Control circuit and optical circuit control method

Abstract

A control unit (41) controls, by using bias voltage (Vb), the optical path difference between two optical paths (3, 4) included in one or more Mach‐Zehnder interferometers (MZI) included in an optical circuit (100). A reference light generation unit (35) generates reference light (λ1, λ2-λN) having N types of wavelengths different from the wavelength of signal light (λ). A reference light input unit (8) performs wavelength multiplexing of the N types of reference light (λ1, λ2-λN) on the signal light (λ) and inputs the resulting light into the optical circuit (100). An optical power monitor (53) monitors the light intensities of the N types of reference light (λ1, λ2-λN) propagated through the Mach‐Zehnder interferometers (MZI) in the optical circuit (100). A normalization unit normalizes monitoring results on the light intensities of the N types of reference light (λ1, λ2-λN). Specifically, the normalization unit performs the normalization so that: when all the optical path differences are optimal, the norm of reference light vectors having the N types of normalized monitoring results as the component thereof becomes zero; and as at least one optical path difference deviates from the optimal value, the norm becomes greater. The control unit (41) changes the bias voltage in such a manner that the norm of the reference light vectors approaches zero.

Description

Control circuit and optical circuit control method

[0001] The present invention relates to a control circuit and an optical circuit control method.

[0002] Optical circuits consisting of Mach-Zehnder interferometers (MZIs) or composite MZIs are widely used as periodic optical filters and optical modulators. In particular, nested MZIs are widely used in IQ optical modulators for generating optical QAM (Quadrature Amplitude Modulation) signals.

[0003] 13 is a diagram showing the configuration of an MZI 91. The MZI 91 is an MZI with the most basic configuration. An optical input port 1 inputs a CW (Continuous Wave) signal light. An optical branching circuit 2 branches the signal light input to the optical input port 1 into two. In many cases, this branching ratio is 1:1. However, it is not necessarily limited to this ratio.

[0004] One of the two beams of light branched by the optical branching circuit 2 propagates through the first optical waveguide 3, and the other beam propagates through the second optical waveguide 4. The optical multiplexing circuit 6 multiplexes the light that has propagated through the first optical waveguide 3 and the light that has propagated through the second optical waveguide 4. The optical multiplexing circuit 6 outputs the multiplexed light to an optical output port 7. The optical path length of the second optical waveguide 4 is determined by the bias voltage V applied by the bias application electrode 5. ｂ For this fine adjustment, the Pockels effect or the thermal expansion of the optical waveguide caused by a heater can be used.

[0005] The difference between the optical path length of the first optical waveguide 3 and the optical path length of the second optical waveguide 4 is defined as the optical path difference ΔL. The intensity of the output light output from the optical output port 7 varies depending on the wavelength λ of the signal light and the optical path difference ΔL. However, the wavelength λ is affected by the refractive index of the optical waveguide. In addition, the optical path length is affected by the circuit configurations of the optical branching circuit 2 and the optical multiplexing circuit 6.

[0006] For simplicity, in this application, the wavelength λ is not the optical wavelength in a vacuum, but the wavelength at the time of propagation inside the optical waveguide. Also, in this application, the optical path length is a value that includes optical phase fluctuations inside the optical branching circuit 2 and the optical multiplexing circuit 6. For example, if the optical phase is delayed by π / 2 (unit: radian, same below) inside the optical multiplexing circuit 6 in the process of light propagating from the second optical waveguide 4 to the optical output port 7, this phase delay is considered to be an increase in the optical path length equivalent to λ / 4. This is because the wavelength λ corresponds to an optical phase of 2π.

[0007] 14A and 14B are diagrams showing the relationship between the optical output intensity P(λ), which is the intensity of the output light from the MZI 91 shown in FIG. 13, and the optical path difference ΔL. Fig. 14A shows the relationship before the occurrence of bias drift, and Fig. 14B shows the relationship after the occurrence of bias drift. Bias drift will be described later.

[0008] As is well known, the optical output intensity P(λ) from the MZI exhibits a sinusoidal response to the optical path difference ΔL. The maximum output state, the minimum output state, and the intermediate state between them are often called the Peak point, the Null point, and the Quad point, respectively. Therefore, these notations are also used in this application. When the optical path difference ΔL is an even multiple (including 0) of λ / 2, it becomes a Peak point, and when the optical path difference ΔL is an odd multiple of λ / 2, it becomes a Null point. It should be noted here that, although the Null point, Quad point, and Peak point generally depend on the wavelength, the Peak point at ΔL = 0 is exceptionally independent of the wavelength.

[0009] Consider using the MZI 91 as an optical filter to suppress light of wavelength λ. In this case, it is sufficient to select one of the null points for wavelength λ. For example, when the bias voltage V applied by the bias application electrode 5 is ｂ The optical path difference ΔL can be set to 0.5λ by adjusting the above. In FIG. 14(a), this state is indicated by a circle with a downward arrow.

[0010] However, the bias voltage V ｂEven if the optical path difference ΔL is always kept constant, the optical output intensity may change over time due to temperature changes outside the MZI 91 or stresses applied thereto. For example, even if the bias is set at the null point and the optical path difference ΔL is 0.5λ as shown in FIG. 14(a), over time, the optical path difference ΔL may become a value different from 0.5λ, as shown in FIG. 14(b). This is called bias drift of the MZI. When the MZI is used as an optical filter as shown in FIG. 13, bias drift causes deterioration of the extinction ratio.

[0011] It is also common to use a single MZI as an optical modulator by applying an external modulation signal to the MZI to modulate the optical path difference ΔL and modulate the output light. In this case, if the output light is biased so that it is at the Quad point when the modulation signal is zero, an NRZ (Non Return to Zero) signal can be obtained. Alternatively, if the output light is biased so that it is at the Null point when the modulation signal is zero, a CS-RZ (Carrier-Suppressed Return-to-Zero) signal can be obtained. In such use cases, bias drift causes deterioration of the optical signal waveform and degrades signal quality.

[0012] It is also common to generate optical QAM signals using an IQ optical modulator configured with multiple nested MZIs rather than a single MZI. Figure 15 is a diagram showing the configuration of a typical IQ optical modulator 92. The IQ optical modulator 92 has a configuration in which an in-phase (in-phase) MZI 50 is installed in the first optical waveguide 3 of the MZI 91 shown in Figure 13, and a quadrature (quadrature phase) MZI 51 is installed in the second optical waveguide 4.

[0013] Inside the In-Phase MZI 50, the I-side optical branching circuit 20 branches the light transmitted through the first optical waveguide 3 into two and outputs one branched light to each of the two optical paths, and the I-side optical multiplexing circuit 21 multiplexes these two branched lights. The optical path difference between the two optical paths inside the In-Phase MZI 50 is modulated in a push-pull manner by a modulation signal I applied via an I-side modulation electrode 22. Here, the In-Phase MZI 50 is modulated by a bias voltage V applied via an I-side bias application electrode 23 so that it becomes a null point at the moment when the modulation signal I is zero. ｂ＿Ｉ is biased by

[0014] Inside the quadrature MZI 51, the Q-side optical branching circuit 24 branches the light transmitted through the second optical waveguide 4 into two and outputs one branched light to each of the two optical paths, and the Q-side optical multiplexing circuit 25 multiplexes these two branched lights. The optical path difference between the two optical paths inside the quadrature MZI 51 is modulated in a push-pull manner by a modulation signal Q applied via a Q-side modulation electrode 26. Here, the quadrature MZI 51 is modulated by a bias voltage V applied via a Q-side bias application electrode 27 so that it becomes a null point at the moment when the modulation signal Q is zero. ｂ＿Ｑ is biased by

[0015] The optical multiplexing circuit 6 multiplexes the modulated light output from the In-Phase MZI 50 and the modulated light output from the Quadrature MZI 51. The optical path difference between the two is determined by the bias voltage V applied via the bias application electrode 5. ｂ＿Ｐｈ The bias voltage is set so that the optical phase difference between the optical electric fields E_I and E_Q is ±π / 4+π×m ｐｈ where m ｐｈ is an arbitrary integer. In other words, the largest MZI consisting of the first optical waveguide 3 and the second optical waveguide 4, which is the parent MZI in the nested MZI structure, has a bias voltage V ｂ As a result, the modulated light output from the optical multiplexing circuit 6 becomes a QAM signal and is output from the optical output port 7.

[0016] In an IQ optical modulator, bias drift of the three types of bias causes distortion of the constellation, i.e., the constellation, which should have a square outline, is distorted into a rectangular or diamond shape, thereby degrading the signal quality.

[0017] For the reasons mentioned above, in optical circuits using MZIs, it is important to have a bias condition monitor that monitors the presence or absence of bias drift, and an automatic bias control that quickly corrects it if it is determined that bias drift has occurred. In particular, in optical circuits that control multiple MZIs and multiple bias voltages as shown in Figure 15, it is extremely important to correctly determine which bias is drifting and whether the bias voltage should be increased or decreased to correct the drift.

[0018] In the case of an optical modulator or optical filter consisting of a single MZI, the bias voltage V ｂ A technique for performing synchronous detection by slightly changing the bias voltage V ｂ Dithering is performed by superimposing a pilot tone with a small amplitude of frequency f on the optical output power P(λ), and minute changes in the optical output power P(λ) are synchronously detected at frequency f. Based on the magnitude of the synchronous detection result and its sign, the bias voltage V ｂ It is possible to determine whether to increase or decrease the correction amount.

[0019] The bias condition monitor and bias drift correction in an IQ optical modulator are much more complicated and difficult than those in an optical modulator consisting of a single MZI, but technical solutions to this problem have already been proposed. One of them is a technique called asymmetric bias dithering (see, for example, Non-Patent Document 1). Asymmetric bias dithering is a technique in which the bias voltage V ｂ＿Ｉ and bias voltage V ｂ＿Ｑ Dithering is performed by superimposing a pilot tone with a small amplitude and in quadrature phase on each of the bias voltages V ｂ＿Ｉ , bias voltage V ｂ＿ＱNot only that, but also the bias voltage V ｂ＿Ｐｈ It is possible to monitor the conditions of all biases, including the bias voltage. The monitoring results include the amount and sign of drift of each bias (whether it has increased or decreased from the optimal value). By feeding back this monitoring result to each bias voltage, it is possible to maintain each bias voltage at an optimum level.

[0020] When an optical modulator is installed in a transmitter and used for commercial services, the bias control of the optical modulator roughly plays two roles. One is to quickly optimize each bias during the transmitter startup sequence. The other is to monitor the presence or absence of bias drift during transmission service operation after the startup sequence is completed, and to perform feedback control to immediately correct any drift that occurs. The purpose of these two controls is essentially the same, but there are differences in the requirements, as described below.

[0021] Immediately after the start of the startup sequence, especially when the wavelength has been newly set, the bias voltages are far from their optimum values. However, since transmission service has not yet started during the startup sequence, trial and error is permitted, such as trying to significantly increase or decrease the bias voltages in order to bring them closer to their optimum values. Furthermore, there is no problem if the amplitude of the pilot tone used for dithering is increased to a certain extent.

[0022] On the other hand, during transmission service operation, it is necessary to maintain the highest quality of the transmission signal at all times. Trial control such as significantly increasing or decreasing the bias voltage cannot be implemented because it would suddenly change the waveform or constellation of the transmission signal. Furthermore, since the pilot tone used for dithering is also considered noise from the perspective of the transmission signal, its amplitude must be kept as small as possible.

[0023] As the number of signal levels in an optical QAM signal increases, the density of each symbol increases, and the penalty imposed by the pilot tone on the transmission signal also increases. However, if the amplitude of the pilot tone is made too small, a trade-off occurs: the sensitivity of the bias condition monitor decreases.

[0024] To solve this problem, we have proposed and reported a ditherless (without pilot tone) bias control technology that enables both bias condition monitoring and feedback control when drift occurs during transmission service operation. This technology utilizes the wavelength dependence of the null point and quad point of the MZI. Details are described in Patent Document 1 and Non-Patent Document 2, but only the points necessary to explain the technical content of this application will be explained here.

[0025] In the MZI 91 shown in FIG. 13, a signal light having a single wavelength λ is input to the optical input port 1. Here, the wavelength λ １ and a first reference light having a wavelength λ ２ Consider a case where a second reference light having a wavelength λ is wavelength-multiplexed with a signal light and these are input to the optical input port 1. １ and wavelength λ ２ is different from the wavelength λ. In this case, the optical output intensity shown in FIG. 14(a) is rewritten as shown in FIG. 16. Here, λ ２ >λ>λ １ The solid line represents the optical output intensity P(λ) of the wavelength λ, and the two dashed lines represent the optical output intensity P(λ) of the first reference light. １ ) and the optical output intensity P(λ) of the second reference beam ２ ) is shown.

[0026] Since the null point and the quad point have wavelength dependency, even if the MZI 91 is biased to the null point at the wavelength λ of the signal light, it is not a null point at the wavelength λ of the reference light. １ ) and P(λ ２ ) does not have the same power.

[0027] Next, consider the case where bias drift occurs and the optical path difference ΔL of the MZI 91 increases slightly from 0.5λ. As shown in FIG. １ ) increases rapidly, and P(λ ２ ) decreases slightly.

[0028] Next, consider the case where a reverse bias drift occurs and the optical path difference ΔL of the MZI 91 is slightly reduced to less than 0.5λ. １ ) decreases, and P(λ ２ ) will increase sharply.

[0029] It should be noted that in both Figures 17 and 18, P(λ) indicated by the black circles shows a slight increase, making it difficult to determine whether the bias drift is positive or negative. However, by introducing a reference beam, it becomes possible to determine the sign.

[0030] If it is possible to learn during the startup sequence how the intensity of the reference light output from the MZI changes when positive or negative bias drift occurs and record this in a data table, it will be possible to estimate the presence or absence of bias drift and the direction of the drift when it occurs in a ditherless manner (without a pilot tone) by monitoring the change in the intensity of the reference light during transmission service operation and comparing it with the data table.

[0031] International Publication No. 2023 / 105930

[0032] Hiroto Kawakami, Takayuki Kobayashi, Mitsuteru Yoshida, Tomoyoshi Kataoka and Yutaka Miyamoto, "Auto bias control and bias hold circuit for IQmodulator in flexible optical QAM transmitter with Nyquist filtering," Optics Express, Vol.22, No,23, pp. 28163-28168, Nov. 2014.Hiroto Kawakami, Yoshiaki Kisaka and Etsushi Yamazaki, "Dither-free Auto Bias Control Technique for In-service Optical IQ modulator Using Reference Pulsed Light," Asia Pacific Conference on Communications (APCC2023) SP2 adopted, Nov. 2023.

[0033] In the ditherless bias control technology described so far, it is necessary to learn in advance how the intensity of the reference light output from the MZI changes when bias drift occurs and record this in a data table.

[0034] The changes in optical intensity shown in Figures 16 to 18 are for a single MZI filter as shown in Figure 13, where no modulation signal is applied and only one type of bias is controlled. In contrast, an IQ modulator as shown in Figure 15 has three types of bias voltages to be controlled, and these can cause bias drift independently of each other, so there are an extremely large number of conditions that must be learned in advance.

[0035] Furthermore, the relative relationship between bias drift and reference light intensity depends on the signal formats of modulated signal I and modulated signal Q. Furthermore, due to the periodicity of the MZI, there are an infinite number of null points and quad points, and the bias drift and change in reference light intensity change depending on which null point or quad point is selected in the startup sequence. This means that it is very difficult to reuse past data when creating the above-mentioned data table, and that the data table must be recreated every time the transmitter is started up, especially when the signal format or wavelength is changed.

[0036] Furthermore, when a change in the reference light intensity due to bias drift is confirmed by actual measurement during transmission service operation, the problem arises of how to specifically compare it with the aforementioned data table. The number of data points that can be recorded in a data table is finite, and because bias drift is an analog change, it is almost impossible for data exactly identical to the measured reference light intensity to exist in the data table. For this reason, the closest data will be found in the data table, but the criteria for determining "closest" must be clearly defined.

[0037] In view of the above circumstances, an object of the present invention is to provide a control circuit and an optical circuit control method that can more easily compensate for bias drift in an optical circuit composed of a single or multiple MZIs in a ditherless manner.

[0038] One aspect of the present invention is a control circuit for controlling at least one optical path difference between two optical paths of each of a single or multiple Mach-Zehnder interferometers included in an optical circuit, the control circuit including: a control unit for controlling the optical path difference with one or multiple bias voltages or bias powers; a reference light generation unit for generating reference lights of N types of wavelengths, where N is a natural number, different from the wavelength of a signal light input to the optical circuit; a reference light input unit for wavelength-multiplexing the N types of reference lights with the signal light and inputting the multiplexed reference lights to the optical circuit; an optical power monitor unit for monitoring the optical intensities of the N types of reference lights after propagating through the single or multiple Mach-Zehnder interferometers included in the optical circuit, or the optical intensities of the N types of reference lights and the signal light; and a monitoring result of the optical intensities of the N types of reference lights by the optical power monitor unit, or and a normalization unit that normalizes each of the monitor results of the optical intensities of the N types of reference light and the optical intensity of the signal light, wherein when a reference light vector is an N-dimensional vector having as its components the monitor results of the normalized optical intensities of the N types of reference light, or an N+1-dimensional vector having as its components the monitor results of the normalized optical intensities of the N types of reference light and the optical intensity of the signal light, the normalization unit performs normalization so that when optical path differences of one or more of the Mach-Zehnder interferometers included in the optical circuit are all optimal, the norm of the reference light vector becomes zero, and the norm increases as at least one optical path difference deviates from the optimal value, and the control unit changes one or more of the bias voltages or the bias powers so that the norm of the reference light vector approaches zero.

[0039] One aspect of the present invention includes a control step of controlling at least one optical path difference of two optical paths of each of a single or multiple Mach-Zehnder interferometers included in an optical circuit by one or multiple bias voltages or bias powers; a reference light generation step of generating reference lights of N kinds of wavelengths, where N is a natural number, different from the wavelength of a signal light input to the optical circuit; a reference light input step of wavelength-multiplexing the N kinds of reference lights with the signal light and inputting the wavelength-multiplexed reference lights into the optical circuit; a monitor step of monitoring the optical intensities of the N kinds of reference lights after propagation through the single or multiple Mach-Zehnder interferometers included in the optical circuit, or the optical intensities of the N kinds of reference lights and the signal light; and a monitor step of monitoring the optical intensities of the N kinds of reference lights by the monitor step, or the optical intensities of the N kinds of reference lights and the and a normalization step of normalizing each of the monitor results of the optical intensities of the signal light, wherein in the normalization step, when a reference light vector is an N-dimensional vector having as its components the monitor results of the optical intensities of the normalized N types of reference light, or an N+1-dimensional vector having as its components the monitor results of the optical intensities of the normalized N types of reference light and the optical intensities of the signal light, normalization is performed such that a norm of the reference light vector becomes zero when optical path differences of one or more of the Mach-Zehnder interferometers included in the optical circuit are all optimal, and the norm increases as at least one optical path difference deviates from an optimal value, and in the control step, one or more of the bias voltages or the bias powers are changed so that the norm of the reference light vector approaches zero.

[0040] The present invention makes it possible to more easily compensate for bias drift in an optical circuit configured with a single or multiple MZIs in a ditherless manner.

[0041] 1 is a diagram showing the configuration of a control circuit according to the first embodiment. FIG. 2 is a flowchart showing the operation of a controller in a start-up sequence according to the first embodiment. FIG. 3 is a flowchart showing the operation of a controller during execution of a transmission service according to the first embodiment. FIG. 4 is a flowchart showing detailed processing of a controller according to the first embodiment. FIG. 5 is a diagram showing the intensities of first and second reference light according to the first embodiment. FIG. 6 is a diagram showing the result of calculation processing according to the first embodiment. FIG. 7 is a diagram showing a part of a data table according to the first embodiment. FIG. 8 is a diagram showing the configuration of a control circuit according to the second embodiment. FIG. 9 is a diagram showing changes in the light intensity of a reference light source according to the second embodiment and the monitoring results obtained by an optical power monitor. FIG. 10 is a diagram plotting reference light vectors according to the fourth embodiment. FIG. 11 is a diagram showing the configuration of an IQ optical modulator according to a variation of the first to fourth embodiments. FIG. 12 is a diagram showing the configuration of an IQ optical modulator according to a variation of the first to fourth embodiments. FIG. 13 is a diagram showing the configuration of an MZI. FIG. 14 is a diagram showing the relationship between the output light intensity of an MZI and the optical path difference. FIG. 15 is a diagram showing the configuration of an IQ optical modulator. FIG. 16 is a diagram showing the output light intensity of an MZI. FIG. 17 is a diagram showing the output light intensity of an MZI.

[0042] Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings. This embodiment relates to a control circuit and a control method for an optical circuit that includes a Mach-Zehnder interferometer (MZI) as a part thereof. In this embodiment, a data table of intensity changes of a reference light used for ditherless monitoring of bias drift in an optical circuit composed of a single or multiple MZIs is created with fewer steps. Furthermore, this embodiment enables comparison of the intensity changes of the reference light actually measured after the data table is created with the data table created during the startup sequence using simple rules.

[0043] <First Embodiment> Fig. 1 is a diagram showing the configuration of a control circuit according to a first embodiment. Fig. 1 shows only functional blocks related to this embodiment. Fig. 1 also shows an object controlled by the control circuit. The object controlled is an optical circuit 100 made up of MZIs. Here, the optical circuit 100 may be an optical filter made up of a single MZI, such as the MZI 91 shown in Fig. 13, or an optical modulator made up of multiple MZIs, such as the IQ optical modulator 92 shown in Fig. 15. In the following description, as an example, the optical circuit 100 is assumed to be the same as the IQ optical modulator 92 shown in Fig. 15.

[0044] The optical circuit 100 includes a modulation signal I, a modulation signal Q, and a bias voltage V ｂ＿Ｉ , bias voltage V ｂ＿Ｑ , and bias voltage V ｂ＿Ｐｈ is given from the outside. The optical circuit 100 has the optical input port 1 and the optical output port 7 already shown in FIG.

[0045] The control circuit includes a wavelength division multiplexing coupler 8, an optical tap circuit 10, an optical filter 12, a wavelength demultiplexing coupler 13, a first reference light source 35-1, a second reference light source 35-2, a controller 41, an optical power monitor 52, a first reference light power monitor 53-1, a second reference light power monitor 53-2, an A / D (Analog-to-Digital) converter 102, a first reference light A / D converter 103-1, a second reference light A / D converter 103-2, a first D / A (Digital-to-Analog) converter 104, a second D / A converter 105, and a third D / A converter 106. The controller 41 includes a storage unit 42. The controller 41 may include an external storage unit 42. The storage unit 42 stores a database.

[0046] The first reference light source 35-1 emits a reference light having a wavelength λ １ The second reference light source 35-2 outputs a first reference light having a wavelength λ ２ The wavelength division multiplexing coupler 8 receives the first reference light output from the first reference light source 35-1, the second reference light output from the second reference light source 35-2, and the signal light having a wavelength λ. １ , and wavelength λ２ are different from each other. The wavelength division multiplexing coupler 8 wavelength-multiplexes the input first reference light, second reference light, and signal light, and outputs the multiplexed light. The first reference light, second reference light, and signal light that have been wavelength-multiplexed by the wavelength division multiplexing coupler 8 are input to the optical input port 1.

[0047] The optical tap circuit 10 taps the wavelength-multiplexed light output from the optical output port 7. The wavelength demultiplexing coupler 13 demultiplexes the wavelength-multiplexed light tapped by the optical tap circuit 10. The optical power monitor 52 monitors the optical intensity of the signal light of wavelength λ demultiplexed by the wavelength demultiplexing coupler 13. The first reference light power monitor 53-1 monitors the optical intensity of the signal light of wavelength λ demultiplexed by the wavelength demultiplexing coupler 13. １ The second reference beam power monitor 53-2 monitors the light intensity of the first reference beam having the wavelength λ 1 separated by the wavelength separation coupler 13. ２ The light intensity of the second reference light is monitored.

[0048] The term "light intensity" used in this embodiment refers to a value integrated over a time period that is sufficiently longer than the reciprocal of the bit rate or baud rate of modulated signal I and modulated signal Q. Light intensity may also be simply referred to as "intensity."

[0049] The other light branched by the optical tap circuit 10 is input to the optical filter 12. The optical filter 12 removes each reference light from the light input from the optical tap circuit 10, and then outputs output light of wavelength λ.

[0050] The controller 41 controls the bias voltages of the optical circuit 100. The controller 41 is a digital circuit that operates according to a program code that has been input in advance and is capable of recording a created data table.

[0051] The A / D converter 102 digitizes the monitoring result of the optical power monitor 52 and inputs it to the controller 41. The first reference beam A / D converter 103-1 digitizes the monitoring result of the first reference beam power monitor 53-1 and inputs it to the controller 41. The second reference beam A / D converter 103-2 digitizes the monitoring result of the second reference beam power monitor 53-2 and inputs it to the controller 41.

[0052] The controller 41 outputs the bias voltage V of the optical circuit 100 via the first D / A converter 104. ｂ＿Ｉ , and outputs the bias voltage V of the optical circuit 100 via the second D / A converter 105. ｂ＿Ｑ , and outputs the bias voltage V of the optical circuit 100 via the third D / A converter 106. ｂ＿Ｐｈ Control.

[0053] In FIG. 1, the number of reference beams is set to two. However, the number of reference beams can be set to N by adding a third reference beam light source 35-3, a fourth reference beam light source 35-4, ..., an Nth reference beam light source 35-N, a third reference beam power monitor 53-3, a third reference beam power monitor 53-4, ..., an Nth reference beam power monitor 53-N, and a third reference beam A / D converter 103-3, a fourth reference beam A / D converter 103-4, ..., an Nth reference beam A / D converter 103-N. Here, N is an integer equal to or greater than three. However, the wavelengths of the reference beams must all be different. Furthermore, it is desirable that approximately half of the reference beams have a longer wavelength than the signal beam, and the other half have a shorter wavelength than the signal beam.

[0054] When the number of reference beams is N, where n is an integer between 1 and N, the nth reference beam source 35-n has a wavelength λ ｎ The wavelength division multiplexing coupler 8 multiplexes the first to Nth reference light beams output by the first to Nth reference light beam sources 35-1 to 35-N, respectively, with signal light having a wavelength λ, and inputs the multiplexed signal light to the optical input port 1. The optical power monitor 52 monitors the optical intensity of the signal light of wavelength λ separated by the wavelength separation coupler 13, and the nth reference light power monitor 53-n monitors the optical intensity of the signal light of wavelength λ separated by the wavelength separation coupler 13. ｎ The A / D converter 102 digitizes the monitoring result of the optical power monitor 52 and inputs it to the controller 41, and the nth reference beam A / D converter 103-n digitizes the monitoring result of the nth reference beam power monitor 53-n and inputs it to the controller 41.

[0055] During the start-up sequence, if necessary, the controller 41 may adjust the bias voltage V ｂ＿Ｉ and V ｂ＿ＱIt is also possible to superimpose a pilot tone on the signal light of wavelength λ and perform dithering. The control circuit also digitally multiplies and smooths the digital data captured via the A / D converter 102 by the pilot tone, and can synchronously detect the dither component (pilot tone) superimposed on the signal light of wavelength λ. This makes it possible to control each bias to an optimal value during the startup sequence using the method described in Non-Patent Document 1, etc.

[0056] As mentioned above, during the startup sequence before the start of transmission service, there is no problem of penalties on the transmission signal due to dithering, so it is possible to increase the amplitude of the pilot tone and increase the signal-to-noise (SN) ratio of the output of the optical power monitor 52.

[0057] FIG. 2 is a flowchart showing the operation of the controller 41 in a start-up sequence, and FIG. 3 is a flowchart showing the operation of the controller 41 while a transmission service is being executed.

[0058] 2, after the start-up sequence starts, in step S1, the controller 41 supplies power to peripheral devices of the optical circuit 100. This may include power to the first reference light source 35-1 and the A / D converter 102 shown in FIG. 1 as well as a light source for signal light not shown in FIG. 1 or a driver amplifier for amplifying the modulated signal I and the modulated signal Q.

[0059] In step S2, the controller 41 uses a pseudo-random signal to generate a dummy modulated signal I and a dummy modulated signal Q, and applies these to the optical circuit 100. The format of these dummy signals is set to be the same as the format of the signal used during the transmission service operation period.

[0060] In step S3, the controller 41 performs bias condition monitoring using dithering as described in Non-Patent Document 1 and the like, and adjusts the bias voltage V ｂ＿Ｉ , V ｂ＿Ｑ , V ｂ＿Ｐｈis at an optimal value. If it is determined in step S4 that any of the biases is not optimal, the controller 41 proceeds to step S5. In step S5, the controller 41 performs feedback control using the monitoring results of step S3 so that the bias voltages become optimal values, and then returns to step S3. On the other hand, if it is determined in step S4 that all of the biases are optimal, the controller 41 proceeds to step S6.

[0061] In step S6, the controller 41 controls the bias voltage V ｂ＿Ｉ , V ｂ＿Ｑ , V ｂ＿Ｐｈ One or more values ​​of the above are intentionally shifted from the optimum value at the same time. At this time, as shown in Figures 16 to 18, the intensities of the N types of reference light beams change differently, but the controller 41 normalizes the intensities of these reference light beams and records them in the database stored in the storage unit 42. Details of normalization will be described later. When M is an integer equal to or greater than 1, step S6 needs to be repeated N x M times.

[0062] After the database is created in step S6, in step S7, the bias control circuit calculates the bias voltage V ｂ＿Ｉ , V ｂ＿Ｑ , V ｂ＿Ｐｈ are restored to their optimum values. The values ​​already determined in step S4 can be used as these optimum values. However, if the database creation time in step S6 is long and there is concern about bias drift during that time, the controller 41 may execute the loop process from step S3 to step S5 again instead of step S7.

[0063] The database is created, and the bias voltage V ｂ＿Ｉ , V ｂ＿Ｑ , V ｂ＿Ｐｈ When the value is returned to the optimum value, the start-up sequence ends and transmission service begins.

[0064] As shown in FIG. 3, after the start of the transmission service, in step S8, the controller 41 applies the transmission modulation signal I and the transmission modulation signal Q to the optical circuit 100 in place of the dummy modulation signal I and the dummy modulation signal Q.

[0065] Thereafter, the controller 41 periodically performs the process of step S9 to monitor the intensities of the N types of standardized reference light.

[0066] In step S10, the controller 41 controls each bias voltage V ｂ＿Ｉ , V ｂ＿Ｑ , V ｂ＿Ｐｈ is the optimum value, and the intensities of the N types of normalized reference light monitored in step S9 are determined to be significantly different from each other. Details of this determination will be described later. If there is a significant difference, bias drift has occurred, and the controller 41 proceeds to step S11 to correct it.

[0067] In step S11, the controller 41 compares the normalized intensity of each reference beam recorded in the database with the current normalized intensity of each reference beam to determine whether the bias has drifted positively or negatively. In step S12, the controller 41 feeds back the determination result in step S11 to the bias voltage, and then returns to step S9. The details of this determination will be described later.

[0068] Fig. 4 is a flowchart showing in more detail the processing of the controller 41 in step S6 in Fig. 2. In order to distinguish the number of steps from those in Fig. 2 and Fig. 3, the number of steps is counted from 100 here.

[0069] Immediately before step S100 is executed, in step S4 of FIG. 2, the bias voltage V ｂ＿Ｉ , bias voltage V ｂ＿Ｑ , and bias voltage V ｂ＿Ｐｈ is set to an optimum value, but the controller 41 adjusts the bias voltage V ｂ＿Ｉ ΔV ｂ＿Ｉ (m), bias voltage V ｂ＿Ｑ ΔVｂ＿Ｑ (m), bias voltage V ｂ＿Ｐｈ ΔV ｂ＿Ｐｈ Shift by (m), where m is the first counter in Loop 1, and Loop 1 is executed while incrementing m by 1 from m=1 to M.

[0070] ΔV ｂ＿Ｉ (m), ΔV ｂ＿Ｑ (m), ΔV ｂ＿Ｐｈ (m) can be either positive or negative, including zero, but ΔV ｂ＿Ｉ (m), ΔV ｂ＿Ｑ (m), ΔV ｂ＿Ｐｈ (m) is ±0.05V πｂｉａｓ The voltage is changed within a range of 0.01V. πｂｉａｓ It is desirable that the V πｂｉａｓ is the half-wave voltage at the bias port of the MZI that the optical circuit 100 has.

[0071] When the change range and step width are the recommended values ​​above, ΔV ｂ＿Ｉ The number of values ​​that (m) can take, including zero, is 0.05 x 2 / 0.01 + 1 = 11. ｂ＿Ｑ (m), ΔV ｂ＿Ｐｈ The same is true for (m), so the total number of possible combinations of biases is 11^3 = 1331. Therefore, M = 1331.

[0072] At the beginning of loop 1, m=1, ΔV ｂ＿Ｉ (1) = ΔV ｂ＿Ｑ (1) = ΔV ｂ＿Ｐｈ (1) = 0. In order to check whether there is bias drift during the loop, ΔV ｂ＿Ｉ (m) = ΔV ｂ＿Ｑ (m) = ΔV ｂ＿Ｐｈ It is advisable to double check that (m) = 0. In this case, M takes a value of 1331 or greater.

[0073] In step S101, the controller 41 records the intensity of the reference light. The reference light to be measured is the intensity of the nth reference light obtained by the nth reference light power monitor 53-n according to the value n of the second counter in the loop 2. Here, if m is 1 and the total bias voltage is at an optimal value, the intensity of the nth reference light obtained at this time is defined as the "nth reference value."

[0074] In step S102, the controller 41 divides the intensity of the reference light obtained in step S101 by the nth reference value. If m=1 when step S102 is executed, the answer obtained will be 1 regardless of n.

[0075] In step S103, the controller 41 subtracts 1 from the calculation result obtained in step S102. If m=1 when step S103 is executed, the obtained answer will be 0 regardless of n.

[0076] The above-described processing of loop 2 from step S101 to step S103 is performed while increasing n by 1 from 1 to N. As a result, the processing of step S101 to step S103 is performed for all of the N types of reference light.

[0077] Here, we will temporarily leave the explanation of FIG. 4 and explain the process from step S101 to step S103 based on data obtained from an actual experiment.

[0078] FIG. 5 shows the measured intensities of the first and second reference beams. The first reference beam is 1490 nm, and the second reference beam is 1625 nm. The main signal is 1547 nm, and the bias voltages are optimized (ΔV ｂ＿Ｉ (1) = ΔV ｂ＿Ｑ (1) = ΔV ｂ＿Ｐｈ When (1)=0), each MZI of the IQ optical modulator 92 is at the Null and Quad points at 1547 nm. πｂｉａｓ was 7V.

[0079] FIG. 5(a) shows the measurement results from m=1 to m=8. ΔV ｂ＿Ｑ (m) and ΔV ｂ＿Ｐｈ (m) is fixed to 0 regardless of m.ｂ＿Ｉ (m) is ±0.2V (±0.029V πｂｉａｓ ) range, and the sweep step size is 0.1 V (0.014 V πｂｉａｓ ) The horizontal axis is ΔV ｂ＿Ｉ (1) ~ ΔV ｂ＿Ｉ (8) shows the voltage change in volts.

[0080] FIG. 5(b) shows the measurement results from m=9 to m=16. ｂ＿Ｉ (m) and ΔV ｂ＿Ｐｈ (m) is fixed to 0 regardless of m, and ΔV ｂ＿Ｑ Only (m) is changed. ΔV ｂ＿Ｑ The sweep range and step width of (m) are the same as those of the aforementioned ΔV ｂ＿Ｉ The horizontal axis is the same as that of (m). ｂ＿Ｑ (9) ~ ΔV ｂ＿Ｑ (16) shows the voltage change in volts.

[0081] FIG. 5(c) shows the measurement results from m=17 to m=24. ΔV ｂ＿Ｉ (m) and ΔV ｂ＿Ｑ (m) is fixed to 0 regardless of m, and ΔV ｂ＿Ｐｈ Only (m) is changed. ΔV ｂ＿Ｐｈ The sweep range and step width of (m) are the same as those of the aforementioned ΔV ｂ＿Ｉ The horizontal axis is the same as that of (m). ｂ＿Ｐｈ (17) ~ ΔV ｂ＿Ｐｈ (24) shows the voltage change in volts.

[0082] When m is 1, 8, 9, 16, 17, and 24, reproducibility is confirmed and all ΔV ｂ＿Ｉ (1) = ΔV ｂ＿Ｑ (1) = ΔV ｂ＿Ｐｈ 5(a) to 5(c), the values ​​indicated by downward arrows are the first reference values, and the values ​​indicated by upward arrows are the second reference values.

[0083] FIG. 6 shows the results of actually performing the calculation processes of steps S102 and S103 on the data of FIG. 5. The horizontal axis is the same as in FIG. 5, but the vertical axis has been changed to represent the normalized intensities of the first and second reference beams. By definition, the first reference beam (1490 nm) and the second reference beam (1625 nm) pass through the origin, where both the horizontal and vertical axes are zero. However, the measured data show a slight discrepancy near the origin. This is due to bias drift that occurred during the sweep.

[0084] Now, let us return to the explanation of FIG. 4. When the loop 2, which is the inner loop of the double loop, is passed, ΔV ｂ＿Ｉ (m), ΔV ｂ＿Ｑ (m), ΔV ｂ＿Ｐｈ In step S104, the controller 41 creates a data table in the database in which the 3+N pieces of data and the value of m are grouped into one row.

[0085] When the number of rows in the data table reaches M and the outer loop, Loop 1, is exited, the flow shown in FIG. 4, i.e., step S106 in FIG. 3, is completed. However, if the time until completion is long to some extent, the bias voltage V ｂ＿Ｉ , V ｂ＿Ｑ , V ｂ＿Ｐｈ To compensate for this, after step S104, steps S105 to S107 are performed in the same manner as steps S3 to S5 in FIG. 2, and the optimum value of the bias voltage V ｂ＿Ｉ , V ｂ＿Ｑ , V ｂ＿Ｐｈ The optimization may be redone.

[0086] Fig. 7 shows a portion of a data table that has actually been created. Fig. 7 shows only cases where m is from 1 to 32, but in actual operation, as mentioned above, it is desirable that the upper limit of m, M, be 1000 or more. When m = 1, m = 8, m = 16, m = 24, and m = 32, ΔV ｂ＿Ｉ(m) = ΔV ｂ＿Ｑ (m) = ΔV ｂ＿Ｐｈ Although (m) = 0, the normalized reference light intensity is slightly out of sync. This is because bias drift occurs during the bias change process, and the bias voltage V ｂ＿Ｉ , V ｂ＿Ｑ , V ｂ＿Ｐｈ This is the result of a slight change in the optimum value of V. In order to compensate for this deviation, the processes from step S105 to step S107 are performed, and V is calculated for m=9, m=17, and m=25. ｂ＿Ｉ (m), V ｂ＿Ｑ (m), V ｂ＿Ｐｈ The optimization of (m) is being redone.

[0087] Focus on a specific m in the data table shown in Fig. 7. An N-dimensional vector consisting of N types of parameters from the first normalized reference beam intensity to the Nth normalized reference beam intensity at the specific m is defined as a "reference beam vector." M types of reference beam vectors are described in the data table, and the reference beam vectors have the following properties.

[0088] (Property 1) The norm of the reference light vector increases when the error of each bias is large, and becomes zero when all the biases are at their optimum values. (Property 2) Each component of the reference light vector does not depend on the wavelength dependency of the loss of the wavelength demultiplexing coupler 13 or the wavelength dependency of the optical / electrical conversion efficiency of the nth reference light power monitor 53-n.

[0089] By utilizing these properties, it becomes possible to compensate for bias drift during a transmission service without dithering. This will be explained in more detail by returning to the processing during a transmission service shown in FIG.

[0090] In step S9 of Fig. 3, it is stated that "the controller 41 monitors the intensities of N types of normalized reference light." Explaining this process in more detail, the controller 41 records the intensity of the nth reference light at a certain time during the transmission service period, as in step S101 of Fig. 4, divides the recorded intensity by the nth reference value, as in step S102 of Fig. 4, and subtracts 1 from the division result to obtain the normalized intensity of the nth reference light, as in step S103 of Fig. 4, repeating this process from n=1 to N. The nth reference value can be obtained by referencing the value for m=1 in the data table already created in step S6 of Fig. 2.

[0091] The normalized intensities of the N types of reference beams obtained in step S9 can be treated as a reference beam vector, similar to when the data table was created. That is, the reference beam vector obtained here is a vector in which the normalized intensities of the nth reference beam (n=1 to N) are arranged.

[0092] In step S10 of FIG. 3, it is stated that "the controller 41 determines whether the normalized intensity of each reference beam is significantly different from the time when each bias was at its optimal value." This process will be explained in more detail. First, the controller 41 calculates the norm of the reference beam vector obtained in step S9. If this norm is zero, or if it is a value that is not zero but is within the measurement error range, it is determined that each bias remains in an optimal state, and the determination in step S10 is NO. Conversely, if the norm of the reference beam vector obtained in step S9 is significantly larger than the measurement error, the determination in step S10 is YES.

[0093] 3, it is stated that "the controller 41 compares the normalized intensity of each reference light recorded in the database with the current normalized intensity of each reference light, and determines whether the bias has drifted positively or negatively." This process will be explained in more detail.

[0094] The controller 41 subtracts the reference beam vector obtained in step S9 from the reference beam vector written in the m-th row of the data table. Next, the controller 41 calculates the norm of the new vector obtained as a result of the subtraction. The controller 41 repeats this calculation process from m=1 to m=M, and finds the m that gives the smallest norm. ｍｉｎ The controller 41 checks which ΔV is recorded in the row of the data table. ｂ＿Ｉ (m ｍｉｎ ), ΔV ｂ＿Ｑ (m ｍｉｎ ), ΔV ｂ＿Ｐｈ (m ｍｉｎ ) is determined to be the drift amount of each bias voltage.

[0095] In step S12 of FIG. 3, it is written that "the controller 41 performs feedback control on the bias voltage based on the result of step S11." As shown in FIG. ｂ＿Ｉ (m ｍｉｎ ), ΔV ｂ＿Ｑ (m ｍｉｎ ), ΔV ｂ＿Ｐｈ (m ｍｉｎ ) contains a sign so that it is possible to determine how much each bias voltage should be increased or decreased.

[0096] It should be noted that during the transmission service period, the reference light vector obtained in step S9 changes continuously due to bias drift, whereas ΔV recorded in the data table ｂ＿Ｉ (m), ΔV ｂ＿Ｑ (m), ΔV ｂ＿Ｐｈ Since (m) is discrete, it is almost impossible for the reference light vector obtained in step S9 to perfectly match the M kinds of reference light vectors listed in the data table. ｂ＿Ｉ (m ｍｉｎ ), ΔV ｂ＿Ｑ (m ｍｉｎ ), ΔV ｂ＿Ｐｈ (m ｍｉｎ ) is used as an estimate of the drift amount of each bias voltage, but it must be processed as it may contain errors.

[0097] As a specific example, assume that the two components of the reference light vector obtained in step S9, i.e., the normalized intensity of the first reference light and the normalized intensity of the second reference light, are −0.034 and 0.030, respectively. In this case, m ｍｉｎ becomes 2.

[0098] According to FIG. 7, ΔV ｂ＿Ｉ (m ｍｉｎ ) = 0.1, ΔV ｂ＿Ｑ (m ｍｉｎ ) = ΔV ｂ＿Ｐｈ (m ｍｉｎ ) = 0, so the bias voltage V ｂ＿Ｉ However, it is desirable to prevent hunting by suppressing the amount of feedback to, for example, 80% of the original value, to -0.08 V, rather than setting it to -0.1 V.

[0099] As described above, the controller 41 changes one or more bias voltages by feedback control that determines a feedback coefficient based on the norm of the reference light vector.

[0100] 1 showing the first embodiment, there are two types of reference light and one signal light, and the reference light vector is a two-dimensional vector. Here, the signal light may be treated as a third reference light, and the reference light vector may be a three-dimensional vector using the output of the optical power monitor 52. When there are N types of reference light, the reference light vector is an (N+1)-dimensional vector.

[0101] Alternatively, it is possible to some extent to use only one reference light and make the reference light vector a one-dimensional vector, that is, a single signed variable. However, since the MZI has periodicity as shown in Figures 16 to 18, if a single reference light is used, it is possible that the same optical output intensity can be obtained with different bias drifts. In order to prevent erroneous determination, it is desirable to use two or more reference lights with wavelengths different from that of the signal light.

[0102] 1 showing the first embodiment, a wavelength demultiplexing coupler 13 is used to separate two types of reference light and signal light, and the intensities of the signal light and reference light are measured using an optical power monitor 52, a first reference light power monitor 53-1, and a second reference light power monitor 53-2. However, by turning the reference light on and off in a time-sharing manner, the wavelength demultiplexing coupler and reference light power monitor can be omitted. The second embodiment will be described focusing on the differences from the first embodiment.

[0103] Figure 8 shows the configuration of a control circuit of the second embodiment. In Figure 8, parts that are the same as those of the first embodiment shown in Figure 1 are given the same reference numerals, and their description will be omitted. The difference between the control circuit shown in Figure 8 and the control circuit shown in Figure 1 is that the wavelength demultiplexing coupler 13 and the first and second reference beam power monitors 53-1 and 53-2 are omitted. Another difference is that in the second embodiment, a controller 41 controls the on / off of the first reference beam light source 35-1 and the second reference beam light source 35-2.

[0104] 9 is a diagram showing a change in the light intensity at the time of light output from each of the first reference light source 35-1 and the second reference light source 35-2 in the second embodiment, and a monitor result output from the optical power monitor 52. Here, time t １ From time t １０ The time required to reach the predetermined value is much longer than the reciprocal of the bit rate or baud rate of the modulated signal I and the modulated signal Q, but is short enough to make the influence of bias drift negligible.

[0105] As shown in FIG. 9A, the first reference light source 35-1 emits a reference light beam at time t ２ From time t ３ and time t ４ From time t ５ 9B, the second reference light source 35-2 is turned on at time t ６ From time t ７ and time t ８ From time t ９and is turned on at certain times, but is turned off by the controller 41 at other times.

[0106] The first reference light, second reference light, and signal light output from the optical circuit 100 are input to the optical power monitor 52. FIG. 9C schematically shows the output of the optical power monitor 52. Because the signal light is not extinguished, the output of the optical power monitor 52 is always a non-zero value. However, the output of the optical power monitor 52 is high during periods A and C, when the first reference light is on. The output of the optical power monitor 52 is also high during periods E and G, when the second reference light is on. However, because the optical loss of the first reference light and the second reference light due to the optical circuit 100 depends on the bias condition, the increase in the amount of increase during periods A and C is generally not the same as the increase in the amount of increase during periods E and D.

[0107] By calculating the average value P(A, C) of the optical power monitor output in periods A and C and the average value P(B, D) of the optical power monitor output in periods B and D, and performing the calculation P(A, C)×(+1)+P(B, D)×(−1), it is possible to obtain the first reference light after passing through the optical circuit 100. Similarly, by performing the calculation P(E, G)×(+1)+P(F, H)×(−1), it is possible to obtain the second reference light after passing through the optical circuit 100.

[0108] Here, time t ｉ and time t ｉ＋１ If the time difference between the input signal and the output signal is always constant, the above calculation process is equivalent to synchronous detection. Therefore, a band-pass filter or low-pass filter may be configured as a digital filter inside the controller 41, and synchronous detection may be performed digitally.

[0109] It should be noted that the intensity of the signal light at the time of passing through the optical circuit 100 can be obtained by averaging the values ​​of the periods B, D, F, and H.

[0110] <Third Embodiment> In the first and second embodiments, before creating the data table shown in FIG. 7, all biases are optimized using bias control using dithering as described in Non-Patent Document 1 (steps S3 to S5). After that, the controller 41 optimizes the bias voltage V ｂ＿Ｉ , V ｂ＿Ｑ , V ｂ＿Ｐｈ ΔV ｂ＿Ｉ (m), ΔV ｂ＿Ｑ (m), ΔV ｂ＿Ｐｈ The data table in FIG. 7 is created by shifting the data by (m).

[0111] If we assume that bias drift that occurs naturally due to temperature changes or other reasons during the data table creation period can be ignored, then ΔV ｂ＿Ｉ (m), ΔV ｂ＿Ｑ (m), ΔV ｂ＿Ｐｈ The value of (m) can be regarded as the difference between the current value of each bias voltage and the optimum value of each bias voltage, so that bias drift can be simulated.

[0112] However, in addition to the shift forcibly caused by the controller 41, bias drift that occurs naturally due to temperature changes and other reasons also occurs at the same time, and if this influence is significant enough to be non-negligible, the correspondence between the bias shift amount shown on the left side of the data table in Fig. 7 and the normalized reference light intensity shown on the right side of the data table in Fig. 7 will become inaccurate. In fact, as mentioned above, there is a slight error when m = 1, m = 8, m = 16, m = 24, and m = 32.

[0113] To solve this problem, bias control may be performed again as shown in steps S105 to S107. However, if this operation is performed frequently, a new problem arises in that the time required to create the database increases.

[0114] To solve these problems, bias condition monitoring is performed at each stage during the process of creating the data table, and the deviation of the shifted bias voltage from the optimum bias voltage is measured and entered on the left side of the data table in Figure 7.

[0115] Since the database is created during the startup sequence, the bias condition monitor can use dithering using a pilot tone as described in Non-Patent Document 1 and the like.

[0116] <Fourth embodiment> In the explanation of the first, second and third embodiments, an example of an actual measurement of the data table is shown in Figure 7. However, this is only a partial example shown for the purpose of explanation, and the data table that must actually be created is much larger, and as mentioned above, M is preferably 1,000 or more. For m=25 to 32 in the table shown in Figure 7, ΔV ｂ＿Ｉ (m) and ΔV ｂ＿Ｑ (m) are swept simultaneously with the same sign, but each bias voltage is shifted individually in a random direction, so ΔV ｂ＿Ｉ (m) and ΔV ｂ＿Ｑ (m) and the case where they drift in the opposite direction, ΔV ｂ＿Ｉ The absolute value of the drift of (m) is ΔV ｂ＿Ｑ In this case, various situations must be considered, such as when the absolute value of the drift of (m) is larger than the absolute value of the drift of (m). For this reason, creating a data table takes a long time. In the fourth embodiment, an embodiment that can reduce the time required for creating a data table will be described.

[0117] Fig. 10 is an XY plot of the two-dimensional reference beam vectors listed in the data table shown in Fig. 7. The horizontal axis represents the normalized intensity of the first reference beam, and the vertical axis represents the normalized intensity of the second reference beam.

[0118] The circle symbol is ΔV ｂ＿Ｉ When the value is increased, it extends to the upper left, and when it is decreased, it extends to the lower right. The triangle symbol indicates ΔV ｂ＿Ｑ When only ΔV is changed, it extends to the upper left when it is increased, and extends to the lower right when it is decreased. ｂ＿Ｐｈ When the value is increased, the image extends downward and to the left, and when the value is decreased, the image extends upward and to the right.

[0119] In the figure, the point marked A is m=7 in the data table, and ΔV ｂ＿Ｉ(7)=-0.1, ΔV ｂ＿Ｑ (7) = ΔV ｂ＿Ｐｈ (7) = 0. The point marked B is m = 15 in the data table, and ΔV ｂ＿Ｑ (15)=-0.1, ΔV ｂ＿Ｉ (15) = ΔV ｂ＿Ｐｈ (15) = 0. The point marked C is m = 31 in the data table, and ΔV ｂ＿Ｉ (31) = ΔV ｂ＿Ｑ (31)=-0.1, ΔV ｂ＿Ｐｈ (31) = 0.

[0120] 10, it should be noted that the reference beam vector from the origin to C is approximately equal to the vector sum of the reference beam vector from the origin to A and the reference beam vector from the origin to B. From this, the following prediction holds:

[0121] m １ and m ２ is an integer between 1 and M, and r １ and ２ When is an integer other than 0, the shift amounts of the three biases are respectively １ ×ΔV ｂ＿Ｉ (m １ ) + r ２ ×ΔV ｂ＿Ｉ (m ２ ), r １ ×ΔV ｂ＿Ｑ (m １ ) + r ２ ×ΔV ｂ＿Ｑ (m ２ ), r １ ×ΔV ｂ＿Ｐｈ (m １ ) + r ２ ×ΔV ｂ＿Ｐｈ (m ２ ), then the reference beam vector in that state can be approximated by the following equation:

[0122] r １ ×{m １ Reference beam vector at ２ ×{m １ reference beam vector at

[0123] m １ =7 (point A), m ２=15 (point B), r １ =r ２ Using the values ​​shown in FIG. 7, we verify whether m=31 (point C) can be estimated under the condition of m=1.

[0124] The normalized intensity of the first reference light is 0.055 for A (m=7), 0.033 for B (m=15), and the sum of the two is 0.088, while the sum is 0.077 for C (m=31).

[0125] The normalized intensity of the second reference beam is −0.028 for A (m=7), −0.044 for B (m=15), and the sum of the two is −0.072, while it is −0.074 for C (m=31). These results show that good approximate values ​​sufficient for practical use can be obtained.

[0126] The controller 41 uses this approximation to estimate the amount of bias voltage shift that was not actually measured and the reference light vector in that case, and records them in the data table, thereby expanding the data table. However, since the value obtained by the linear sum is only an approximation, in order to suppress errors, １ and ２ It is desirable that the signs of the s and s are kept between 0 and 1.

[0127] <Variations of the First to Fourth Embodiments> In the embodiments described so far, the optical circuit 100 is the same as the IQ optical modulator 92 shown in FIG. 15. However, the optical circuit 100 is not limited to an IQ optical modulator, and may be an optical modulator composed of a single MZI. In this case, the modulation signal I and the modulation signal Q are replaced with a single modulation signal. Also, the bias voltage V ｂ＿Ｉ , V ｂ＿Ｑ , V ｂ＿Ｐｈ is a single bias voltage V ｂ is replaced by.

[0128] The optical circuit 100 is not limited to an optical modulator, and may be an optical filter consisting of a single MZI such as the MZI 91 shown in Fig. 13. In that case, the modulation signal I and the modulation signal Q are not required, and the processing of step S2 is also not required.

[0129] The IQ optical modulator 92 shown in FIG. 15 is a type that is currently widely used as a commercially available product, but a configuration like the IQ optical modulator 400 shown in FIG. 11, which has multiple monitor ports, is also theoretically possible. FIG. 11 is a diagram showing the configuration of the IQ optical modulator 400 with multiple monitor ports. The IQ optical modulator 400 differs from the IQ optical modulator 92 in that it has an I monitor port 301, a Q monitor port 302, and a Ph monitor port 300. The I monitor port 301 branches and outputs the signal multiplexed by the I-side optical multiplexing circuit 21, the Q monitor port 302 branches and outputs the signal multiplexed by the Q-side optical multiplexing circuit 25, and the Ph monitor port 300 branches and outputs the signal multiplexed by the optical multiplexing circuit 6. The I monitor port 301 is connected to a bias voltage V ｂ＿Ｑ , V ｂ＿Ｐｈ , and the Q monitor port 302 is not affected by the bias voltage V ｂ＿Ｉ , V ｂ＿Ｐｈ Since the bias is not affected by the

[0130] This embodiment can also be applied to an IQ optical modulator 400 having a plurality of monitor ports as shown in Fig. 11. Fig. 12 is a diagram showing the configuration of a control circuit in this case. Compared to the control circuit of the second embodiment shown in Fig. 8, the control circuit shown in Fig. 12 needs to newly add an I optical power monitor 303, a Q optical power monitor 304, an I A / D converter 305, and a Q A / D converter 306.

[0131] Optical power monitor 52, I optical power monitor 303, and Q optical power monitor 304 each receive all of the signal light, first reference light, and second reference light. By using the time sharing described in the second embodiment and in FIG. 8, it is possible to monitor the intensity of the signal light, the intensity of the first reference light, and the intensity of the second reference light at I monitor port 301, Q monitor port 302, and Ph monitor port 300, respectively.

[0132] Here, the controller 41 controls the bias voltage V ｂ＿Ｉ The controller 41 measures the reference light vector while shifting only the bias voltage Vｂ＿Ｑ The controller 41 measures the reference light vector while shifting only the bias voltage V ｂ＿Ｐｈ The reference light vector is measured while shifting only the phase of the reference light, and a data table dedicated to Ph is created.

[0133] During the transmission service period, the controller 41 cyclically refers to the three types of data tables and performs feedback control so that the norm of the reference light vector approaches 0 at any monitor port output.

[0134] In the embodiments described so far, the optical path difference ΔL of each MZI is controlled by the bias voltage. However, if an MZI of the type that finely adjusts ΔL using heater heat rather than the Pockels effect is used as the optical circuit 100, control is possible by replacing the bias voltage described in the embodiments so far with bias power. However, since power is not proportional to voltage or current, when shifting the bias power, the controller 41 needs to determine the outputs of the first D / A converter 104, the second D / A converter 105, and the third D / A converter 106 using a conversion formula that takes into account the heater resistance.

[0135] The controller 41 provided in the control circuit of the above-described embodiment can also be realized by a computer and a program, and the program can be recorded on a recording medium or provided via a network.

[0136] According to the above-described embodiment, it is possible to shorten the learning period required for a control circuit that can ditherlessly monitor and compensate for bias drift in an optical circuit composed of a single or multiple MZIs. Also, it is possible to more efficiently utilize the data table created during the learning period.

[0137] According to the embodiment described above, the control unit controls at least one optical path difference among the optical path differences of two optical paths of each of a single or multiple Mach-Zehnder interferometers included in the optical circuit. The control circuit has a control unit, a reference light generation unit, a reference light input unit, an optical power monitor unit, and a normalization unit. The control unit controls the optical path difference of the Mach-Zehnder interferometer included in the optical circuit by one or multiple bias voltages or bias powers. The control unit corresponds to, for example, the controller 41 in the embodiment. Where N is a natural number, the reference light generation unit controls N types of wavelengths λ different from the wavelength λ of the signal light input to the optical circuit. １ ~λ Ｎ The reference light generating unit generates respective reference light beams. The reference light generating unit corresponds, for example, to the first reference light source 35-1 to the Nth reference light source 35-N in the embodiments. The reference light input unit wavelength-multiplexes N types of reference light beams with signal light beams and inputs the result to the optical circuit. The reference light input unit corresponds, for example, to the wavelength division multiplexing coupler 8 in the embodiments. The optical power monitoring unit monitors the optical intensities of the N types of reference light beams after propagation through one or more Mach-Zehnder interferometers included in the optical circuit, or the optical intensities of the N types of reference light beams and the signal light beams. The optical power monitoring unit corresponds, for example, to the optical power monitor 52 and the first reference light power monitor 53-1 to the Nth reference light power monitor 53-N in the embodiments. The normalizing unit normalizes the monitoring results of the optical intensities of the N types of reference light beams, or the monitoring results of the optical intensities of the N types of reference light beams and the signal light beams, obtained by the optical power monitoring unit. The normalizing unit corresponds, for example, to the controller 41 in the embodiments. The reference light vector is an N-dimensional vector having as its components the monitor results of the optical intensities of N types of normalized reference light, or an N+1-dimensional vector having as its components the monitor results of the optical intensities of N types of normalized reference light and the optical intensities of the signal light. The normalization unit performs normalization so that the norm of the reference light vector is zero when all optical path differences of a single or multiple Mach-Zehnder interferometers included in the optical circuit are optimal, and the norm increases as at least one optical path difference deviates from the optimal value. The control unit changes one or more bias voltages or bias powers so that the norm of the reference light vector approaches zero.

[0138] The control unit may change one or more bias voltages or bias powers by feedback control that determines a feedback coefficient based on the norm of the reference light vector.

[0139] During the start-up sequence period of the optical circuit, there may be a learning period in which the control unit shifts one or more bias voltages or bias powers from their optimum values ​​and creates a data table showing the correspondence between the amount of bias voltage shift or the amount of bias power shift and the reference light vector obtained by shifting the amount of shift.

[0140] The control unit may measure how far one or more bias voltages or bias powers deviate from the optimum value by applying dithering to one or more bias voltages or bias powers, and record the measurement results in a data table.

[0141] After creating the data table, the control unit may take a linear sum of the multiple reference light vectors recorded in the created data table to estimate each component of the reference light vector corresponding to one or more shift amounts of bias voltage or bias power that were not recorded in the data table, and add the estimated reference light vector to the data table.

[0142] After the start-up sequence of the optical circuit is completed, the control unit monitors the reference light vector, and if it detects that the norm of the monitored reference light vector is significantly different from zero, it may identify a reference light vector from among the multiple reference light vectors listed in the data table that is closest to the reference light vector at the time of detection, and control the optical path difference based on one or more bias voltage shift amounts or bias power shift amounts recorded in the data table corresponding to the identified reference light vector.

[0143] Although an embodiment of the present invention has been described above in detail with reference to the drawings, the specific configuration is not limited to this embodiment, and includes designs within the scope of the gist of the present invention.

[0144] REFERENCE SIGNS LIST 1 Optical input port 2 Optical branching circuit 3 First optical waveguide 4 Second optical waveguide 5 Bias application electrode 6 Optical multiplexing circuit 7 Optical output port 8 Wavelength multiplexing coupler 10 Optical tap circuit 12 Optical filter 13 Wavelength demultiplexing coupler 20 I-side optical branching circuit 21 I-side optical multiplexing circuit 22 I-side modulation electrode 23 I-side bias application electrode 24 Q-side optical branching circuit 25 Q-side optical multiplexing circuit 26 Q-side modulation electrode 27 Q-side bias application electrode 35-1 First reference light source 35-2 Second reference light source 41 Controller 42 Storage unit 50 In-Phase MZI 51 Quadrature MZI 52 Optical power monitor 53-1 First reference light power monitor 53-2 Second reference light power monitor 91 MZI 92 IQ optical modulator 100 Optical circuit 102 A / D converter 103-1 A / D converter for first reference light 103-2 A / D converter for second reference light 104 First D / A converter 105 Second D / A converter 106 Third D / A converter

Claims

1. A control circuit for controlling at least one optical path difference among the optical path differences of two optical paths each of a single or a plurality of Mach-Zehnder interferometers included in an optical circuit, the control circuit including: a control unit that controls the optical path difference by one or more bias voltages or bias powers; a reference light generation unit that generates reference lights of N types each having a wavelength different from the wavelength of a signal light input to the optical circuit, where N is a natural number; a reference light input unit that wavelength-division multiplexes the N types of reference lights with the signal light and inputs the multiplexed lights to the optical circuit; an optical power monitor unit that monitors the optical intensities of the N types of reference lights after propagation through the single or the plurality of Mach-Zehnder interferometers included in the optical circuit, or monitors the optical intensities of the N types of reference lights and the optical intensity of the signal light; and a normalization unit that normalizes each of the monitoring results of the optical intensities of the N types of reference lights by the optical power monitor unit, or the monitoring results of the optical intensities of the N types of reference lights and the optical intensity of the signal light. When an N-dimensional vector having the monitoring results of the optical intensities of the N types of normalized reference lights as components, or an (N + 1)-dimensional vector having the monitoring results of the optical intensities of the N types of normalized reference lights and the optical intensity of the signal light as components is defined as a reference light vector, the normalization unit performs normalization such that the norm of the reference light vector becomes zero when all the optical path differences of the single or the plurality of Mach-Zehnder interferometers included in the optical circuit are optimal, and the norm increases as at least one optical path difference deviates from the optimal value. The control unit changes one or more of the bias voltages or the bias powers so that the norm of the reference light vector approaches zero. Control circuit.

2. The control circuit according to claim 1, wherein the control unit determines one or more of the bias voltages or the bias powers by feedback control that determines a feedback coefficient based on the norm of the reference light vector.

3. During a startup sequence period of the optical circuit, the control unit shifts one or more of the bias voltages or the bias powers from an optimal value, and creates a data table showing a correspondence relationship between a shift amount of the bias voltage or a shift amount of the bias power and the reference light vector obtained by shifting the shift amount. The control circuit according to claim 1.

4. The control unit measures how far the one or more bias voltages or bias powers deviate from the optimum value by applying dithering to the one or more bias voltages or bias powers, and records the measurement result in the data table. The control circuit according to claim 3.

5. After creating the data table, the control unit estimates each component of the reference optical vector corresponding to the shift amount of one or more bias voltages or bias powers not recorded in the data table by taking the linear sum of the plurality of reference optical vectors recorded in the data table, and adds the estimated reference optical vector to the data table. The control circuit according to claim 3.

6. After the start-up sequence of the optical circuit ends, the control unit monitors the reference optical vector. When it detects that the norm of the monitored reference optical vector is significantly different from zero, it identifies a reference optical vector close to the reference optical vector at the time of detection among the plurality of reference optical vectors listed in the data table, and controls the optical path difference based on the shift amount of one or more bias voltages or the shift amount of the bias power recorded in the data table corresponding to the identified reference optical vector. The control circuit according to claim 3.

7. A control step of controlling at least one of the optical path differences of two optical paths each of a single or a plurality of Mach-Zehnder interferometers included in an optical circuit by one or a plurality of bias voltages or bias powers; a reference light generation step of generating reference lights of N types each having a wavelength different from the wavelength of a signal light input to the optical circuit, where N is a natural number; a reference light input step of wavelength-division multiplexing the N types of reference lights onto the signal light and inputting the multiplexed light into the optical circuit; a monitoring step of monitoring the optical intensities of the N types of reference lights after propagating through the single or the plurality of Mach-Zehnder interferometers the optical circuit has, or the optical intensities of the N types of reference lights and the optical intensity of the signal light; and a normalization step of normalizing each of the monitoring results of the optical intensities of the N types of reference lights by the monitoring step, or the monitoring results of the optical intensities of the N types of reference lights and the optical intensity of the signal light. In the normalization step, when an N-dimensional vector having the monitoring results of the optical intensities of the N types of normalized reference lights as components, or an (N + 1)-dimensional vector having the monitoring results of the optical intensities of the N types of normalized reference lights and the optical intensity of the signal light as components is defined as a reference light vector, normalization is performed such that the norm of the reference light vector becomes zero when all of the optical path differences of the single or the plurality of Mach-Zehnder interferometers included in the optical circuit are optimal, and the norm increases as at least one of the optical path differences deviates from the optimal value. In the control step, one or a plurality of the bias voltages or the bias powers are changed so that the norm of the reference light vector approaches zero. An optical circuit control method.

Images