Optical transmitter

Abstract

Disclosed is an optical transmitter that makes it possible to simultaneously realize the expansion of a limited band of a DAC and a simplified configuration. In the optical transmitter, the band expansion of an electrical region is limited to the double thereof, the result is combined with a band expansion of N times an optical region, and it is possible to omit multiplication clock circuits, large optical filters, and the like by means of specific digital signal processing. As a result of the simplified configuration, it is possible to provide an optical transmitter whereby a discretionary optical signal of which the band is extended 2N times beyond a limited DAC operating bandwidth is output. The specific digital signal processing includes filtering the following by a filter coefficient corresponding to the configuration of the optical transmitter: a digital division signal obtained by band division of a target signal to be generated; and a folded digital division signal of the digital division signal. The band expansion of the optical region can be flexibly expanded by means of the number of branches in an optical modulation unit. The present invention can also be easily applied to the generation of polarization-multiplexed light.

Description

Optical Transmitter

[0001] The present invention relates to an optical transmitter for optical fiber communications.

[0002] To meet the growing demand for communications, active research is being conducted to increase the speed of optical fiber communication systems. Optical transmitters for digital coherent communication systems used in medium- to long-distance communications employ one polarization-multiplexed optical modulator per transmitting laser. Separate IQ optical modulators are provided for two orthogonal polarizations, and each IQ optical modulator applies modulation signals to the in-phase (I) and quadrature (Q) components of the carrier light, respectively, creating a two-dimensional (two) modulation space. A total of four-dimensional modulation space is utilized with one polarization-multiplexed IQ optical modulator.

[0003] The I and Q components of the carrier light are modulated by a multi-level baseband electrical signal output from a digital-to-analog converter (DAC). One DAC is used per modulation dimension, and four DACs are used per polarization multiplexed optical transmitter. If the electro-optic (EO) bandwidth of the driver amplifier driving the optical modulator and the optical modulator itself is sufficiently large, the bandwidth of the optical signal that the optical transmitter can output is twice the bandwidth of the electrical signal from the DAC. Here, the bandwidth of the electrical signal is the bandwidth of the positive frequency component, and the analog bandwidth of the DAC is an important factor in determining the transmission data rate per optical transceiver.

[0004] To improve the data rate of optical transmitters, the analog bandwidth of DACs themselves is being expanded. In addition, in optical modulators, technologies that use multiple DACs for one modulation space to generate desired optical signals by signal synthesis in the analog domain are being actively investigated.

[0005] Patent Document 1 discloses a signal generator that uses multiple DACs and high-speed analog devices to generate a signal with a wider bandwidth and higher quality than the output bandwidth of a single DAC. ＢThe digital signal processing unit includes M DACs having M×f Ｂ It is possible to generate a wideband signal of such high quality. The band-extended electrical signal from this signal generator is input to an IQ optical modulator, resulting in a band-extended optical transmitter. In the following description, this function of extending the band of the modulated signal beyond the band of the DAC alone is referred to as "M-fold band extension," and the value corresponding to M is referred to as "parallel number." The band extension in Patent Document 1 is band extension in the electrical domain.

[0006] A method for achieving functionality equivalent to the high-speed analog device described in Patent Document 1 in the optical domain using an optical comb and an optical filter is also widely known. In the optical transmitter described in Non-Patent Document 1, the frequency components of an optical comb are extracted one by one using an optical filter, and each frequency component is driven by a different IQ optical modulator. A method for generating a wideband optical signal by combining the output light from multiple IQ optical modulators again using an optical filter is also described. In this method, an optical filter with a sufficiently high extinction ratio is used to extract the frequency components of the optical comb. Because the baseband electrical signals modulating each IQ optical modulator are treated as independent spectral slices, the 2M-input, M-output filter described in Patent Document 1 is not required. The bandwidth extension described in Non-Patent Document 1 is an optical domain bandwidth extension.

[0007] International Publication No. 2020 / 054173 International Publication No. 2022 / 244115

[0008] H. Mardoyan et al., “Transmission of Single-Carrier Nyquist-Shaped 1-Tb / s Line-Rate Signal over 3,000 km,” in proc. OFC2015, paper W3G.2.

[0009] However, both of the above-mentioned electrical and optical bandwidth extension technologies have drawbacks in terms of the feasibility of related components, device size, and complexity, making it difficult to achieve wider bandwidth extension.

[0010] One aspect of the present invention is an optical transmitter comprising: a clock source that outputs a clock signal with a frequency of B / (2N) (N is an integer of 2 or more); an optical comb generator that generates an optical comb by modulating continuous wave light (CW light) with the clock signal; a digital signal processing unit; 4N digital-to-analog converters (DACs), each outputting an electrical signal having a bandwidth of B / (2N); 2N electrical signal generation units driven by the clock signal, each receiving two of the electrical signals as input and outputting an electrical signal whose bandwidth is doubled; an optical branching unit that branches the optical comb into N optical paths; N optical IQ modulators corresponding to the N branched optical combs, each using two of the 2N bandwidth-extended electrical signals as drive signals and outputting modulated optical signals; and an optical modulation unit having an optical delay circuit for receiving the modulated optical signals from the N optical IQ modulation units, and an optical coupler for combining the modulated optical signals from the N optical IQ modulation units, wherein the digital signal processing unit has a target signal generation unit for virtually generating a target signal having a bandwidth of 2B, a spectrum division unit for dividing the target signal into 4N signals having a bandwidth of B / (2N) and generating 4N digital divided signals corresponding to signals obtained by frequency-shifting each of the divided signals to baseband, a folding unit for obtaining 4N folded digital divided signals by inverting each of the digital divided signals on the frequency axis around a frequency B / (4N) and taking the complex conjugate, and an 8N x 4N filter unit that receives the 4N digital divided signals and the 4N folded digital divided signals as inputs and outputs corresponding digital signals to the 4N DACs.

[0011] According to the present invention, it is possible to provide an optical transmitter that has a bandwidth expansion function of four times or more and has a simple configuration.

[0012] FIG. 1 is a diagram illustrating the configuration of an optical transmitter according to a first embodiment. FIG. 2 is a diagram illustrating a first configuration example of an electrical signal generating unit. FIG. 3 is a diagram illustrating a third configuration example of an electrical signal generating unit. FIG. 4 is a diagram illustrating the spectrum division operation of a digital signal processing unit. FIG. 5 is a diagram illustrating the band expansion operation of an electrical signal generating unit. FIG. 6 is a diagram illustrating the convolution of an electrical signal in an optical IQ modulator. FIG. 7 is a list illustrating the definitions of signals, etc., in an optical transmitter according to the present disclosure. FIG. 8 is a diagram illustrating the configuration of an optical transmitter according to a second embodiment. FIG. 9 is a diagram illustrating the configuration of an optical transmitter according to a third embodiment. FIG. 10 is a diagram illustrating the operation of an optical comb generator suitable for the optical transmitter according to the present disclosure. FIG. 11 is a diagram illustrating the configuration of an optical comb generator suitable for the optical transmitter according to the present disclosure.

[0013] The optical transmitter disclosed herein combines the features of bandwidth expansion in both the electrical and optical domains, and presents a new configuration that overcomes the drawbacks of conventional bandwidth expansion techniques. While the optical transmitter includes a doubled bandwidth expansion in the electrical domain and an N-fold bandwidth expansion in the optical domain using an optical modulation unit with N branched paths, the clock circuit and optical filter in the optical transmitter can be significantly simplified compared to conventional configurations. The following description will first provide an overview of the challenges involved in bandwidth expansion in conventional optical transmitters, and then describe the configuration and operation of the optical transmitter disclosed herein.

[0014] In the following description, the analog bandwidth of a DAC refers to the upper limit frequency of an analog signal that the DAC can output without significant signal degradation. Typically, the frequency at which the strength of the analog signal output from the DAC is attenuated by a certain value compared to near DC is defined as the analog bandwidth of the DAC. The amount of reduction in signal strength that defines the analog bandwidth of the DAC is set to an arbitrary value depending on the spectral shape of the signal to be generated, the characteristics of the transmitting and receiving device, etc. Typically, the analog bandwidth of the DAC is defined using a threshold of approximately 3 to 6 dB, and at most approximately 20 dB, reduction from the signal strength near DC.

[0015] In both the electrical domain (Patent Document 1) and the optical domain (Non-Patent Document 1 and Patent Document 2), multiple DACs can be used to generate a desired optical signal by combining analog signals. However, both of these conventional configurations have the following problems.

[0016] When the bandwidth is expanded in the electrical domain, the overall configuration of the optical transmitter is simple, but it is difficult to increase the number of parallel circuits and the output bandwidth compared to when the bandwidth is expanded in the optical domain. When implementing a wideband analog signal generation function in the electrical domain as in Patent Document 1, components such as analog multiplexers, mixers, and combiners can be integrated within an analog IC. Furthermore, the driver amplifier that drives the optical modulator can also be integrated, allowing the main components of the optical transmitter to be implemented compactly. However, when the bandwidth is three times or more the bandwidth of the DAC alone, i.e., when the number of parallel circuits is three or more, a multiplication clock is required (Figures 11, 12, 17, and 18 of Patent Document 1). The clock that drives the analog signal generation unit has a basic frequency f ｃｌｋ = f Ｂ Not only the clock, but also the 2nd floor ｃｌｋ , 3f ｃｌｋ A multiplied clock such as the above is required.

[0017] Usually, the fundamental frequency f ｃｌｋ = f Ｂ is set to a frequency close to the upper limit frequency that the DAC can output. Generating a clock that is twice or three times the fundamental frequency is difficult, and the loss caused by the wiring pattern is large, making implementation difficult. The bandwidth after expansion is also limited by the analog signal generation unit and its output wiring.

[0018] While it is easy to increase the number of parallel connections and output bandwidth in optical bandwidth expansion, the optical transmitter configuration becomes complex. In optical transmitters using an optical comb and optical filters, as in Non-Patent Document 1, an optical comb is generally generated by clock-modulating continuous wave (CW) light with an external modulator. The number of frequency components in the optical comb can be easily increased by increasing the amplitude of the driving clock signal, cascading multiple external modulators, or adding a nonlinear medium in the downstream stage. However, as the number of frequency components increases, the number of optical modulators also increases, requiring driver amplifiers and optical phase control mechanisms for the optical modulators. Bandwidth expansion in the optical domain also poses the problem of the overall size and complexity of the optical transmitter.

[0019] Furthermore, separating the frequency components of an optical comb with a high extinction ratio generally requires a large optical filter, which increases the overall size of the optical transmitter. Furthermore, the operation of the optical filter depends not only on the frequency spacing of the optical comb but also on the absolute frequency of the light source. This necessitates adjusting the operating point of the optical filter to match the light source frequency, further complicating control.

[0020] The present invention has been made in view of the above problems, and provides an optical transmitter having a bandwidth expansion function of four times or more and a simple configuration.

[0021] The material for forming the optical comb generator and optical IQ modulator described later is, for example, LiNbO, which has the Pockels effect, which is a type of electro-optic effect. ３ Multi-component oxide crystals such as GaN (LiN), and GaAs-based and InP-based compound semiconductors capable of refractive index modulation by the Pockels effect and the quantum confined Stark effect (QCSE) can be used. Furthermore, Si and SiGe semiconductors with pn junctions capable of refractive index modulation by the carrier plasma effect, and polymers with the electro-optical effect such as chromophores can also be used. The optical transmitter disclosed below has novel features in its functional block configuration, and the effects obtained do not depend on the material of the optical circuit that implements the optical modulation section.

[0022] The optical transmitter disclosed herein combines N-fold bandwidth expansion in the optical domain while limiting bandwidth expansion in the electrical domain to 2x, and uses unique digital signal processing to eliminate the need for a multiplication clock circuit or a large optical filter. The present invention provides an optical transmitter with a simplified configuration that outputs an arbitrary optical signal whose bandwidth is expanded 2N times, exceeding the limited operating bandwidth of a DAC. The unique digital signal processing includes filtering digitally divided signals obtained by band-dividing a target signal to be generated and their folded digitally divided signals using filter coefficients appropriate for the optical transmitter configuration. The bandwidth expansion in the optical domain can be flexibly expanded by adjusting the number of branches in the optical modulation section. The present invention can also be easily applied to the generation of polarization-multiplexed light.

[0023] [First Embodiment] Fig. 1 is a diagram showing the configuration of an optical transmitter according to the first embodiment. The optical transmitter 1000 achieves a bandwidth expansion function (number of parallel circuits = 4) that is four times the bandwidth of a single DAC. The optical transmitter 1000 includes a digital signal processing unit 1100, eight DACs 1201 to 1208, and four electrical signal generation units 1301 to 1304, and outputs a bandwidth-expanded electrical signal 102. The digital signal processing unit 1100 includes an encoding / mapping unit 1110, a target signal generation unit 1121, a spectrum splitter 1131, a folding unit 1141, and a 16x8 filter 1151.

[0024] The electrical signal generating units 1301 to 1304 correspond to the analog broadband electrical signal generating units in the prior art (Patent Document 1), and generate a wideband electrical signal with the band expanded from, for example, two narrowband electrical signals.

[0025] The optical transmitter 1000 further includes a transmission laser 1400, an optical comb generator 1500, an optical modulation unit 1600, and a clock source 1700. The optical modulation unit 1600 has an optical branching unit 1610 that branches the generated optical comb into two optical paths, two optical IQ modulators 1621 and 1622, two optical delay units 1631 and 1632, and an optical coupler 1641. The electrical signal generation units 1301 to 1304 and the optical comb generator 1500 are driven by a clock with a frequency of B / 4 sent from the clock source 1700.

[0026] The functions of each block included in the digital signal processing unit 1100 are implemented as part of a digital signal processor (DSP). The DACs 1201 to 1208 may be built into the DSP or may be implemented as separate integrated circuits. The electrical signal generating units 1301 to 1304 that output the wideband analog electrical signal, i.e., the band-extended electrical signal 102, may have various configurations.

[0027] 2 is a diagram showing a first example of the configuration of the electrical signal generating unit 1300-1. The electrical signal generating unit 1300-1 is configured using an analog multiplexer 211, and is supplied with a clock having a frequency of B / 4 from a clock source 1700.

[0028] 3 is a diagram showing a second example of the configuration of the electrical signal generating unit 1300-2. The electrical signal generating unit 1300-2 is composed of a mixer 321 and a combiner 331, and is supplied with a clock having a frequency of B / 4 from a clock source 1700.

[0029] 4 is a diagram showing a third example of the configuration of the electrical signal generating unit. The electrical signal generating unit 1300-3 has an IQ modulator type configuration consisting of mixers 421 and 422, a combiner 431, and a 90-degree phase shifter 441. A clock having a frequency of B / 4 is supplied from a clock source 1700.

[0030] The wiring lengths from the DACs 1201 to 1208, via the electrical signal generation unit, to the optical IQ modulators 1621 and 1622 are basically all adjusted to be equal lengths. In other words, the wiring lengths of the wiring patterns on the integrated IC and the PCB or other substrate on which the IC is mounted are designed to be equal lengths among the eight electrical paths.

[0031] The oscillation frequency of the transmitting laser 1400 is f ｃ The optical comb generator 1500 can be a linear optical phase modulator, a Mach-Zehnder optical modulator, or a cascade connection of these. Furthermore, the optical comb generator 1500 may be configured to include a nonlinear optical element after the optical modulator to increase the number of frequency components and increase the comb bandwidth.

[0032] The optical modulation unit 1600 may be formed by integrally forming all of its components on a single substrate, by directly connecting multiple substrates, or by partially using spatial optical system components. However, including a fiber connection within the optical modulation unit 1600 is undesirable because it makes the optical phase unstable. The transmitting laser 1400 may be provided separately from the optical transmitter and supply CW light to the optical comb generator 1500.

[0033] A normal IQ modulation circuit of two parallel Mach-Zehnder modulators can be used as the optical IQ modulators 1621 and 1622. The optical delay units 1631 and 1632 impart a time difference τ between the optical signals of the two paths in the optical modulation unit 1600. That is, the optical delay units 1631 and 1632 impart a time delay τ to the optical signal that passes from the optical branching unit 1610 to the optical IQ modulator 1621 and then to the optical coupler 1641, and then to the optical IQ modulator 1622 and then to the optical coupler 1641, respectively. Specifically, it is desirable that τ be approximately 1 / (2B).

[0034] The output optical signal 103 of the optical transmitter 1000 has an optical frequency f ｃ It has a signal spectrum ranging from ±B, and its signal bandwidth is approximately 2B. As will be described later, DACs 1201 to 1208 each have a bandwidth of B / 4, so that optical transmitter 1000 in Fig. 1 has a four-fold bandwidth expansion function. An outline of the operation of the bandwidth expansion in optical transmitter 1000 will be described below.

[0035] In the optical transmitter 1000 of the first embodiment, the bandwidth is doubled in the electrical domain and then doubled again in the optical domain. That is, narrowband electrical signals 101 with a bandwidth of approximately B / 4 from each of the DACs 1201 to 1208 are doubled in the electrical domain by the electrical signal generation units 1301 to 1304. The doubled-bandwidth electrical signal 102 is then doubled in the optical domain by the optical modulation unit 1600, and an output optical signal 103 is obtained.

[0036] As will be described later, output optical signal 103, whose bandwidth has been expanded fourfold, is a signal in which the output signals from DACs 1201 to 1208 overlap in the frequency domain. The manner in which the frequency domains overlap is determined by the configurations and operating conditions of electrical signal generators 1301 to 1304, optical comb generator 1500, and optical delay units 1631 to 1632. By performing appropriate arithmetic processing in digital signal processor 1100 according to the manner in which the frequency domains of the outputs from the multiple DACs overlap and the manner in which they overlap, which are determined by the hardware configuration of the optical transmitter, it is possible to generate any modulated optical signal within the fourfold expanded bandwidth.

[0037] Therefore, to actually fabricate the optical transmitter 1000, first determine the specific configurations and operating conditions of the electrical signal generators 1301 to 1304, the optical comb generator 1500, and the optical delay units 1631 to 1632. Then, determine the parameters used in the digital signal processor 1100 in accordance with these configurations and operating conditions. Specifically, as will be described later, it is necessary to follow the procedure of determining the coefficients of the 16×8 filter 1151.

[0038] The function of the digital signal processing unit 1100 is basically implemented as part of the function of the DSP. The detailed operating principle of the optical transmitter 1000 and the procedure for determining the coefficients used in the filters in the digital signal processing unit will be described below. In the following description, the frequency domain spectral structures of the electrical signals and optical signals in each section of the optical transmitter 1000 will be described with reference to FIGS. 5 to 7. Furthermore, the names of signals and their definitions explained in the operating principle and the procedure for determining the filter coefficients are listed in the table of FIG. 8. While the following description will not provide a complete explanation of the definitions in FIG. 8, please refer to the definitions of each signal in the table of FIG. 8 as appropriate when a signal name is mentioned for the first time.

[0039] As already mentioned, each sub-block of the digital signal processing unit 1100 in FIG. 1 represents a signal processing function that is performed by calculation.

[0040] In the following explanation, we first explain the configuration and operation of an optical transmitter that is simpler than the prior art and capable of bandwidth expansion with a large number of parallel connections, assuming that the filter coefficients of digital signal processing unit 1100 have already been determined, and then explain the procedure for determining the coefficients used in the filter of digital signal processing unit 1100.

[0041] In the digital signal processing unit 1100, the encoding / mapping unit 1110 performs FEC encoding on the input transmission bit sequence, and then maps the encoded bit sequence to a complex symbol sequence. A known training symbol may be inserted into the symbol sequence as needed. In the optical transmitter of this embodiment, the complex symbol sequence does not necessarily have to be a single-carrier sequence, but may be a multi-carrier sequence in which multiple symbol sequences are allocated to multiple RF subcarriers. The signal bandwidth can also be any value equal to or less than 2B.

[0042] The target signal generator 1121 virtually generates a desired signal waveform U around the optical carrier frequency, which is ultimately output Z103 from the optical transmitter 1000, in response to the input complex symbol sequence. For example, when a single-carrier signal is used as the target signal, a desired pulse waveform can be simply convolved with one symbol sequence received from the encoding and mapping unit 1110. When a multi-carrier signal is used as the target signal, a desired pulse waveform can be convolved with each of multiple symbol sequences received from the encoding and mapping unit 1110, and then the resulting signals can be added. However, the spectral components of the target signal must be generally concentrated within ±B. Therefore, the symbol rate for a single-carrier signal must basically be 2B or less.

[0043] 5 is a diagram illustrating the spectrum division operation of the digital signal processing unit. In FIG. 5, the upper part shows the spectrum U of the target signal. ｔｏｔ (f) is the target signal spectrum U １ (f) ~U ８ The spectrum of the target signal sent from the target signal generator 1121 is expressed as U ｔｏｔ (f), the optical carrier frequency is f ｃ When the spectrum 500 of the optical signal to be generated is expressed in complex notation as shown in the upper part of FIG. ｔｏｔ (f-f ｃ ) The spectrum splitter 1131 can be expressed as U ｔｏｔ (f-f ｃ ) to f ｃThe spectrum divider virtually divides the spectrum 501-1 to 501-8 into eight (4 x 2) parts each having a width of approximately B / 4 each around the spectrum U. The spectrum divider further virtually shifts each of the divided spectra 501-1 to 501-8 to the baseband. As a result, as shown in the lower part of FIG. １ (f) ~U ８ Eight digital split signals 511 to 518 represented by (f) are output. These spectra U １ (f) ~U ８ (f) will be summarized as follows:

[0044] As shown in Figure 5, the relationship between the spectra before and after division is Ｘｔｏｔ (f-f ｃ ) = U １ (f + B - f ｃ ) + U ２ (f+3B / 4-f ｃ ) + U ３ (f+B / 2-f ｃ ) + U ４ (f + B / 4 - f ｃ ) + U ５ (f-f ｃ ) + U ６ (f-B / 4-f ｃ ) + U ７ (f-B / 2-f ｃ ) + U ８ (f-3B / 4-f ｃ ) holds true. Each of the digital split signals has a bandwidth of B / 4.

[0045] The folding unit 1141 converts the digital divided signal U １ (f) ~U ８ (f) is inverted around B / 8 on the frequency axis to take the complex conjugate, and eight folded digital divided signals are output. The spectrum of the eight folded digital divided signals is expressed as follows using the tilde symbol, where the superscript asterisk represents the complex conjugate.

[0046] The eight digital divided signals and eight folded digital divided signals are input to a 16x8 filter 1151, which outputs eight signals. Here, 16x8 means 16 inputs and 8 outputs. The coefficients of the 16x8 filter 1150 are expressed as an 8x16 matrix, as will be described later. How the filter coefficients of the 16x8 filter 1150 are determined will be described later. The eight output signals of the 16x8 filter 1150 are sent to DACs 1201 to 1208, respectively, and output from each DAC as narrowband analog electrical signals 101(V) with a bandwidth of approximately B / 4.

[0047] 6 is a diagram illustrating the band expansion operation of the electrical signal generation unit. FIG. 6(a) shows the spectrum of the input signal V to the electrical signal generation unit, and FIG. 6(b) shows the spectrum of the output signal W from the electrical signal generation unit. The spectrum of each of the electrical signals 101 output from the DACs 1201 to 1208 is considered to be divided into two, a positive frequency side and a negative frequency side, as shown in FIG. 6(a). Since the electrical signal 101 from the DAC is a real signal, the negative frequency side component of the spectrum is equal to the complex conjugate obtained by inverting the sign of the frequency of the positive frequency side component. For example, if the positive frequency side component of the spectrum of the first analog electrical signal output from DAC 1201 is V １ If (f), the negative frequency component can be expressed as follows using the tilde symbol mentioned above:

[0048] The second to eighth analog electrical signals output from the DACs 1202 to 1208 can also be expressed in the same manner as the above equations.

[0049] Each of the electrical signal generators 1301-1304 receives as input two of the eight narrowband electrical signals 101 from the DACs 1202-1208. The electrical signal generators widen the bandwidth of the output electrical signal 102 by generating images of their input signals around a clock signal with a frequency of B / 4. The electrical signal generators are driven by a clock with a frequency of B / 4, with the input electrical signal 101 having a bandwidth of approximately B / 4 and the output electrical signal 102 having a bandwidth of B / 2. The relative amplitudes and phases of the images, and between the images and the original input signal, depend on the configuration of the electrical signal generators 1301-1304 (see FIGS. 2-4).

[0050] The wideband analog electrical signal 610 obtained from the electrical signal generating unit shown in Fig. 6B is divided into a positive frequency side and a negative frequency side, and each of these is further divided into two bands with a bandwidth of approximately B / 4. Specifically, the divided components 611 and 613 of the positive frequency side components of the spectrum of the first analog electrical signal output from the electrical signal generating unit 1301 are divided into W １ (f), W ２ The divided components of the positive frequency side components of the spectrum of the second analog electrical signal output from the electrical signal generating unit 1302 are expressed as W (f−B / 4) from the side close to DC. ３ (f), W ４ Similarly, the divided components of the positive frequency side components of the spectrum of the third analog electrical signal output from the electrical signal generating unit 1303 are expressed as W ５ (f), W ６ (f−B / 4), and the divided components of the positive frequency side components of the spectrum of the fourth analog electrical signal output from the electrical signal generating unit 1304 are expressed as W ７ (f), W ８ This is expressed as (f-B / 4).

[0051] In FIG. 6B, the electrical output of the electrical signal generating unit 1301 includes divided components 611 and 613 of the positive frequency side component, i.e., W １ (f), W ２The negative frequency components 612 and 614 corresponding to the positive frequency components 611 and 613 can be expressed as follows, starting from the side closest to DC:

[0052] The second to fourth analog electrical signals output from the electrical signal generating units 1302 to 1304 can be expressed in a similar manner. １ (f), W ２ (f) is V １ (f), V ２ Using (f) and a response matrix R of 2 rows and 4 columns determined by the configuration of the electrical signal generating unit 1301, it can be expressed as follows:

[0053] In the above equation, the superscript [1] in the R matrix indicates that the response corresponds to the electrical signal generator 1301. If the four electrical signal generators 1301 to 1304 were configured with identical components with no individual differences, no response differences would appear due to the superscripts [1] to [4]. Therefore, the R rows of the four electrical signal generators can be considered to be identical.

[0054] Using the relationship between W, R, and V in the above equation, W is calculated for all four electrical signal generating units. １ (f) ~ W ８ (f) and V １ (f) to V ８ The relationship in (f) can be summarized as follows: As also defined in the table of FIG. 8, the response matrix R Ｒ (f) (bold) and R Ｌ (f) (bold) is the left and right halves of the response matrix representing the transformation from V to W. Also, since it is not possible to highlight bold in this specification, when (bold) is shown at the first mention of a parameter, it represents a vector (matrix) with elements.

[0055] Response matrix R Ｒ (f) and R ＬThe element (f) includes the response of the wiring between the DACs 1201 to 1208 and the electrical signal generators 1301 to 1304, as well as the response and individual differences of the DACs 1201 to 1208 and the electrical signal generators 1301 to 1304 themselves.

[0056] The optical comb generator 1500 in FIG. 1 is driven by a clock with a frequency of B / 4, and f ｃ An optical comb having frequency components at ±kB / 4 (k=0, 1, 2, ...) is generated. The generated optical comb is sent to the optical modulation unit 1600. Here, the complex amplitude of the frequency components corresponding to k=0, 1, 2, ... in the optical comb is expressed as a ｋ The optical comb does not need to be flat over the entire band, but the absolute value of the amplitude of the second-order component, i.e., |a ±２ As a guideline, adjust the frequency where the absolute value of the amplitude is maximum. ｋ |a ｍａｘ When this is the case, |a ±２ |≧a ｍａｘ It is desirable that the ratio be about 1 / 2.

[0057] The optical modulation unit 1600 receives the band-expanded electrical signal 102 as a baseband signal and an optical comb as carrier light. The input optical comb is branched into two by the optical branch 1610, and the branched optical combs are modulated by the band-expanded electrical signals in the optical IQ modulators 1621 and 1622. Specifically, in the optical IQ modulator 1621, the I and Q components of the optical comb are modulated by the band-expanded electrical signals output from the electrical signal generators 1301 and 1302, respectively. Similarly, in the optical IQ modulator 1622, the I and Q components of the optical comb are modulated by the band-expanded electrical signals output from the electrical signal generators 1303 and 1304, respectively. When the modulation operation in the optical IQ modulator is viewed in the frequency domain, the band-expanded electrical signals are convolved with the respective frequency components of the optical comb in each optical IQ modulator, resulting in a band-expanded spectrum.

[0058] FIG. 7 is a diagram showing a schematic diagram of convolution of an electrical signal in an optical IQ modulator. For simplicity, it is assumed that the optical IQ modulator is ideal, with a flat EO frequency response and zero chirp in each of the I and Q modulation sections. FIG. 7 shows only the convolution of the I component signal, for example, within the modulation operation of the optical IQ modulator. The upper part of FIG. 7 shows the spectrum of an exemplary optical comb 710 from the optical comb generator 1500, including up to the fifth order components (a ±５ A linear phase modulator is used as the optical comb generator 1500, and the half-wave voltage is V π , the peak-to-peak voltage V of the drive signal ｐｐ About 2V π This assumes a case where the

[0059] Below the spectrum of the optical comb 710, the components of the band-extended electrical signal that are superimposed on each frequency component of the optical comb and output are shown. For example, the frequency f ｃ Similarly, a spectrum 700 is generated in which the band-extended electrical signal is convolved with the frequency f ｃ +B / 4, spectrum 702 is generated, and the frequency f ｃ +B / 2, spectrum 704 is generated. ｃ For −B / 4, spectrum 701 is generated, and the frequency f ｃ For −B / 2, spectrum 703 is produced.

[0060] In FIG. 7, the variable part is omitted to avoid the notation becoming complicated. ｃ +kB / 4~f ｃ The variable part of the dotted spectral component in the region of +(k+1)B / 4 is (f-kB / 4). －２ W １ Haa －２ W １ (f+B / 2), notation a ＋２ W ２ Haa ＋２ W ２ As can be seen from FIG. 7, the absolute amplitude value |a±２ If | is large enough, then f ｃ -B to f ｃ There will be some output optical signal components from the optical IQ modulator in the +B frequency range.

[0061] Although FIG. 7 shows only the convolution of the I component signal in the output of one optical IQ modulator 1621, when showing the convolution of the Q component of the optical IQ modulator 1621, the W １ , W ２ respectively. ３ , W ４ and multiplying the whole by the imaginary unit j. In addition, when expressing the convolution of the I and Q components of the output of another optical IQ modulator 1622, W １ , W ２ , W ３ , W ４ respectively. ５ , W ６ , W ７ , W ８ The figure is obtained by replacing

[0062] As described above, when the delays provided by the optical delay units 1631 and 1632 are taken into consideration in addition to the spectra of the I and Q components of the two optical IQ modulators, the output optical signal 103 is finally obtained from the optical coupler 1641. The output optical signal 103 of the optical transmitter 1000 and the band-expanded electrical signal 102, i.e., W １ (f) ~ W ８ (f) can be obtained.

[0063] Regarding the delay, the time delay of τ is equivalent to multiplying the spectrum by exp(-2πfτ) in the frequency domain. ｃ -B to f ｃ +B range is set to f following the example of the target signal in FIG. ｃ The spectrum is virtually divided into eight parts each having a width of approximately B / 4 around Z. The divided spectrum is virtually shifted to the baseband, and the resulting spectrum is then １ (f) to Z ８In this case, taking into consideration the spectrum obtained by convolution of the electrical signal in the optical IQ modulator in FIG. 7, the substitution for Z, and the delay, the following relationship is obtained:

[0064] The Hadamard product represents an operation of taking the element-by-element product of matrices. ＋ (bold) and A － (bold) are coefficients for the positive frequency component and negative frequency component of the band-extended electrical signal 102, respectively, and D(f) (bold) represents the delay imparted by the optical delay units 1631 and 1632. P(f) (bold) is an 8x16 matrix. In the above explanation, an ideal optical modulator was assumed, so in equation (8), all columns other than those containing D(f) are frequency-independent. In an actual optical transmitter, all elements of P(f) are frequency-dependent, reflecting the difference in characteristics between the two optical IQ modulators, the difference in characteristics between the IQ components of each optical IQ modulator, the EO response characteristics, and the response characteristics of other parts of the optical modulation unit 1600.

[0065] Up to this point, the operation of the entire optical transmitter has been described assuming that the filter coefficients of filter 1150 have been determined. To operate as an optical transmitter capable of outputting any optical modulated signal, the filter coefficients corresponding to the configuration of the optical transmitter must be determined and available. In other words, the filter coefficients must be available for the arithmetic processing performed by digital signal processing unit 1100. For example, the filter coefficients may be stored in memory. The filter coefficients of filter 1150 can be determined as follows.

[0066] Based on the above equations (1) to (9), the output V(f) (bold) of the electrical signal 101 from the DACs 1201 to 1208 and the f of the optical output signal 103 of the optical transmitter are ｃ -B to f ｃThe relationship between V(f) and the spectral component Z(f) (bold) in the frequency range of +B can be analytically expressed. Then, by back-calculating from the relationship between V(f) and Z(f), it is possible to determine the filter coefficients of the filter 1150 in the digital signal processing unit that are required to obtain the desired signal as Z(f). In other words, if the filter coefficients of the 16×8 filter 1150 are determined so that Z(f) = U(f) for the target signal U(f) from the target signal generator, a four-fold bandwidth extension can be achieved in the optical transmitter 1000.

[0067] The following describes how to determine the filter coefficients of the 16×8 filter 1150. In equations (2) and (6), if Z(f)=U(f) and the relationship between V(f) (bold) and U(f) (bold) is found, the following equation is obtained: Furthermore, the following equation is obtained from equation (11):

[0068] Here, G(f) is a 16x16 matrix, and the upper half of G(f) is an 8x16 matrix, Ｕ (f) (boldface) gives the following relationship:

[0069] In formula (15), G Ｕ Note that the subscript U in (f) stands for upper, not for the target signal U(f). ｕ (f) is a matrix for inversely calculating the output signal V(f) of the DACs 1201 to 1208 from the target signal U(f), and the filter coefficients of the 16×8 filter 1150 are G Ｕ (f) can be used.

[0070] In fact, the filter coefficient G Ｕ To find (f) (bold), first Ｕ After giving an appropriate initial value such as an identity matrix to (f), a known signal is transmitted from the DAC as V(f) (bold). An optical signal is received from the optical transmitter, and the received signal Z(f) is converted to H Ｌ (f) (bold), H Ｒ(f) (bold) is sufficient. Ｌ (f), H Ｒ (f) is basically a response specific to the configuration of the analog log signal generating unit, optical comb generator 1500, and optical delay unit of the optical transmitter, and does not depend on V(f).

[0071] The optical transmitter 1000 disclosed herein combines a 2x bandwidth expansion in the electrical domain with an Nx (2x in this embodiment) bandwidth expansion in the optical domain, thereby eliminating the need for multiplied clocks and optical filters that were necessary in conventional technology, and simplifying the overall configuration of the optical transmitter.

[0072] Therefore, the optical transmitter 1000 of the present disclosure includes a clock source 1700 that outputs a clock signal with a frequency of B / (2N) (N is an integer equal to or greater than 2), an optical comb generator 1500 that generates an optical comb by modulating continuous wave light (CW light) with the clock signal, a digital signal processing unit 1100, 4N digital-to-analog converters (DACs) 1201 to 1208, each of which outputs an electrical signal having a bandwidth of B / (2N), and 2N digital-to-analog converters (DACs) driven by the clock signal. The electrical signal generation units 13-1 to 1304 each receive two of the electrical signals as input and output an electrical signal whose bandwidth has been doubled; an optical branching unit 1610 that branches the optical comb into N optical paths; N optical IQ modulators 1621 and 1622 that correspond to the N branched optical combs and each use two of the 2N electrical signals 102 whose bandwidth has been expanded as a drive signal and output a modulated optical signal; and an optical modulation unit 1600 having optical delays 1631, 1632 that provide delays, and an optical combiner 1641 that combines the modulated optical signals from the N optical IQ modulation units. The digital signal processing unit 1100 can be implemented as having a target signal generation unit 1121 that virtually generates a target signal having a bandwidth of 2B, a spectrum division unit 1131 that divides the target signal into 4N signals having a bandwidth of B / (2N) and generates 4N digital divided signals that correspond to signals obtained by frequency-shifting each of the divided signals to baseband, a folding unit 1141 that obtains 4N folded digital divided signals by inverting each of the digital divided signals on the frequency axis around a frequency B / (4N) and taking the complex conjugate, and an 8N×4N filter unit 1151 that receives the 4N digital divided signals and the 4N folded digital divided signals as inputs and outputs corresponding digital signals to the 4N DACs.

[0073] The optical transmitter 100 shown in FIG. 1 is a functional block diagram and does not necessarily reflect the physical layout of the actual components. For example, the electrical signal generators 1301-1304 are drawn close to the DACs 1201-1208 for clarity. In practice, it is desirable to place the electrical signal generators 1301-1304 and the IQ modulators 1621 and 1622 as close as possible to minimize the transmission distance of the broadband analog signal 102(W). Furthermore, to consolidate the functions of the analog electronic devices, it is desirable to integrate the driver amplifier arrays for the IQ modulators 1621 and 1622 and the electrical signal generators 1301-1304, consolidate and integrate the functions of the analog electronic devices, and then mount them close to the optical circuits of the IQ modulators 1621 and 1622.

[0074] The optical delay units 1631 and 1632 may be arranged either before or after the optical IQ modulators 1621 and 1622 in the optical path branched into two (N=2) by the optical branch 1610. However, depending on which side of the optical IQ modulator they are arranged on, it is necessary to adjust either the wiring length from the DACs 1201 to 1208 to the optical IQ modulators 1621 and 1622 or the filter coefficients of the 16×8 filters, or both. As described above, the optical delay units 1631 and 1632 impart relative delay to the two optical paths and may simply be waveguides or optical paths in a spatial optical system with different lengths. For example, an implementation is conceivable in which the optical delay unit 1631 is the waveguide of the shortest path connecting the optical IQ modulator 1621 and the optical coupler 1641, and the optical delay unit 1632 is a waveguide that is a predetermined length longer than the optical delay unit 1631.

[0075] The optical comb generator 1501 can be formed as an optical circuit on the same substrate as the optical modulation unit 1600. The functions of the optical comb generator 1501, the optical branching unit 1621, and the optical delay units 1631 and 1632 may be replaced by a combination of an input-side coupler and a differential phase modulation unit (Patent Document 2).

[0076] DACs 1201 to 1208 need to be synchronized with the clock from clock source 1700. The specific method of synchronization is determined appropriately depending on the designs of DACs 1201 to 1208 and digital signal processing unit 1000. For example, a clock with a frequency of B / 4 may be input to DACs 1201 to 1208, or DACs 1201 to 1208 and clock source 1700 may be synchronized with the internal clock of digital signal processing unit 1000. Clock source 1700 itself may be formed integrally with the digital signal processing unit.

[0077] Let us again focus on the differences between the optical transmitter or signal generator of the prior art and the optical transmitter 1000 of the present disclosure. In contrast to a configuration in which a four-fold bandwidth is achieved solely through optical bandwidth expansion technology (e.g., Non-Patent Document 1), the optical transmitter 1000 of the present disclosure also utilizes a two-fold bandwidth expansion in the electrical domain, thereby reducing the number of parallel optical signals to two and simplifying the optical transmitter configuration. Furthermore, the optical transmitter 1000 of the present disclosure does not require the optical filter required in the prior art optical bandwidth expansion technology for separating the frequency components of the optical comb. The optical transmitter 1000 of the present disclosure also has a simpler configuration in that the optical modulation unit 1600 only needs to include optical delay units 1631 and 1632, which can be implemented using waveguides of different lengths.

[0078] Combining the conventional double-bandwidth expansion in the electrical domain with the double-bandwidth expansion in the optical domain can achieve a fourfold overall bandwidth expansion while limiting the number of parallel optical signals to two. However, a simple combination of conventional configurations requires a quadruple, double, or other multiplied clock. For example, in the signal generator of Patent Document 1, a configuration with double-bandwidth expansion in the optical domain as described in Non-Patent Document 1 can be connected to the downstream of a configuration with M=2 for electrical domain bandwidth expansion, ultimately outputting an optical signal with a bandwidth of 2B. In such a simple combination configuration, the downstream optical transmitter must set the spacing of the frequency components of the optical comb so that the output optical signals from each optical IQ modulator do not overlap on the frequency axis. In this case, the clock frequency of the optical comb generator must be B, which is four times the clock frequency (B / 4) that drives the electrical signal generator.

[0079] Instead of the optical band expansion described in Non-Patent Document 1, the technique described in Patent Document 2 allows for half of the bandwidth of the output optical signals from each optical IQ modulator to overlap. Even so, the clock frequency of the differential optical phase modulation unit corresponding to the optical comb generator is B / 2. B / 2 is twice the clock frequency driving the electrical signal generator. It is also possible to thin out the optical comb by setting the clock of the optical comb generator to B / 4 and adding an optical filter in the downstream stage to extract every fourth or every second frequency component. However, such optical filters are generally large in size and require a complex mechanism for frequency adjustment.

[0080] As described above, simply combining the bandwidth expansion in the electrical domain and the bandwidth expansion in the optical domain in the prior art requires a clock that is multiplied by 4, 2, etc. The optical transmitter disclosed herein uses only a clock with a frequency of B / 4, eliminating the need for a multiplied clock at all, and also achieving a significant simplification in the configuration of the synchronization signal.

[0081] [Second Embodiment] The optical transmitter 1000 of the first embodiment described above outputs a modulated output optical signal 103 obtained by combining modulated light into which a bandwidth-expanded electrical signal is convolved with each of the branched optical combs. The configuration of the optical transmitter 1000 can be expanded to generate a polarization multiplexed signal using two orthogonal polarizations. In the following embodiment, the components of the optical transmitter 1000 of the first embodiment are roughly parallelized to provide two systems, so that the single-polarization configuration of FIG. 1 can be easily expanded to generate a polarization multiplexed optical signal.

[0082] Fig. 9 is a diagram showing the configuration of an optical transmitter according to the second embodiment. As with Fig. 1, the optical transmitter 2000 shown in Fig. 2 is also a functional block diagram, and does not show the physical layout of the components in an actual optical transmitter.

[0083] The optical transmitter 2000 includes a digital signal processing unit 2100, sixteen DACs 2201-2216, and eight electrical signal generating units 2301-2308, and outputs a bandwidth-expanded electrical signal 102. The optical transmitter 2000 further includes a transmission laser 2400, an optical comb generator 2500, an optical modulation unit 2600, and a clock source 2700. Four of the eight DACs, together with their corresponding electrical signal generating units, generate the bandwidth-expanded electrical signal 102 for a first polarization of two orthogonal polarizations. The remaining four of the eight DACs, together with their corresponding electrical signal generating units, generate the bandwidth-expanded electrical signal 102 for a second polarization of two orthogonal polarizations.

[0084] The digital signal processing unit 2100 includes an encoding / mapping unit 2110, target signal generation units 2121 and 2122, spectral splitters 2131 and 2132, folding units 2141 and 2142, and 16×8 filters 2151 and 2152. The optical modulation unit 2600 includes an optical branching unit 2610 that branches the optical comb into four optical paths, four optical IQ modulators 2621 to 2624, four optical delay units 2631 to 2634, optical couplers 2641 and 2642, and a polarization multiplexer 2650.

[0085] Electrical signal generating sections 2301 to 2308 and optical comb generator 2500 are driven by a clock with a frequency of B / 4 sent from clock source 2700 .

[0086] In the optical transmitter 2000 shown in Figure 9, among the components of the optical transmitter 1000 shown in Figure 1, the sub-block included in the digital signal processing unit 1100 and the sub-block located on the branched optical path of the optical modulation unit 1600 are arranged in two parallel configurations. In the digital signal processing unit 1100, target signals U for two orthogonal polarizations are generated from the encoding / mapping unit 2110. Corresponding 16x8 filters 2151 and 2152 are provided for each target signal. In the optical modulation unit 1600, output optical signals are obtained from two optical couplers 2641 and 2642, respectively. Two independent output optical signals 103 are orthogonally polarization-multiplexed in a polarization multiplexer, and a multiplexed output optical signal 104 is obtained.

[0087] Therefore, the optical transmitter 2000 of this embodiment can be implemented as comprising: a digital signal processing unit 2100 configured in two parallel processes to generate an expanded electrical signal for a first polarization and an expanded electrical signal for a second polarization orthogonal to the first polarization; a DAC and corresponding electrical signal generating unit for the first polarization; a DAC and corresponding electrical signal generating unit for the second polarization; and an optical modulation unit 2600 configured in two parallel processes to generate first modulated light for the first polarization and second modulated light for the second polarization, and further including a polarization multiplexer 2650 that orthogonally polarization combines the first modulated light and the second modulated light.

[0088] An element unique to the optical transmitter 2000 is the polarization multiplexer 2650 in the optical modulation unit 2600. As the polarization multiplexer 2650, a combination of a polarization rotator and a polarization beam coupler can be used, similar to the polarization multiplexer of a commonly used polarization multiplexing IQ modulator.

[0089] [Third Embodiment] Fig. 10 is a diagram showing the configuration of an optical transmitter according to a third embodiment. As with Figs. 1 and 9, the optical transmitter 3000 shown in Fig. 10 is also a functional block diagram and does not show the physical layout of the components in an actual optical transmitter.

[0090] The optical transmitter 3000 includes a digital signal processing unit 3100, twelve DACs 3201 to 3212, and six electrical signal generation units 3301 to 3306, and outputs a bandwidth-expanded electrical signal 102. The optical transmitter 3000 further includes a transmission laser 3400, an optical comb generator 3500, an optical modulation unit 3600, and a clock source 3700.

[0091] The digital signal processing unit 3100 has an encoding / mapping unit 3110, a target signal generation unit 3121, a spectrum splitter 3131, a folding unit 3141, and a 24×12 filter 3151. The optical modulation unit 3600 has an optical branching unit 3610 that branches the optical comb into three optical paths, three optical IQ modulators 3621 to 3623, three optical delay units 3631 to 3634, and an optical coupler 3641.

[0092] Electrical signal generating units 3301 to 3306 and optical comb generator 3500 are driven by a clock with a frequency of B / 4 sent from clock source 3700 .

[0093] In the optical transmitter 3000 of this embodiment, the number of parallel optical regions is increased from two to three compared to the configurations of the first and second embodiments, and an overall bandwidth expansion of six times is realized. ｃ -3B / 2~f ｃ +3B / 2, resulting in an optical signal with a bandwidth of approximately 3B. The details of each component are the same as those of the optical transmitter 1000 shown in FIG. 1, so a description thereof will be omitted here. The operating principle of the optical transmitter 3000 can be explained in the same way as in the first embodiment, except that in the formulas used in the explanation of the first embodiment, the subscript numbers 1 to 8 are changed to 1 to 12, and 1 to 4 are changed to 1 to 6, respectively.

[0094] The output optical comb of the optical comb generator 3700 is the absolute value of the amplitude of the fourth-order component, i.e., |a ±４ In addition, the absolute value of at least one (pair) of the amplitudes of the 0th to 2nd order components, i.e., |a ０ |or|a ±１ |or|a ±２ | is also adjusted so that it is large enough. ｋ |a ｍａｘ When this is the case, |a ±４ |≧a ｍａｘ / 2 and |a ０ |≧a ｍａｘ / 2 or |a ±１ |≧a ｍａｘ / 2 or |a ±２ |≧a ｍａｘ It is desirable to satisfy at least one of the following conditions:

[0095] Furthermore, in the optical delay units 3631 to 3633, it is desirable to delay the optical signal passing through optical delay unit 3632 by approximately τ=1 / (3B) and the optical signal passing through optical delay unit 3633 by approximately 2τ=2 / (3B) based on the optical signal passing through optical delay unit 3631.

[0096] In the same manner as the expansion from the first embodiment (parallel number 2) to this embodiment (parallel number 3), it is also possible to increase the parallel number of the optical domain to 4, 5, 6, ... Even if the parallel number is increased, the driving clock frequency can remain the same as the driving clock frequency of the electrical signal generator, and there is no need to generate a multiplied wave. However, in order to increase the bandwidth of the optical comb (the number of frequency components), configuration changes are necessary, such as increasing the driving clock amplitude, cascading optical modulators for optical comb generation in multiple stages, or using optical nonlinear elements.

[0097] Furthermore, in the same manner as the expansion from the first embodiment to the second embodiment, the configuration of this embodiment or a configuration in which the number of parallel optical regions is further increased to 4, 5, or 6 can also be expanded for generating polarization multiplexed signals.

[0098] [Configuration Example of Optical Comb Generator] The optical comb generator 1500 of the optical transmitter 1000 of the first embodiment shown in FIG. 1 and the optical comb generator 2500 of the optical transmitter 2000 of the second embodiment shown in FIG. 9 can use various configurations described above. Here, an optical comb generator using a push-pull driven Mach-Zehnder optical modulation circuit, which is suitable for the optical transmitter of the present disclosure, is shown. By driving this optical comb generator under specific conditions, the functions of the optical transmitter of the present disclosure can be more efficiently realized. Specifically, the Mach-Zehnder optical modulation circuit is biased to its maximum transmission point, and the peak-to-peak amplitude of the clock signal that push-pull drives it is set to approximately three times the half-wave voltage. By using this configuration, the second-order component of the output optical comb can be made larger than the other components.

[0099] In the convolution of signal components into each frequency component of the optical comb as explained in FIG. 7, as long as the frequency components of the optical comb have ±2nd order components, the desired frequency f ｃ -B to f ｃ +B. Furthermore, if only the ±2nd-order components of the optical comb are present and no other components exist, there is almost no overlap between the spectral components in the optical domain. In the optical transmitter disclosed herein, even if there is overlap between the spectral components, the desired signal can be generated by using digital signal processing and appropriate 16x8 filter coefficients.

[0100] However, temperature changes, aging, and other factors in the device can change the relative intensity and relative phase between frequency components, resulting in non-optimal 16x8 filter coefficients. Such temperature changes, aging, and other factors can leave unwanted signal components in areas of spectral overlap, causing distortion in the output optical signal. From this perspective, it is desirable to minimize spectral overlap, and ideally, only the ±2nd order frequency components of the optical comb are strong, with the others being as weak as possible. A simple configuration for generating a waveform close to an ideal optical comb for the optical transmitters of the first and second embodiments is described below.

[0101] The driving voltage waveform is v(t), and the half-wave voltage is V π When the push-pull Mach-Zehnder optical modulator is biased to the maximum transmission point, the maximum transmission occurs when v(t) = 0, and V(t) = V π The output optical field e(t) for the CW optical input is expressed as s(t) = cos {πv(t) / (2V π )} where v(t) is the frequency f ｃｌｋ (In this example, f ｃｌｋ =B / 4), Peak-to-peak amplitude is V ｐｐ That is, v(t) = (Vpp / 2) sin(2πf ｃｌｋ In this case, the output optical field s(t) is expressed by the following equation:

[0102] In the above equation, J ｋ is the kth order Bessel function of the first kind. That is, s(t) is f ｃｌｋ The optical field amplitude of the k-th order component is J ｋ {πV ｐｐ / (4V π )}. In other words, odd-order components can be removed simply by biasing to the maximum transmission point. ｐｐ By optimizing the above, the ±2nd order components can be made relatively large.

[0103] 11A and 11B are diagrams illustrating the operation of an optical comb generator suitable for the optical transmitter of the present disclosure. ｐｐ / V π and the Bessel functions of the first kind J from 0th to 4th order (k = 0, 1, 2, 3, 4) ｋ {πV ｐｐ / (4V π )}. The first and third order components shown by the dotted lines are cut from the output optical comb by biasing it to the maximum transmission point. Focusing on the zeroth, second and fourth orders, V ｐｐ / V π By setting the value to about 3, the magnitude of the second-order component can be made larger than the zeroth-order and fourth-order components.

[0104] 11(b) shows the measured spectrum of the output optical signal when the push-pull Mach-Zehnder optical modulation circuit is biased to the maximum transmission point. ｃｌｋ 28GHz, V ｐｐ to 3V π The optical filter was not used. The horizontal axis represents the relative frequency (GHz) with respect to the optical carrier frequency. The ±56 GHz components, which correspond to the ±2nd order, are the main components, with the intensity extinction ratio for the 0th order (0 GHz) component being approximately 10 dB, and the intensity extinction ratio for the ±4th order (±112 GHz) component being approximately 15 dB. Due to imperfections in the balance between the arms of the Mach-Zehnder optical modulation circuit, the ±1st order and +3rd order components are slightly visible, but the intensity extinction ratio for these components is also close to 20 dB.

[0105] 12 is a configuration diagram of an optical comb generator using a push-pull drive type Mach-Zehnder optical modulation circuit. The optical comb generator 800 includes an optical branching unit 801, a differential phase modulation arm 802, a DC bias adjustment unit 803, and an optical coupler 804. CW light 810 is input to the optical comb generator 800, a clock is input from a clock source to the differential phase modulation arm 802, and an optical comb 811 is output. The bias described above is provided to the DC bias adjustment unit 803.

[0106] As described above, the push-pull Mach-Zehnder optical modulation circuit is biased to the maximum transmission point, and V ｐｐto 3V π By driving the filter at about 1 / 2, it is possible to emphasize ±2nd order components relative to the clock frequency and suppress other components. In the optical transmitters of the first and second embodiments, it is possible to suppress signal distortion due to deviations from the optimum point of the 16x8 filter coefficients. ｐｐ The setting range is 2.7V, where the absolute values ​​of the 0th and ±4th order amplitudes are less than half of the absolute value of the ±2nd order amplitudes. π ~3.7V π It is desirable to do so.

[0107] As described above in detail, the optical transmitter of the present disclosure simultaneously achieves the bandwidth expansion of the DAC and a simplified configuration in which a multiplying clock circuit and a large optical filter are omitted.

[0108] The present invention can be used in an optical transmitter in optical communications.

Claims

1. An optical transmitter comprising: a clock source that outputs a clock signal with a frequency of B / (2N) (N is an integer of 2 or more); an optical comb generator that generates an optical comb by modulating continuous wave light (CW light) with the clock signal; a digital signal processing unit; 4N digital-to-analog converters (DACs), each outputting an electrical signal having a bandwidth of B / (2N); 2N electrical signal generation units driven by the clock signal, each receiving two of the electrical signals as input and outputting an electrical signal with a bandwidth expanded by two times; an optical branching unit that branches the optical comb into N optical paths; N optical IQ modulators corresponding to each of the N branched optical combs, each using two of the 2N bandwidth-expanded electrical signals as drive signals and outputting a modulated optical signal; an optical delay unit that imparts a different time delay to each of the N optical paths; and an optical combiner that combines the modulated optical signals from the N optical IQ modulators. the digital signal processing unit comprises: a target signal generation unit that virtually generates a target signal having a bandwidth of 2B; a spectrum division unit that divides the target signal into 4N signals having a bandwidth of B / (2N) and generates 4N digital divided signals that correspond to signals obtained by frequency-shifting each of the divided signals to baseband; a folding unit that obtains 4N folded digital divided signals by inverting each of the digital divided signals on the frequency axis around the frequency B / (4N) and taking the complex conjugate; and an 8N x 4N filter unit that receives the 4N digital divided signals and the 4N folded digital divided signals as inputs and outputs corresponding digital signals to the 4N DACs.

2. The optical transmitter according to claim 1, comprising: the digital signal processing unit, which is configured in two parallel processes to generate an expanded electrical signal for a first polarization and an expanded electrical signal for a second polarization orthogonal to the first polarization; a DAC and corresponding electrical signal generating unit for the first polarization; a DAC and corresponding electrical signal generating unit for the second polarization; and the optical modulation unit, which is configured in two parallel processes to generate first modulated light for the first polarization and second modulated light for the second polarization, and further includes a polarization multiplexer that orthogonally polarizes and multiplexes the first modulated light and the second modulated light.

3. An optical transmitter according to claim 1 or 2, wherein N=2, and the optical comb generator is a Mach-Zehnder type optical modulation unit biased to the point of maximum transmission and driven by a clock signal of frequency B / 4 whose peak-to-peak amplitude is approximately three times the half-wave voltage.

4. An optical transmitter according to claim 1 or 2, further comprising a transmitting laser for supplying said CW light.

Images