as low-attenuation hot switches, are power-hungry and cannot be implemented in network architectures with a large number of switches, such as modern optical switching networks. BRIEF DESCRIPTION OF THE DRAWINGS

[0005] To easily identify the discussion of any particular element or act, the most significant digit or digits in a reference number refer to the drawing ("Figure") number in which the element or act is first introduced.

[0006] Figure 1 A network architecture implementing a reduced crosstalk switch is shown according to some example embodiments.

[0007] Figure 2 A Mach-Zehnder Interferometer (MZI) reduced crosstalk switch is shown in accordance with some example embodiments.

[0008] Figure 3 Response graphs of a multiple quantum well (MQW) modulator are shown according to some example embodiments.

[0009] Figure 4A coupler that may be integrated into an optical switch to compensate for phase modulator based attenuation is shown according to some example embodiments.

[0010] Figure 5 A switch architecture with two couplers is shown according to some example embodiments.

[0011] Figure 6 A flow chart of a method for routing light through a reduced crosstalk switch is shown, according to some example embodiments.

[0012] Figure 7 A flow chart of a method for fabricating a power efficient reduced crosstalk optical switch is shown, according to some example embodiments.

[0013] Figure 8 An expansion switch that may be used to reduce crosstalk is shown according to some example embodiments. DETAILED DESCRIPTION

[0014] The following description includes systems, methods, techniques and instruction sequences that embody illustrative embodiments of the present disclosure. In the following description, for the purpose of explanation, many specific details are set forth in order to provide an understanding of the various embodiments of the subject matter of the present invention. However, it will be apparent to those skilled in the art that embodiments of the subject matter of the present invention can be practiced without these specific details. Typically, known instruction instances, protocols, structures and techniques are not necessarily shown in detail.

[0015] An optical packet switch is a reconfigurable, low-latency optical routing network component used in some data communications networks. Some switching networks can be implemented using photonic integrated circuit (PIC) technology (e.g., using Benes or Banyan architectures). An example of a switch component used in a switching network is a 2x2 Mach Zehnder interferometer (MZI), which uses a coupler to separate an incoming light beam into multiple beams, each directed to a different path. One or more phase modulators are used to create a phase difference between the beams on the different paths. A second coupler is used to interfere with the beams on the multiple paths and couple the beams to an output port. The phase difference causes constructive and destructive interference on several possible output paths, determining which path the signal is routed to. The 2x2 MZI has two input and two output ports, and uses one or more phase modulators to switch between its "cross" and "bar" transmission states.

[0016] Typically, the cardinality (number of ports) of a switching network is limited by the crosstalk, insertion loss, and power consumption of the implemented switch. Furthermore, the total crosstalk and insertion loss of a given switching network can be calculated by summing contributions from all components in the optical path through the switching network. As a component in an optical packet switch, the MZI has quantifiable crosstalk and insertion loss.

[0017] Herein, when referring to an individual switch, crosstalk is the ratio of the power of beams leaving two output ports, where the beams originate from a single input port. One of the output ports is the intended path. The switch is configured so that the port of the intended path is the construction port ("constructive port") of the output coupler, and the second output port is the destructive port ("destructive port") of the output coupler. In addition, in the context of a switching network including multiple beams along an optical path and multiple switches, crosstalk is the ratio of the power of a beam at an output port or along an optical path to the total power of another beam at the same location in the switching network due to crosstalk of an individual switch.

[0018] Crosstalk can originate from several potential sources within a given switch. Some sources include couplers with incorrect split ratios, and waveguides manufactured with insufficient tolerances that result in random changes in insertion loss or optical path length. Additional sources of crosstalk include amplitude changes caused by phase modulator elements. In silicon photonic modulators using carrier plasma modulation, changes in phase are accompanied by changes in insertion loss. Changes in insertion loss result in amplitude imbalance in the beam, which interferes with the output coupler of the MZI, which results in crosstalk. Silicon modulators based on carrier plasma dispersion effects can operate at frequencies in the GHz range and can be used in optical packet switches. Fast modulation in PICs can also be achieved by using modulators using the quantum confined Stark effect or the Franz-Keldysh effect; however, when used in switches, they suffer from the same amplitude imbalance problem. Amplitude imbalance sets a lower limit on the crosstalk that can be achieved in switch elements and further limits the size of the entire packet switch.

[0019] While some thermal phase modulators exhibit little or no amplitude imbalance, thermal phase modulators have high power consumption and slow rise / fall times (eg, microseconds), thus limiting their usefulness in packet switching.

[0020] To this end, an improved switch specifically configured to optimize the split ratio of the coupler can be implemented to offset the crosstalk of amplitude imbalance caused by absorption from the phase modulator of the switch. In some embodiments, light that has been phase-shifted from the phase modulator is input to a multimode interference component (MMI), such as a 2×2 MMI. The MMI is manufactured with tapered sides so that the power split ratio compensates for the amplitude imbalance. The configured power split ratio results in perfect or near-perfect destructive interference at one output of the 2x2 MMI. The improved switch can be integrated into a high radix ("radix") count switching network to eliminate crosstalk between different light beams traversing the network.

[0021] As used herein, the term beam is interpreted as an electromagnetic wave (e.g., light, optical signal) that can propagate along a channel (e.g., a waveguide, a fiber optic cable). A beam can be encoded to carry one or more signals, such as a data stream. A beam can be separated (e.g., split) into components. The components of a beam are themselves beams, and each of the beams may include one or more signals of the original beam. Unless otherwise specified, routing, splitting, phase shifting, or otherwise modifying is a manipulation of the beam itself, not the underlying signal or signals that a given beam may carry. For example, if a beam is separated into two component beams, each of the two component beams may include all of the signals in the original beam or a portion of the signals in the original beam.

[0022] Figure 1 An example network architecture 100 implementing a reduced crosstalk switch according to some example embodiments is shown. The network architecture 100 includes a plurality of endpoints 105A-105F devices (e.g., computers such as laptops, desktops, or servers in a data center) that send optical data back and forth through an optical switching network 110. The optical switching network 110 includes: a plurality of switches 115A-115D that route light carrying signals between the plurality of endpoints 105A-105F. For example, Figure 1 As shown, endpoint device 105C (tablet computer) and endpoint device 105E (data center) send information via switch 115C and switch 115B, as shown by the dashed arrows between the endpoints.

[0023] although Figure 1Only four switches 115A-115D are shown in FIG, but the optical switching network 110 may include multiple switches to increase the overall bandwidth of the network 110. In addition, each switch may include multiple sub-assembly switches as part of an expansion switch, as discussed in further detail below. Multiple switches 115A-115D can simultaneously route multiple light-carrying signals between multiple endpoints 105A-105F. As discussed, one possible source of crosstalk is amplitude imbalance within the switch, resulting in incomplete destructive interference, causing noise on the "victim" signal due to unintentional crosstalk from the "interferer" signal. For example, a 2×2 MZI switch operating in a "strip" state may have a small portion of light passing through a "cross" state. In some example embodiments, one or more of the multiple switches 115A-115D have a phase modulator and an MMI coupler that is configured to offset the amplitude imbalance caused by the phase modulator. Eliminating amplitude imbalance in a power-efficient manner reduces crosstalk in the optical switching network 110, thereby allowing the optical switching network 110 to integrate more switches and route a larger amount of information between a greater number of endpoints.

[0024] Figure 2 An exemplary crosstalk reduction switch 200 is shown in accordance with some example embodiments. The switch 200 is configured as a 1x2 Mach-Zehnder interferometer (MZI) switch using symmetric couplers with three ports: A, B, and C. Although the switch 200 can operate in either direction (i.e., light is input to the 1x2 coupler 205 and output at the 2x2 coupler 220, or vice versa), the following discussion assumes that light traverses the switch 200 from left to right (i.e., light is input to port A and outputs from ports B and / or C).

[0025] like Figure 2 As shown, 1×2 coupler 205 has one input port (port A) and two output ports (unlabeled ports connected to phase modulators 210 and 215). In some example embodiments, 1×2 coupler 205 is a 1×2 MMI or waveguide Y-junction. Each of the output ports of 1×2 coupler 205 outputs to a corresponding phase modulator, phase modulator 210 and phase modulator 215. Phase modulators 210 and 215 output to 2×2 coupler 220, which has two input ports and two output ports. In some example embodiments, 2×2 coupler 220 is a 2×2 MMI. The dashed lines connecting the corresponding components of the architecture of switch 200 are channels for optical data, such as waveguides, optical fibers.

[0026] refer to Figure 2 A “phase arm” refers to the components that make up the path between optocouplers 205 and 220 along the top or bottom of switch 200 .

[0027] The phase modulators 210 and 215 may have a phase shift that is adjusted to determine whether all of the light entering (or leaving) port A leaves (or enters) via port B, whether a small portion of the light leaves (or enters) ports B and C, or whether all of the light from port A leaves (or enters) via port C. In some example embodiments, a light beam is input to port A of the 1×2 coupler 205 and is symmetrically split into the top and bottom outputs of the 1×2 coupler 205. Each of the output beams of the 1×2 coupler 205 has exactly the same phase and 50% of the original power (e.g., 50% of the power of the light input to port A). The phase modulators 210 and 215 may phase shift the light so that when the light is coupled in the 2×2 coupler 220, the phase difference results in destructive interference at one output port of the 2×2 coupler 220 and constructive interference at the other output port.

[0028] In some example embodiments, the switch 200 operates in two states, where in each state only one of the phase modulators is turned on. In the first state, the phase modulator 215 is turned on and has a phase shift of π / 2 applied to the light, and the phase modulator 210 is turned off and has a zero phase shift. In the first state, all light input to the switch 200 leaves the top output B of the switch 200 because the two beams interfere constructively at port B and destructively at port C. In the second state, the phase modulator 215 is turned off and has a zero phase shift, while the phase modulator 210 is turned on and has a π / 2 phase shift. In the second state, all light input to the switch 200 leaves port C because the signals interfere constructively at port C and destructively at port B.

[0029] If the amplitudes of the components of light entering the two inputs of the 2×2 coupler 220 are the same, then for the correct phase shift, perfect destructive interference occurs in port B or C, for which no light is intended to exit. However, if the amplitudes of the two components of light entering the two inputs of the 2×2 coupler 220 are not the same, then perfect destructive interference does not occur, resulting in crosstalk. In addition, some phase modulators with desirable characteristics (e.g., power efficiency, GHz response time) have absorption of the phase shifted light that varies with the amount of phase shift, resulting in amplitude imbalance and crosstalk in the 2×2 coupler 220.

[0030] Figure 3 An example response graph 300 of a multiple quantum well (MQW) modulator is shown according to some example embodiments. A multiple quantum well (MQW) modulator modulates the phase and amplitude of light by changing the complex refractive index of the modulator by changing the applied electric field. As shown, as the MQW modulator gradually shifts the phase of the light, the amplitude of the light gradually decays. Return Figure 2, so if the phase modulator 210 is an MQW modulator and is turned on with a π / 2 phase shift, the amplitude of the phase-shifted light will be lower than the amplitude of the non-phase-shifted light passing through the phase modulator 215.

[0031] In some example embodiments, a transfer matrix may be used to model light passing through the switch 200. The column vector is shown in Expression 1:

[0032]

[0033] Expression 1 represents the amplitude and relative phase of light in a waveguide mode at a cross section along the switch 200 or at the input or output of a coupler. In some example embodiments, the number of elements of the column vector is the same as the number of waveguides at the cross section, and it is assumed that each waveguide carries light only in the fundamental mode.

[0034] The 1x2 coupler 205 can be modeled as a transfer matrix that splits an input mode (e.g., light input to the 1x2 coupler 205) into two modes with the same phase, as would occur in a 1x2 MMI or waveguide Y-junction. In some example embodiments, the input to the 1x2 coupler 205 is as follows:

[0035]

[0036] If you operate on an input field distribution to create an output field distribution, then the output fields (e.g. two modes in two waveguides) are a function of the input field (e.g. one mode in one waveguide). That is:

[0037]

[0038] As will be appreciated by one of ordinary skill in the art, Kramers-Kronig analysis can yield expressions for absorption and phase change as a function of carrier density in silicon. In this application, operating at 1310 nm: (1) dn = -7.7625e-023*N^1.05 - 4.7863e-018*P^.805; (2) dabs (1 / cm) = 1.7925E-20*N.^1.14 + 5.9858E-20*P.^1.1. N and P represent the number of nanometers per cm 3 The electron and hole densities.

[0039] The phase modulator arm can be set to use two operating states with phase shifts of (0, π / 2) and (π / 2, 0). The field transmission coefficient of the modulator with a phase shift of π / 2 relative to the field transmission coefficient of the modulator with zero phase shift will be given by t. For both operating states, the transfer matrix of the phase modulator arm will be:

[0040]

[0041] In expression 4:

[0042]

[0043] If an MMI implementation with a 50% split ratio is used, the transfer matrix of the output 2x2 coupler 220 is as follows:

[0044]

[0045] The transfer matrix of the switch in the two operating states for light entering the left side (into the 1×2 coupler 205) is as follows:

[0046]

[0047]

[0048] Expression 7 can be simplified to:

[0049]

[0050] If there is no loss, then t = 1 and the expression simplifies to:

[0051] In the case of loss, the crosstalk (in dB) in the two switching states is as follows:

[0052]

[0053] As described for light entering the switch 200 through the side with a single port (i.e., entering port A), the switch will be configured so that one of the two output ports (e.g., port B, port C) is the intended output port and the constructed port, and any light coming out of the destructive port will cause crosstalk. If the switch 200 is operated in the opposite direction, with two different light beams entering two ports on the same side (i.e., ports B and C), only one of the two signals will be intentionally passed to the single port on the other side (i.e., port A). Any portion of the other light beam that exits the single port on the opposite side will cause crosstalk.

[0054] In some example embodiments, the phase modulators (e.g., phase modulators 210 and 215) are based on the carrier plasma dispersion effect and can modulate light by changing the complex refractive index of the modulator based on the amount of charge carriers (e.g., N and P) injected into the semiconductor material of the modulator. According to some example embodiments, if the phase modulators 210 and 215 are modulators based on carrier-plasma dispersion, when operating at 1310nm, and if the phase modulators 210 and 215 are each 100 microns long, carriers are injected into the phase modulators 210, 215 so that the N and P concentrations are ~1.7E18 / cm^3, which produces a shift of π / 2 in one arm while the other arm is in the off state. The resulting field transmission t is then 0.916, with an insertion loss of 0.77dB in the modulated phase arm. In addition, the crosstalk in this configuration is 27.1dB.

[0055] If the power split ratio of the 2×2 coupler 220 is specially configured, the crosstalk of the architecture 100 of the switch 200 can be eliminated. For a rectangular 2×2 coupler, also known as a multi-mode interferer (MMI), several power split ratios can be obtained, as understood by those of ordinary skill in the art:

[0056]

[0057] Where P c is the power output of one output port of the 2x2 coupler 220, and P b is the power output from the other port of the 2x2 coupler 220 .

[0058] According to some example embodiments, while some phase modulators (such as MQW modulators or phase modulators based on carrier-plasma dispersion effects) are power efficient, they can cause attenuation in the phase-shifted light, resulting in amplitude imbalances between channels entering the coupler, where the light in the channels is subject to interference, and subsequent crosstalk at the output of the coupler. To make matters worse, modern optical switching networks require a large number of consecutive switch stages, thereby increasing the likelihood of crosstalk. Crosstalk in such networks cannot be simply fixed by implementing non-attenuating phase modulators (such as thermal-based phase modulators). This is because low-attenuation phase modulators require considerable power to operate, and implementing those phase modulators in high radix count networks will result in power-hungry, commercially impractical architectures.

[0059] Figure 4An example coupler 400 is shown that can be integrated into an optical switch to compensate for phase modulator-based attenuation according to some example embodiments. Coupler 400 is a 2x2 optical coupler having a length L and a width W0, and two inputs and outputs, including input A, output A, input B, and output B. Typically, coherent light can be input into input A and / or input B, mixed or interfered, and emitted from output A and / or output B depending on the phase difference and amplitude of the inputs. For example, if the light entering inputs A and B has the same amplitude but a phase difference of π / 2, constructive interference may occur for output A, which indicates a high signal, while destructive interference may occur for output B, indicating a low signal. By controlling the relative phase of the light at the two inputs of the coupler, the input light can be routed to one output or the other.

[0060] As shown, coupler 400 has a tapered midsection (“tapered midsection”) whose sides are offset (e.g., indented) from the sides of the rectangular coupler by a distance |dW|. In addition, coupler 400 has a gap 405 having a width W1. The tapered sides and gap 405 of coupler 400 can be configured to change the power distribution ratio of coupler 400 due to the difference in the paths of the light superimposed in coupler 400. Specifically, the normalized power “P c " can be:

[0061] P c =cos(0,5·π·dΩ) 2 Expression 11

[0062] Where dΩ is the normalized width change. The other output “P b The normalized power of ” is: P b =1-P c Furthermore, the normalized width change dΩ depends on the width W0 and the taper dW of the input face of the coupler 400:

[0063]

[0064] In some example embodiments, the split ratio may be configured to compensate for the loss of the phase arm that may be connected to the coupler. The cross power split ratio S of the 2×2 coupler 220 may be configured as given in Expression 13, where the field transmission t refers to the ratio of the field transmission coefficient of the phase modulator with a π / 2 phase shift to the field transmission coefficient of the phase modulator with a zero phase shift:

[0065]

[0066] Expression 13 results in the transfer matrix of the 2×2 coupler 220 being as follows:

[0067]

[0068] In some example embodiments, in a first switching state, phase modulator 210 has a phase shift of π / 2, and phase modulator 215 has a zero phase shift (e.g., an off state). In a second switching state, phase modulator 210 is in an off state, and phase modulator 215 has a phase shift of π / 2. The transfer matrix of switch 200 in both switching states simplifies to these expressions, where all transmitted light exits only from one port, so there will be no crosstalk

[0069]

[0070] If operating in the reverse direction, for example as a 2×1 coupler, the transfer matrix in the two switching states is:

[0071]

[0072]

[0073] Expression 16 simplifies to show that there is no crosstalk when operating as a 2x1 coupler:

[0074]

[0075] In some example embodiments, the switch 200 may have only one phase modulator. That is, two beam components are output from the 1×2 coupler 205, and one of the beam components is input to the only phase modulator (e.g., phase modulator 210) in the switch 200, and the beam is coupled in the 2×2 coupler 220 to compensate for the imbalance caused by the phase modulator. In some embodiments having only one phase modulator, the beam components output from the 1×2 coupler 205 will have an uneven distribution ratio.

[0076] Figure 5 A switch 500 with two 2×2 couplers 505, 520 is shown according to some example embodiments. In the switch 500, two beams are input into the 2×2 coupler 505 (e.g., one beam input port A, the other beam input port B). The 2x2 coupler 505 outputs to two phase modulators 510 and 515. The two phase modulators 510 and 515 output to another 2×2 coupler 520. Figure 5 In , a “phase arm” refers to the components between 2x2 couplers 505 and 520 that make up the path along the top or bottom of switch 500 .

[0077] In switch 500, if both the bottom and top phase arms have the same length and both phase arms are in the off state (e.g., phase modulators 510 and 515 are off), perfect cross coupling exists if the sum of the split ratios of the two couplers is 100%. However, when switch 500 is manufactured, manufacturing defects may occur that cause amplitude imbalance and crosstalk. In some embodiments, one or more of the couplers of switch 500 may be tuned by changing the coupling ratio to compensate for manufacturing defects. In some example embodiments, a coupling ratio having Figure 4 Such tuning is achieved by using a coupler in the middle of the tapered portion shown in . Specifically, for example, the 2×2 coupler 520 can be tuned to create a cross power split ratio given by the following expression, where t is the ratio of the field transmission coefficient from the phase modulator with a π phase shift to the field transmission coefficient of the phase modulator with a zero phase shift:

[0078]

[0079] A coupler with a cross power split ratio of S&1-S can exhibit the following transfer matrix:

[0080]

[0081] In the case where couplers 505 and 520 are placed in series, the matrix The transfer matrix of the switch 500 obtained by separating the lossless pairs of equal-length phase arms represented by is the transfer matrix of the crossover state without crosstalk, which is independent of the field transmission value t, where t is assumed to be a non-negative real number.

[0082] If one of the phase arms is modulated so that the phase shift between the arms is separated by 180° (e.g. (0,π)), and if the ratio of the field transmission of the modulated arm at 180° phase shift to the field transmission of the modulated arm at 0° phase shift is t, the transfer matrix for the phase section (“phase section”) will be: Furthermore, the transmission of switch 500 is:

[0083]

[0084] Expression 20 is simplified to:

[0085]

[0086] If the input mode is and The output mode is and For the first mode There is no crosstalk. If the other arm in the phase section is modulated to have a 180° phase shift of the field transfer t, given by the transfer matrix indicates that the second mode can be avoided In this case, the transmission of switch 500 is:

[0087]

[0088] If the switch is operated in the reverse direction (eg, light passes from 2x2 coupler 520, through phase modulators 510 and 515, and out of 2x2 coupler 505), the first state of the phase portion is represented by And the transmission of switch 500 is as follows:

[0089]

[0090] In the reverse embodiment, in mode Light entering switch 500 in the opposite direction will not have crosstalk.

[0091] According to some example embodiments, switch 500 can operate in three different states: a passive crossbar state, a first bar state, and a second bar state. In some example embodiments, configuring switch 500 to operate in one of the three states can be useful in a network architecture where not all possible paths carry signals at the same time. In the following discussion, A→D can be interpreted as a beam from port A through port D across the entire switch 500 (e.g., through the superposition in couplers 505 and 520, etc.).

[0092] In the passive crossover state, light can pass from A→D or from D→A. Similarly, light can pass from B→C and C→B. In the crossover state, both phase modulators 510 and 515 are in the off state, and crosstalk does not occur.

[0093] In the first stripe state, the phase modulator 510 operates with a 180° phase shift, the phase modulator 515 is turned off, and the field transmission is the t: stripe state. In the first stripe state, the light passing through C→A and B→D has no crosstalk. On the other hand, the light passing through A→C and D→B exhibits crosstalk due to imbalance.

[0094] In the second bar state, the phase modulator 515 operates with a 180° phase shift, the phase modulator 510 is off, and the field transmission is the t: bar state. In the second bar state, the light passing through A→C and D→B has no crosstalk. On the other hand, the light passing through C→A and B→D exhibits crosstalk due to the imbalance.

[0095] In the first and second states, the crosstalk in dB is:

[0096]

[0097] This is an improvement over the scheme for implementing a 50% coupler (i.e., instead of a split ratio of S & 1-S), where the crosstalk is higher for all input states, regardless of which arm is modulated:

[0098]

[0099] In some example embodiments, phase modulators 510 and 515 are carrier injection based modulators with a length of 150 microns. In these embodiments, to achieve a 180 degree phase shift, carriers are injected such that the N and P concentrations are ~2.3E18 / cm^3, which results in a field transmission field t=0.830 and an insertion loss of 1.61 dB in the modulated phase arm. If couplers 505 and 520 are 50% couplers, the crosstalk in the strip states of the two inputs is 20.7 dB.

[0100] In some example embodiments, couplers 505 and 520 use power split ratios of 1 / (1+t) and t / (1+t), respectively. In those embodiments, one or both of phase modulators 510, 515 may be tapered to compensate for imbalance, as described above. In those embodiments, there will be optimal zero crosstalk in one direction, while 14.6 dB of crosstalk may occur in the other direction.

[0101] In some example embodiments, the 2x2 switch 500 is configured as a 1x2 switch by removing one of the ports and one of the phase modulators. Control of the modified 2x2 switch architecture is simplified compared to the switch architecture 200 (1x2 switch embodiment) because there is one less phase modulator to manage. In these example embodiments, a light beam that traverses a path where the phase modulator is removed can pass from an initial coupler (e.g., coupler 505) to another coupler (e.g., coupler 520) via a passive optical waveguide without passing through the phase modulator.

[0102] Figure 6 6 is a flowchart of a method 600 for routing light through a reduced crosstalk switch according to some example embodiments. In some example embodiments, a single switch (e.g., switch 200, switch 500) implements method 600, while in some example embodiments, a switching network (e.g., discussed below) implements method 600. Figure 8 The expansion switch 800) implements the method 600.

[0103] At operation 605, the coupler receives light, such as one or more light beams. Figure 5 , the coupler 505 receives the first beam at port A and receives the second beam at port B. At operation 610, the coupler 605 routes one or more received beams to a plurality of phase arms. Figure 5, the 2x2 coupler outputs light to the top phase arm and the bottom phase arm of switch 500, as described above.

[0104] At operation 615, the light on the phase arm is phase shifted using one or more phase modulators. For example, the phase modulator 510 is a carrier injection based phase shifter that is configured to phase shift the light on the top phase arm by π / 2, as described above. In addition, in some example embodiments, although the phase modulator 510 is configured to impart a phase shift, the phase modulator 515 is in an off state and the light on the bottom phase arm maintains its initial phase. At operation 220, the light on the top and bottom phase arms are combined in a second coupler that is configured to compensate for attenuation caused by the modulation. Continuing with the example, if the phase modulator 510 is a carrier injection based phase modulator, the modulated light on the top phase arm may be attenuated relative to the light on the bottom phase arm. To compensate for the attenuation, the second coupler has a distribution ratio (e.g., t 2 / (1+t 2 )), which compensates for the attenuation of the light on the top phase arm. At operation 625, the second coupler sends a low crosstalk output at one or more output ports. For example, if the second coupler has two output ports, due to the attenuation compensation, complete destructive interference may occur for one output port, and complete constructive interference may occur for one output port.

[0105] Figure 7 A flow chart of a method 700 for manufacturing a power efficient reduced crosstalk optical switch according to some example embodiments is shown. At operation 705, a phase modulator phase shifts light. At operation 710, the modulated light is analyzed to determine an amount of attenuation caused by the phase modulator. At operation 715, an initial split ratio of the coupler is selected. For example, if the coupler is a rectangular coupler, P is selected from possible split ratios. c / P b =50 / 50 distribution ratio (see Expression 10 above). At operation 720, the coupler is shaped to minimize crosstalk. For example, if the coupler is a rectangular coupler, the middle portion of the coupler can gradually reduce the distance |dW| until the initial distribution ratio becomes the distribution ratio (e.g., t 2 / (1+t 2 )). In some example embodiments, which allocation ratio is selected and gradually modified depends on different design considerations, including, for example: beam intensity, cardinality count, phase modulator type, attenuation amount.

[0106] Figure 8An exemplary expansion switch 800 that can be used to reduce crosstalk according to some example embodiments is shown. As shown, the expansion switch 800 includes a plurality of switch subassemblies 805A-805D, each of which is a separate switch, such as switch 200 or 500. The expansion switch 800 can reduce or eliminate crosstalk by ensuring that only one light beam carrying a signal passes through any one of the sub-switch components at a time. To this end, as the radix count increases, additional sub-assembly switches can be added to expand the entire switch architecture and reduce or eliminate the possibility of crosstalk. In some example embodiments, in order to maintain efficient power consumption as the radix count and sub-switch count increase, one or more of the sub-switches implement a carrier injection-based phase modulator and shaping coupler as described above. In this way, the optical switching network 110 can increase its port count and subsequent bandwidth while being power efficient and maintaining low crosstalk.

Claims

1. A method for propagating light using an optical switch, the method comprising: receiving, using one or more waveguides of the optical switch, a first light and a second light, the first light being at a first optical power and the second light being at a second optical power, the second optical power being lower than the first optical power due to attenuation, the optical switch having a physical shape that compensates for the attenuation; forming compensation light using the physical shape of the optical switch; and The compensating light is transmitted to one of a plurality of ports in the optical switch.

2. The method according to claim 1, further comprising: Input light is received using one or more input ports of the optical switch.

3. The method according to claim 2, further comprising: The input light is separated into the first light and the second light. The method according to claim 3 , wherein the input light is separated into the first light and the second light by a coupler.

5. The method of claim 1, wherein the optical switch comprises a coupler having a physical shape that compensates for attenuation differences of light in the coupler.

6. The method of claim 5, wherein the coupler has tapered sides that modify superposition of light in the coupler.

7. The method of claim 6, wherein the superposition of the lights in the couplers is modified by the tapered sides so that a power splitting ratio of the couplers compensates for differences in attenuation of the lights in the couplers.

8. The method of claim 7, wherein the power allocation ratio is t^2 / (1+t^2), where t is the ratio of field transmission coefficients of optical components in the optical switch.

9. The method of claim 7, wherein the power allocation ratio is t / (1+t), where t is a ratio of field transmission coefficients of optical components in the optical switch.

10. The method of claim 1, wherein the attenuation is due to one or more optical components in the optical switch. The method of claim 10 , wherein the one or more optical components include one or more optical modulators.

12. The method according to claim 11, further comprising: The second light is modulated using the one or more light modulators.

13. The method of claim 11, wherein the one or more light modulators include a phase modulator that modulates the second light using phase modulation.

14. The method of claim 13, wherein modulating the second light comprises shifting a phase of the second light in the optical switch by approximately 90 degrees using the phase modulator.

15. The method of claim 13, wherein modulating the second light comprises shifting a phase of the second light in the optical switch by approximately 180 degrees using the phase modulator.

16. The method of claim 13, wherein the one or more optical modulators modulate the second light based on a number of positive and negative carriers embedded in the one or more optical modulators.

17. The method of claim 13, wherein the one or more light modulators include a first modulator and a second modulator, and wherein the second modulator modulates the second light when the first modulator is inactive. The method of claim 17 , wherein an inactive first modulator passes the first light unmodulated.

19. The method of claim 1, wherein the compensating light is transmitted to one of the plurality of ports of an optical network.

20. An optical switch, comprising: One or more waveguides for receiving a first light and a second light, the first light being at a first optical power, the second light being at a second optical power, the second optical power being lower than the first optical power due to attenuation, the optical switch having a physical shape that compensates for the attenuation; as well as One or more ports are used to output the compensation light formed using the physical shape of the optical switch.

Images