JP2016009915A - Digital filter and digital coherent receiver - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce the number of errors in symbol determination, even if the statistical trend is different from symbol to symbol.SOLUTION: A digital filter 10 is used for demodulation of a signal subjected to multilevel phase modulation. The digital filter 10 includes a storage unit 11 for storing a plurality of tap coefficients set for each symbol that is identified in the multilevel phase modulation, a selection unit 12 for selecting a first tap coefficient cfrom the plurality of tap coefficients based on the first coordinate information xinputted on a complex plane, and an estimation unit 13 for estimating the second coordinate information aon the complex plane based the first coordinate information xand first tap coefficient c, and outputting the second coordinate information a.

Description

  The present invention relates to a digital filter and a digital coherent receiver.

  Development of a digital coherent optical transmission system is in progress to increase the capacity of the core network. In a digital coherent optical transmission system, an optical signal generated by modulation such as phase shift keying (PSK) or quadrature amplitude modulation (QAM) and polarization multiplexing is transmitted to a transmitter (transmitter). ) To the receiving side (receiver) via the optical fiber. On the receiving side, an optical signal received via an optical fiber is demodulated by digital signal processing such as chromatic dispersion compensation and polarization processing through polarization separation and intradyne detection using local oscillation light.

  Various configurations and methods have been proposed for carrier phase recovery, which is one of the digital signal processing. For example, Patent Document 1 and Non-Patent Document 1 disclose carrier phase recovery using a BPS (Blind-Phase-Search) method. Patent Document 2 and Non-Patent Document 2 disclose carrier phase recovery using a combination of a decision-oriented PLL (Phase Locked Loop) and a decision-oriented equalizer.

US Patent Application Publication No. 2011/0318021 US Patent Application Publication No. 2011/0150503 Specification

Journal of Light wave technology, Vol27, No8 IEEE Trans. On Signal Processing Vol.40 No.6 June 1992

  By the way, the amount of information transmitted by one modulation is 8 values in polarization multiplexed QPSK (Quadrature Phase Shift Keying), 16 values, 64 values or 256 values in QAM, and is multi-valued to increase capacity. Development is underway. However, since the interval between each point (hereinafter referred to as “symbol”) on the constellation map becomes narrow due to the multi-value conversion, more accurate phase compensation processing is required.

  In the digital signal processing method described above, based on the statistical data of all symbols, for example, feed-forward compensation such as blind phase search or maximum likelihood method (Maximum Likelihood Method), or a decision-oriented approach, for example. Feedback compensation is performed. However, in either method, compensation is performed so as to minimize the average error for all symbols, so that when individual symbols show different error vector tendencies, it is optimal for each symbol. It is not a safe compensation.

  For example, imperfections caused by the reception front end (imbalance and variation between in-phase and quadrature components, etc.) cause distortion and bias on the constellation map, or differences in statistical trends among individual symbols. Sometimes. For this reason, it is necessary to perform compensation corresponding to this, and in particular, it becomes important in the case of multilevel.

  One embodiment of the present invention provides a digital filter and a digital coherent receiver having a structure capable of reducing the number of errors in symbol determination even when a statistical tendency is different for each symbol.

  A digital filter according to one embodiment of the present invention is a digital filter used in demodulation of a multi-level phase modulated signal. The digital filter includes a storage unit that stores a plurality of tap coefficients set for each symbol identified in multi-level phase modulation, and a plurality of tap coefficients based on the input first coordinate information on the complex plane. A selection unit that selects one tap coefficient; and an estimation unit that estimates second coordinate information on the complex plane based on the first coordinate information and the first tap coefficient, and outputs the second coordinate information.

  According to one aspect of the present invention, the number of errors in symbol determination can be reduced even when the statistical tendency differs for each symbol.

It is a block diagram which shows typically the digital coherent receiver which concerns on one Embodiment. FIG. 2 is a functional block diagram of a digital signal processing unit in FIG. 1. It is a block diagram which shows typically the digital filter which concerns on one Embodiment. FIG. 4 is a configuration diagram illustrating an example of the register of FIG. 3. (A) is a figure which shows the temporary determination of FIG. 3, (b) is a figure which shows this determination of FIG. (A) is a figure which shows an example of the constellation before the equalizer by a digital filter in case the digital coherent receiver of FIG. 1 receives DP-16QAM signal light, (b) is a digital coherent receiver of FIG. It is a figure which shows the constellation after the equalizer by a digital filter at the time of receiving 16QAM signal light. (A) is a figure which shows distortion of the constellation of (a) of FIG. 6, (b) is a figure which shows distortion of the constellation of (b) of FIG.

[Description of Embodiment of Present Invention]
First, the contents of the embodiment of the present invention will be listed and described.

  A digital filter according to one embodiment of the present invention is a digital filter used in demodulation of a multi-level phase modulated signal. The digital filter includes a storage unit that stores a plurality of tap coefficients set for each symbol identified in multi-level phase modulation, and a plurality of tap coefficients based on the input first coordinate information on the complex plane. A selection unit that selects one tap coefficient; and an estimation unit that estimates second coordinate information on the complex plane based on the first coordinate information and the first tap coefficient, and outputs the second coordinate information.

  According to this digital filter, a tap coefficient is set for each symbol identified in multilevel phase modulation. A first tap coefficient is selected from a plurality of tap coefficients based on the input first coordinate information, and second coordinate information is estimated based on the first coordinate information and the first tap coefficient. For this reason, even if the statistical tendency of deviation from the ideal value differs for each symbol, the equalizer operation is performed so that the error vector for each symbol is minimized. As a result, the number of errors in symbol determination can be reduced even when the statistical tendency differs for each symbol.

  The estimation unit may include a multiplier that multiplies the first coordinate information and the first tap coefficient, and a main determination unit that determines which symbol is based on a multiplication result of the multiplier. In this case, by multiplying the first coordinate information and the first tap coefficient, it is possible to perform an equalization process that minimizes the error vector for each symbol even when the statistical tendency differs for each symbol. Since the symbol is determined based on the equalized coordinate information, the number of errors in symbol determination can be reduced.

  The digital filter according to another aspect of the present invention may further include an updating unit that updates the first tap coefficient based on a difference between the multiplication result and the second coordinate information. In this case, the tap coefficient for each symbol can be finely corrected and converged to optimize the tap coefficient.

  The updating unit may update the first tap coefficient using an LMS (Least Mean Square) algorithm. Since the LMS algorithm is performed by simple calculation, the update of the first tap coefficient can be simplified.

  A selection unit configured to determine which symbol is a symbol based on the first coordinate information; and a tap coefficient set for the symbol determined by the temporary determination unit from a plurality of tap coefficients. And a selector that selects as a coefficient. In this case, it is possible to temporarily determine which symbol is based on the first coordinate information, and perform equalization processing using the tap coefficient set for the temporarily determined symbol. Thereby, a tap coefficient suitable for the first coordinate information can be used.

  A digital coherent receiver according to an aspect of the present invention includes the digital filter described above. In this digital coherent receiver, it is possible to reduce the number of errors in symbol determination even when the statistical tendency differs for each symbol.

[Details of the embodiment of the present invention]
Specific examples of the digital filter and the digital coherent receiver according to the embodiment of the present invention will be described below with reference to the drawings. In addition, this invention is not limited to these illustrations, is shown by the claim, and intends that all the changes within the meaning and range equivalent to a claim are included.

  FIG. 1 is a configuration diagram schematically illustrating a digital coherent receiver according to an embodiment. The digital coherent receiver is a device that receives multilevel phase-modulated signal light via an optical fiber and extracts an original symbol. Here, multilevel phase modulation includes multilevel (M-value) quadrature amplitude modulation (M-QAM), multilevel (M-value) phase shift keying (M-PSK), and the like. The phase shift keying includes quaternary phase shift keying (QPSK), differential quadrature phase shift keying (DQPSK), and polarization multiplexed quaternary phase shift keying (DP−). QPSK; Dual Polarization-Quadrature Phase ShiftKeying) and the like.

  As shown in FIG. 1, the digital coherent receiver 1 includes a local light source 2, a reception front end 3, an AD converter 4, and a digital signal processing unit 5. The local light source 2 is, for example, a laser oscillator, and generates local light (hereinafter referred to as “LO light”) having substantially the same frequency as the carrier frequency of the signal light received by the digital coherent receiver 1. The local light source 2 supplies LO light to the reception front end 3.

  The reception front end 3 includes a polarization separator, an optical phase hybrid circuit, and a PD (balanced light receiving element) / TIA (transimpedance amplifier). The polarization separator separates the signal light received from the optical fiber into two orthogonal polarization components (X polarization and Y polarization). The optical phase hybrid circuit causes the in-phase component and the quadrature component of the LO light received from the local light source 2 to interfere with each polarization component, and the in-phase interference component (I component) and the quadrature interference component (Q Component). The I component and Q component are each output as two complementary signals in pairs. PD / TIA receives complementary signals of each component, converts them into voltage signals XI signal, XQ signal, YI signal and YQ signal, and converts each converted signal to an AD converter. 4 is output.

  The AD converter 4 samples the XI signal, the XQ signal, the YI signal, and the YQ signal output by the reception front end 3, and the XI signal and the XQ signal, which are digital signals, are sampled. , YI signal and YQ signal, respectively.

  The digital signal processing unit 5 performs digital signal processing described later on the XI signal, XQ signal, YI signal, and YQ signal sampled by the AD converter 4. The digital signal processing unit 5 is, for example, an application specific IC (ASIC) that executes signal processing calculations at high speed.

  FIG. 2 is a functional block diagram of the digital signal processing unit 5. As shown in FIG. 2, the digital signal processing unit 5 includes a front end compensation unit 51, a chromatic dispersion compensation unit 52, a clock recovery unit 53, a polarization separation / PMD compensation unit 54, and a frequency offset compensation unit 55. A phase recovery unit 56, and a symbol determination unit 57. In FIG. 2, the XI signal and the XQ signal are collectively expressed as XI + jXQ, and the YI signal and the YQ signal are collectively expressed as YI + jYQ.

  The front end compensation unit 51 corrects various mismatches caused by the reception front end 3. For example, the front end compensation unit 51 performs scale adjustment and skew adjustment between channels (XI signal, XQ signal, YI signal, and YQ signal), phase error correction between IQs, and the like.

  The chromatic dispersion compensation unit 52 compensates the chromatic dispersion for the digital signal output from the front end compensation unit 51. This chromatic dispersion is a linear waveform distortion that occurs due to the occurrence of different propagation times for each wavelength component of an optical signal in an optical transmission line such as an optical fiber and does not vary with time. The chromatic dispersion compensation unit 52 is, for example, an equalizer with a fixed tap coefficient, and has an inverse characteristic of waveform distortion.

  The clock recovery unit 53 extracts a clock signal from the digital signal output from the chromatic dispersion compensation unit 52, and performs identification reproduction of the digital signal using the clock signal. The correction can also be referred to as correction in the time axis direction.

  The polarization separation / PMD compensation unit 54 separates the digital signal output from the clock recovery unit 53 into signal components of vertical polarization and horizontal polarization, and generates distortion caused by polarization mode dispersion (PMD). To compensate. This PMD is generated by generating a different propagation time for each polarization component in an optical transmission line such as an optical fiber. This compensation is mainly correction of the amplitude direction of the multilevel signal.

  The frequency offset compensation unit 55 compensates the frequency offset for the digital signal output from the polarization separation / PMD compensation unit 54. The frequency offset is a frequency difference (offset) between the signal light (transmitter light source) and the LO light (local light source 2).

  As a result of the frequency offset compensation performed by the frequency offset compensator 55, the phase recovery unit 56 has a phase remaining in that state for a digital signal whose rotation on the complex plane is roughly corrected according to a certain frequency offset. A phase shift (phase offset) from a desired symbol arrangement is compensated by performing a relatively high-speed rotation process for noise (phase fluctuation) as compared with frequency offset compensation.

  The constellation map is, for example, a plot of the I component and Q component of each input symbol on the complex plane as coordinate information. When the plotted positions are viewed in a two-dimensional polar coordinate system, the moving radius represents the amplitude of the multilevel signal, and the declination represents the phase of the multilevel signal. Here, the M symbols identified in the M-value phase modulation are respectively assigned coordinates on the complex plane. The area assigned to each symbol is an area centered on the symbol coordinate of each symbol on the complex plane.

  The symbol determination unit 57 determines the position of the symbol on the constellation map based on the coordinate information calculated by the phase recovery unit 56.

Here, a digital filter according to an embodiment will be described. FIG. 3 is a configuration diagram schematically illustrating a digital filter according to an embodiment. The digital filter 10 is a digital filter used in demodulating a multi-level (M-value) phase-modulated signal, and is a decision-oriented equalizer. The digital filter 10 has a part of the function of the phase recovery unit 56 of FIG. 2 and the function of the symbol determination unit 57, and inputs the first coordinate information _xn phase-recovered in the previous processing unit, and it outputs the second coordinate information a _n to the subsequent processing unit.

Here, the first coordinate information _xn is an input symbol of the digital filter 10 and is a complex number corresponding to a coordinate value (vector) on the complex plane. The second coordinate information a _n, a complex number corresponding to the symbol decision result, i.e. the coordinate values on the complex plane of the decided symbol. The subscript n indicates the nth (current) time slot. The digital filter 10 is a digital filter for X polarization, but a digital filter similar to the digital filter 10 is provided for Y polarization.

The digital filter 10 includes a storage unit 11, a selection unit 12, an estimation unit 13, and an update unit 14. The first coordinate information _xn is input to the selection unit 12, the estimation unit 13, and the update unit 14.

  The storage unit 11 stores M tap coefficients set for each of M symbols identified in the M-value phase modulation. The storage unit 11 includes M registers R (storage elements). The M registers R are respectively associated with M symbols identified in the M-value phase modulation. Of the M registers R, a register Ri (i = 1 to M) stores a tap coefficient ci set for the i-th symbol. The register Ri outputs the stored tap coefficient ci to the selection unit 12.

The storage unit 11 selects a register Ri corresponding to a symbol determined by a provisional determination unit 21 described later according to the control signal sel output from the selection unit 12. The storage unit 11 writes the first tap coefficient c _n, which is updated by the update unit 14 to the register Ri selected in the edge of the next clock signal. Details of the register R will be described later.

The selection unit 12 selects the first tap coefficient cn ₋₁ from the plurality of tap coefficients based on the first coordinate information x _n on the complex plane input from the outside (the previous processing unit). The selection unit 12 includes a provisional determination unit 21 and a selector 22. The provisional determination unit 21 determines which symbol is based on the first coordinate information _xn , and outputs a control signal sel for controlling the selector 22 to the selector 22. The temporary determination unit 21 also outputs the control signal sel to the storage unit 11.

  The provisional determination unit 21 includes, for example, each symbol, symbol coordinate information indicating a symbol coordinate assigned to each symbol on the complex plane, and region information regarding coordinates defining a region assigned to each symbol on the complex plane. Are stored in association with each other. In the following description, using a constellation map means using symbol coordinate information and area information of each symbol.

Specifically, the provisional determination unit 21 determines, using the constellation map, which region on the complex plane the input first coordinate information x _n belongs to, and which symbol is the determined region. It is determined whether it is assigned to. The temporary determination unit 21, the selector 22, and the storage unit 11 are connected by, for example, M signal lines corresponding to M symbols, and the temporary determination unit 21 enables a signal line corresponding to the determined symbol. Thus, the control signal sel is output to the selector 22, and the selector 22 is controlled so as to select the tap coefficient ci corresponding to the determined symbol.

A signal line set for the i-th symbol is connected to the register Ri. Temporary decision unit 21, by enabling a signal line corresponding to the determined symbols, select register Ri, first tap coefficient c _n the next clock signal is updated by the update unit 14 to the selected register Ri It is controlled so that it is written at the edge.

The selector 22 selects a tap coefficient set for the symbol determined by the temporary determination unit 21 from a plurality of tap coefficients as the first tap coefficient. The selector 22 is an M-input 1-output selector. The selector 22 receives the outputs (tap coefficients c) of the M registers R, and from the register Ri corresponding to the enabled signal line, depending on which signal line is enabled by the temporary determination unit 21. Are selected and output to the estimating unit 13 and the updating unit 14 as the first tap coefficient cn _-1 .

Estimation unit 13, based on the first coordinate information _{x n,} and the first tap coefficient _{c n-1} estimates the second coordinate information _{a n} in the complex plane, and outputs the second coordinate information _{a n.} The estimation unit 13 includes a multiplier 31 and a main determination unit 32. The multiplier 31 calculates (multiplies) the product of the first coordinate information x _n and the first tap coefficient c _n−1 as shown in the equation (1), and updates the multiplication result z _n to the main determination unit 32 and updates To the unit 14. The multiplication result z _n can be said to be a value obtained by equalizing the first coordinate information x _n .

The main determination unit 32 determines which symbol is based on the multiplication result z _n of the multiplier 31. Similar to the provisional determination unit 21, the main determination unit 32 includes, for example, each symbol, symbol coordinate information indicating the symbol coordinate assigned to each symbol on the complex plane, and an area assigned to each symbol on the complex plane. Are stored in association with the area information relating to the coordinates that define. The main determination unit 32 uses the constellation map to determine which region on the complex plane the multiplication result z _n belongs to, and to which symbol the determined region is assigned. This determination unit 32 outputs the determined symbol coordinate information assigned to the symbol on the second coordinate information a _n estimated by the updating section 14 and the outside (the subsequent processing unit). That is, the present determination unit 32, the multiplication result closest symbol coordinates z _n is estimated as the second coordinate information a _n.

Updating unit 14 first updates the tap coefficients _{c n-1} based on the difference between the multiplication result _{z n} and the second coordinate information _{a n.} The updating unit 14 updates the first tap coefficient cn ₋₁ using, for example, an existing LMS algorithm. The update unit 14 includes an adder 41 and a tap coefficient update unit 42. The adder 41 calculates the sum of the negative number of the multiplication result z _n and the second coordinate information a _n (addition). That is, the adder 41, as shown in equation (2), the multiplication result by subtracting the second coordinate information _{a n} from _{z n,} a coordinate error _{e n} from the second coordinate information _{a n} multiplication result _{z n} calculate. The adder 41 outputs the coordinate error _{e n} to the tap coefficient updating unit 42.

As shown in the equation (3), the tap coefficient update unit 42 includes a first tap coefficient c _n−1 , a tap coefficient update step size μ (corresponding to an integral gain), a coordinate error e _n , using conjugate 1 coordinate information x _n, calculates a first tap coefficient c _n, which is updated. That is, the tap coefficient updating unit 42 uses the LMS algorithm, the step size mu, and coordinate error e _n, that the product of the conjugate of the first coordinate information x _n is subtracted from the first tap coefficient c _n-1 Thus, the tap coefficient is finely corrected and converged. Tap coefficient updating unit 42 outputs the first tap coefficient c _n, which is updated in the storage unit 11.

Here, the register R will be described in detail. FIG. 4 is a configuration diagram illustrating an example of the register R. The register R includes, for example, a selector 61 and a D-type flip-flop (D-FF) 62. The D-FF 62 outputs the value input to the input terminal D from the output terminal Q to the selector 61 as the first tap coefficient cn ₋₁ at the edge of the clock signal. Selector 61, D-FF 62 first tap coefficients _{c n-1} output from the inputs the first tap coefficient _{c n,} which is updated by the update unit 14, the control signal sel output from the temporary judgment unit 21 either the first tap coefficient _{c n-1} and the first tap coefficient _{c n,} which is updated and output to the input terminal D of the D-FF 62 in accordance with the.

Selector 61, for example, the control signal sel is enabled (e.g., high level), then at output a first tap coefficient c _n, which is updated by the update unit 14, the control signal sel is disabled (e.g., low level) If there is, the first tap coefficient cn ₋₁ output from the D-FF 62 is output. Thus, the control signal sel is also used as a write control signal of the register R, in synchronization with the next clock edge, the register selected by the control signal sel, the first tap coefficient c _n, which is updated is written .

Next, the operation of the digital filter 10 will be described. The operation of the digital filter 10 is started by inputting the first coordinate information _xn , which is an input symbol that has been subjected to carrier phase recovery processing by the phase recovery unit 56 of FIG. First, the provisional determination unit 21 determines which region on the complex plane the input first coordinate information _xn belongs to, and determines which symbol the determined region is assigned to. Then, the temporary determination unit 21 outputs a control signal sel to the selector 22 by enabling a signal line corresponding to the determined symbol (temporary determination step).

Subsequently, the selector 22 selects the tap coefficient ci from the register Ri corresponding to the signal line enabled by the provisional determination unit 21, and the multiplier 31 and the tap coefficient update unit 42 as the first tap coefficient cn _−1. (Tap coefficient selection step). Then, the multiplier 31 multiplies the first coordinate information x _n and the first tap coefficient cn ₋₁ , and outputs the multiplication result z _n to the main determination unit 32 and the adder 41 (equalization step).

Subsequently, the determination unit 32 determines to which region on the complex plane the multiplication result z _n belongs, and determines to which symbol the determined region is assigned. Then, the determination unit 32, a symbol coordinate information assigned to the determined symbol estimates that second coordinate information a _n outputs to the processing unit of the adder 41 and the latter stage (the determination step). The adder 41, the multiplication result from z _n by subtracting the second coordinate information a _n calculates coordinates errors e _n, and outputs the coordinate error e _n to the tap coefficient updating unit 42 (coordinate error calculation step).

Subsequently, the tap coefficient updating unit 42 uses the LMS algorithm, subtracting the step size mu, and coordinate error e _n, the product of the conjugate of the first coordinate information x _n from the first tap coefficient c _n-1 it allows calculating a first tap coefficient c _n, which is updated, and outputs a first tap coefficient c _n, which is updated in the storage unit 11. Then, the storage unit 11 selects the register Ri corresponding to the symbol determined by the temporary determination unit 21 in accordance with the control signal sel output from the temporary determination unit 21, and is updated to the selected register Ri. writes the tap coefficients c _n in synchronization with the next clock edge (tap coefficient update step). The above processing is repeated for each time slot (clock cycle).

As described above, the digital filter 10 selects the first tap coefficient cn ₋₁ based on the symbol determination result (provisional determination) of the first coordinate information x _n that is the carrier phase recovery output of the previous stage, and the first coordinate. After performing an equalization process obtained by multiplying the information _xn and the first tap coefficient cn ₋₁ , a symbol determination result (main determination) is performed. That is, after the carrier phase recovery process of the phase recovery unit 56, an equalization process (one-tap decision-oriented applied filter) is further added. Moreover, the results of the determination to calculate the coordinate error e _n the basis, and updates the first tap coefficient c _n-1 which is selected.

Next, operational effects of the digital coherent receiver 1 and the digital filter 10 will be described. 5A is a diagram illustrating the provisional determination of FIG. 3, and FIG. 5B is a diagram illustrating the main determination of FIG. In other words, FIG. 5A shows the position (first coordinate information x _n ) of the input symbol after carrier phase recovery in the previous stage of the digital filter 10, and FIG. 5B shows the position of FIG. 5A. The position of the input symbol (multiplication result z _n ) after the input symbol is equalized by the multiplier 31 of the digital filter 10 is shown.

  As shown in FIGS. 5A and 5B, the complex plane is divided into a plurality of regions D by the determination boundary line B. This region D is assigned to each symbol identified in the M-value phase modulation, and an input symbol located in the region D is determined as a symbol of the region D.

  In FIG. 5A, the input symbol Smaj and the input symbol Smin belong to the region D1. The input symbol Smin is separated from the group of major input symbols Smaj due to the imperfection of the reception front end 3. However, the input symbol Smin is supposed to be determined as a symbol in the adjacent (upper) region D2, and therefore an error (incorrect determination) occurs.

  On the other hand, as shown in FIG. 5B, after equalization by the multiplier 31 of the digital filter 10, the group of major input symbols Smaj moves to the center side of the region D1, and the input symbol Smin is originally It is expelled to the area D2 of the symbol to be determined. This eviction of the input symbol Smin contributes to a reduction in the number of symbol errors.

  6A is a diagram illustrating an example of a constellation before the equalizer by the digital filter 10 when the digital coherent receiver 1 receives DP-16QAM signal light, and FIG. 6B is a diagram illustrating the digital coherent receiver 1. It is a figure which shows the constellation after the equalizer by the digital filter 10 when DP-16QAM signal light is received. 6A and 6B, the horizontal axis represents the complex real part of the X polarization, and the vertical axis represents the complex imaginary part of the X polarization. In FIGS. 6A and 6B, the error symbol is indicated by “◯”.

  As shown in FIG. 6A, the number of error symbols is 23 at the position of the input symbol after carrier phase recovery in the previous stage of the digital filter 10. On the other hand, as shown in FIG. 6B, after equalization by the multiplier 31 of the digital filter 10, the number of error symbols is reduced to eight. In this example, the number of error symbols is reduced to less than half by the equalization processing of the digital filter 10, and the margin for the correction capability of the subsequent FEC (Forward Error Correction) is increased. As described above, the digital filter 10 can suppress the influence of imperfection of the reception front end 3 and reduce the number of errors in symbol determination.

  FIG. 7A is a diagram showing distortion of the constellation of FIG. 6A, and FIG. 7B is a diagram showing distortion of the constellation of FIG. 6B. 7A and 7B, the horizontal axis represents the complex real part of the X polarization, and the vertical axis represents the complex imaginary part of the X polarization. FIG. 7 shows a state in which the arrangement of the center points (average values) of input symbols is linearly approximated by the least square method for each row and each column in order to visualize the distortion of the constellation.

  As shown in FIG. 7A, in the constellation before the equalizer by the digital filter 10, each straight line is inclined obliquely, and distortion is recognized. On the other hand, as shown in FIG. 7B, in the constellation after the equalizer by the digital filter 10, the inclination of each straight line is corrected horizontally and vertically to compensate the distortion, and the influence of the reception front end 3 is suppressed. It was. Thus, although the arrangement of the input symbols of the digital filter 10 was inclined, it was confirmed that it was corrected by the equalization processing of the digital filter 10. 6A and 6B and FIG. 7A and FIG. 7B show constellations relating to X polarization. Usually, since the characteristics do not change between the X polarization and the Y polarization, the same effect can be obtained with respect to the Y polarization.

As described above, in the digital filter 10, the tap coefficient c is prepared for each symbol identified in the multilevel phase modulation. Further, in the digital filter 10, the first tap coefficient c _n−1 is selected from the plurality of tap coefficients c based on the input first coordinate information x _n , and the first coordinate information x _n and the first tap coefficient c are selected. second coordinate information _{a n} is estimated based on _n-1. For this reason, even if the statistical tendency of deviation from the ideal value differs for each symbol, the equalizer operation is performed so that the error vector for each symbol is minimized. As a result, the number of errors in symbol determination can be reduced even when the statistical tendency differs for each symbol.

Further, the digital filter 10 tentatively determines which symbol is based on the first coordinate information _xn, and sets the tap coefficient set for the tentatively determined symbol as the first tap coefficient cn _−1. The equalization process is used. This equalization process is performed by multiplying the first coordinate information _xn and the first tap coefficient cn _-1 . Thereby, a tap coefficient suitable for the first coordinate information _xn is used, and equalization processing can be performed such that the error vector for each symbol is minimized even when the statistical tendency differs for each symbol. Since symbol determination is performed based on the coordinate information (multiplication result z _n ) subjected to equalization processing, the number of errors in symbol determination can be reduced.

  For updating the tap coefficients, a general LMS algorithm is used as an update algorithm for the tap coefficients of the adaptive filter. Thereby, the tap coefficient is updated (optimized) by feeding back the result of the symbol determination by the main determination unit 32 to the tap coefficient. For this reason, the tap coefficient can be converged to an optimum value that is different for each symbol. Further, since the LMS algorithm is performed by simple calculation, updating of the tap coefficient can be simplified.

  As described above, even when the error vector has a different tendency for each symbol, the optimum tap coefficient can be obtained. For example, distortion of the constellation due to imperfection of the reception front end 3 such as IQ orthogonality can be reduced, and the number of errors can be reduced. Can be reduced. As a result, it is possible to increase the margin for the error correction capability of the subsequent FEC. The digital filter 10 is an adaptive operation and can be realized with a simple configuration.

  The digital filter and the digital coherent receiver according to the present invention are not limited to the above embodiment. For example, the digital filter 10 does not need to include the update unit 14. That is, it is sufficient that M tap coefficients are set for each of M symbols identified in the M-value phase modulation, and updating of the tap coefficients is not necessarily required.

  Further, feedback compensation of the digital filter 10 may be combined with conventional feedforward type compensation.

  DESCRIPTION OF SYMBOLS 1 ... Digital coherent receiver, 10 ... Digital filter, 11 ... Memory | storage part, 12 ... Selection part, 13 ... Estimation part, 14 ... Update part, 21 ... Temporary determination part, 22 ... Selector, 31 ... Multiplier, 32 ... Book Determination unit, 41... Adder, 42... Tap coefficient update unit, R.

Claims (6)

A digital filter used in demodulation of a multi-level phase modulated signal,
A storage unit for storing a plurality of tap coefficients set for each symbol identified in the multi-level phase modulation;
A selection unit that selects a first tap coefficient from the plurality of tap coefficients based on the input first coordinate information on the complex plane;
An estimation unit that estimates second coordinate information on a complex plane based on the first coordinate information and the first tap coefficient, and outputs the second coordinate information;
Digital filter with   The estimation unit includes a multiplier that multiplies the first coordinate information and the first tap coefficient, and a main determination unit that determines which symbol is based on a multiplication result of the multiplier. The digital filter according to 1.   The digital filter according to claim 2, further comprising an updating unit that updates the first tap coefficient based on a difference between the multiplication result and the second coordinate information.   The digital filter according to claim 3, wherein the updating unit updates the first tap coefficient using an LMS (Least Mean Square) algorithm.   The selection unit is a temporary determination unit that determines which symbol is based on the first coordinate information, and a tap coefficient that is set for the symbol determined by the temporary determination unit from the plurality of tap coefficients. The digital filter according to any one of claims 1 to 4, further comprising: a selector that selects the first tap coefficient.   A digital coherent receiver comprising the digital filter according to any one of claims 1 to 5.