Adaptive equalizer, adaptive equalization method, and optical communication system

Abstract

The tap coefficient control circuit (12) sets the tap coefficient updated by the second tap coefficient updater (11) as the initial value of the convergence action of the tap coefficient of the first digital filter (13) updated by the first tap coefficient updater (14), arranges the tap coefficients set as the initial value in descending order according to the degree of contribution to the convergence action of the first tap coefficient updater (14), judges the tap coefficients above the upper specified number as valid, judges the tap coefficients below the specified number as invalid, sets the tap coefficients of the first digital filter (13) corresponding to the tap coefficients judged to be invalid to zero and does not use them in the calculation of the first tap coefficient updater (14) until the next judgment result comes out.

Description

Technical Field

[0001] The present disclosure relates to an adaptive equalizer, an adaptive equalization method, and an optical communication system for compensating for characteristics of a transmission path in data communication. Background Art

[0002] In coherent optical communications, digital signal processing is used on the receiving side to compensate for signal distortion, enabling high-capacity transmission exceeding tens of Gbit / s. Digital signal processing primarily involves wavelength dispersion compensation, frequency control / phase adjustment, polarization multiplexing and separation, and polarization dispersion compensation.

[0003] Polarization multiplexing and separation, as well as polarization dispersion compensation, are primarily performed through adaptive equalization. Adaptive equalizers in digital signal processing utilize digital filters. By setting tap coefficients calculated to offset the distortion of the transmitted signal, the transmitted signal can be compensated. The tap coefficients of a digital filter are equivalent to the impulse response of the filter characteristics. These coefficients are updated sequentially to adapt to changing conditions over time, and the adaptive equalizer compensates for changes in the state of polarization (SOP).

[0004] To update the tap coefficients of the digital filter that makes up an adaptive equalizer, a successive update algorithm, such as the Constant Modulus Algorithm (CMA), is typically used. The tap coefficients converge according to this algorithm, settling on a predetermined value. Therefore, in an adaptive equalizer, the greater the number of taps, the greater the computational effort. Furthermore, as computational effort increases, power consumption also increases. Reducing the number of taps reduces computational effort, thereby reducing the adaptive equalizer's power consumption.

[0005] In the past, there was no established method for dynamic control that would not cause performance degradation. Therefore, low power consumption was achieved by limiting the number of taps from the tap center. In other words, the tap coefficients at both ends were reduced. However, when the DGD (Differential Group Delay) load is large, the tap coefficients at both ends are required. The DGD load represents the delay difference between the horizontally polarized signal and the vertically polarized signal. Therefore, by limiting the number of taps from the tap center, the compensation accuracy is degraded.

[0006] Methods have been proposed to determine the optimal number of taps based on the equalization performance of an adaptive equalizer. For example, methods have been proposed that detect the group delay time difference between the polarizations of polarization-multiplexed light and determine the number of taps in the adaptive equalizer based on this group delay time difference (see, for example, Patent Document 1), and methods that control the number of taps based on the error between the received pilot signal and the original pilot signal (see, for example, Patent Document 2).

[0007] Prior art literature

[0008] Patent Literature

[0009] Patent Document 1: Japanese Patent Application Laid-Open No. 2018-182620

[0010] Patent Document 2: Japanese Patent Application Laid-Open No. 2010-118817 Summary of the Invention

[0011] Problems to be solved by the invention

[0012] However, the method of determining the number of taps based on desired equalization performance has a problem in that it is difficult to achieve low power consumption by reducing the number of taps while obtaining high equalization performance.

[0013] The present disclosure has been made to solve the above-mentioned problems, and an object thereof is to provide an adaptive equalizer, an adaptive equalization method, and an optical communication system that can achieve low power consumption while achieving high equalization performance.

[0014] Means used to solve problems

[0015] The adaptive equalizer disclosed in the present invention is characterized in that it comprises: an adaptive filter having a first digital filter and a first tap coefficient updater, wherein the first digital filter compensates for the distortion of the input signal, and the first tap coefficient updater adaptively updates the tap coefficients of the first digital filter through a convergence action according to the waveform state of the input signal; a second digital filter that compensates for the distortion of the input signal; a second tap coefficient updater that adaptively updates the tap coefficients of the second digital filter through a convergence action according to the waveform state of the input signal; and a tap coefficient control circuit that updates the tap coefficients of the second digital filter through a convergence action according to the waveform state of the input signal. The tap coefficients updated by the digital updater are set as the initial values ​​of the convergence action of the tap coefficients of the first digital filter updated by the first tap coefficient updater. For each tap coefficient set to the initial value, they are arranged in descending order according to the degree of contribution to the convergence action of the first tap coefficient updater. The tap coefficients above the upper specified number are judged to be valid, and the tap coefficients less than the specified number are judged to be invalid. The tap coefficients of the first digital filter corresponding to the tap coefficients judged to be invalid are set to zero and are not used for the calculation of the first tap coefficient updater until the next judgment result comes out.

[0016] Effects of the Invention

[0017] According to the present disclosure, it is possible to achieve low power consumption while obtaining high balanced performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is a block diagram showing the configuration of the transmission side of the optical communication system according to the embodiment.

[0019] Figure 2 This is a block diagram showing the configuration of the receiving side of the optical communication system according to the embodiment.

[0020] Figure 3 This is a diagram showing the structure of the adaptive equalizer according to the first embodiment.

[0021] Figure 4 : is a diagram showing the structure of the first digital filter.

[0022] Figure 5 It is a diagram showing the details of each FIR filter.

[0023] Figure 6 This is a diagram showing an example of the adaptive equalizer according to the first embodiment.

[0024] Figure 7 This is a flowchart of the tap coefficient validity / invalidity determination algorithm of the tap coefficient control circuit.

[0025] Figure 8This is a diagram showing the operation timing of the adaptive equalizer according to the first embodiment.

[0026] Figure 9 This is a diagram showing the structure of an adaptive equalizer according to the second embodiment.

[0027] Figure 10 This is a diagram showing an example of an adaptive equalizer according to the second embodiment. DETAILED DESCRIPTION

[0028] An adaptive equalizer, an adaptive equalization method, and an optical communication system according to an embodiment will be described with reference to the accompanying drawings. Identical or corresponding components are denoted by the same reference numerals, and duplicate descriptions may be omitted.

[0029] Implementation method 1.

[0030] Figure 1 1 is a block diagram showing a configuration of a transmission side of an optical communication system according to an embodiment. The optical communication system according to this embodiment includes a transmission signal processor 1 and an optical transmitter 2 on the transmission side.

[0031] The transmit signal processor 1 is a circuit that performs predetermined processing on input data. Specifically, the transmit signal processor 1 separates the input data into horizontally polarized data and vertically polarized data, and performs processing such as error correction coding, band-limiting filtering, and modulation mapping on each data. The processed horizontally polarized and vertically polarized signals are then output to the optical transmitter 2.

[0032] Optical transmitter 2 is a circuit that converts horizontally polarized and vertically polarized signals into optical signals and transmits the converted optical signals. It includes a signal light source 2a (signal LD), two 90° combiners 2b and 2c, and a polarization combiner 2d. The 90° combiners 2b and 2c modulate the output light of signal light source 2a with the horizontally polarized and vertically polarized signals, respectively, thereby converting these signals into optical signals. Polarization combiner 2d combines the converted horizontally polarized and vertically polarized signals. The combined optical signal is transmitted to the receiving end via optical fiber transmission path 3.

[0033] Figure 2 The optical communication system of this embodiment includes an optical receiver 4, an AD converter 5, a chromatic dispersion compensator 6, an adaptive equalizer 7, and a decoder 8 on the receiving side.

[0034] The optical receiver 4 is a circuit that receives an optical signal, converts the received optical signal into an electrical signal, and outputs it. The optical receiver 4 includes a polarization separator 4a, a local oscillator light source 4b (local oscillator LD), two 90° hybrid circuits 4c and 4d, and a photoelectric converter 4e. The polarization separator 4a separates the optical signal into two orthogonal polarization components, namely, X polarization (horizontal polarization) and Y polarization (vertical polarization). The 90° hybrid circuits 4c and 4d combine the output light of the local oscillator light source 4b with the polarizations of the optical signal output from the polarization separator 4a, further separating the polarizations of the optical signal into an in-phase (I) component and a quadrature (Q) component. The photoelectric converter 4e converts the components of the optical signal output from the 90° hybrid circuits 4c and 4d into electrical signals, which are then output as an X polarization signal and a Y polarization signal. Hereinafter, the X polarization signal and the Y polarization signal are referred to as received signals. The above-described structure for obtaining the X polarization signal and the Y polarization signal is merely an example and is not limited to the above-described structure.

[0035] A / D converter 5 converts the received signal output from optical receiver 4 into a digital signal. As the optical signal propagates through optical fiber transmission path 3, the signal waveform is distorted by chromatic dispersion. Chromatic dispersion compensator 6 estimates the magnitude of this distortion based on the received signal output from A / D converter 5 and compensates for the distortion of the received signal caused by chromatic dispersion.

[0036] Furthermore, when the X-polarized signal and the Y-polarized signal are combined and transmitted on the transmitting side, and separated on the receiving side, polarization mode dispersion causes polarization fluctuations, distorting the signal waveforms. Adaptive equalizer 7 performs equalization processing to compensate for the distortion caused by polarization fluctuations in the received signal output from chromatic dispersion compensator 6. Polarization separation is initially performed by optical receiver 4, but is further refined by adaptive equalizer 7. Decoder 8 decodes the received signal output from adaptive equalizer 7 to reconstruct the original data (i.e., the input data to transmit signal processor 1).

[0037] As described above, the adaptive equalizer 7 primarily compensates for distortion caused by polarization fluctuations, but it can also compensate for distortion caused by frequency fluctuations or phase fluctuations. Therefore, the adaptive equalizer 7 is useful not only for transmitting and receiving signals synthesized from the transmitting side by combining the X- and Y-polarization signals, but also for transmitting and receiving signals using only one polarization. Therefore, the adaptive equalizer of the present invention is not limited to transmitting and receiving signals synthesized from the transmitting side by combining the X- and Y-polarization signals, but is also useful for transmitting and receiving signals using only one polarization.

[0038] Figure 3This is a block diagram showing an adaptive equalizer according to Embodiment 1. The adaptive equalizer 7 includes an adaptive filter 9, a second digital filter 10, a second tap coefficient updater 11, and a tap coefficient control circuit 12. The adaptive filter 9 includes a first digital filter 13 and a first tap coefficient updater 14. Here, "tap coefficients" refer to coefficients that are multiplied by the signals of each of the multiple taps that constitute the digital filter. In this specification, "tap coefficients" generally refers to a set of coefficients. However, individual tap coefficients may be shown for explanation.

[0039] The first digital filter 13 compensates the input signal. The compensated result is supplied to the first tap coefficient updater 14. The polarization state of the input signal changes with time. Therefore, the first tap coefficient updater 14 adaptively updates the tap coefficients of the first digital filter 13 according to the polarization state of the input signal through CMA. Figure 3 In FIG1 , the line supplying the input signal of the first digital filter 13 to the first tap coefficient updater 14 is omitted. However, in the tap coefficient convergence operation described later, it is shown that the tap coefficient convergence operation is actually performed based on the input signal of the first digital filter 13. The updated tap coefficient is set in the first digital filter 13. In CMA, the tap coefficient is successively updated and converged to a predetermined value so that the output of the first digital filter 13 becomes the original value.

[0040] In this specification, the term "updating" of tap coefficients is used both for updating during a convergence operation when obtaining tap coefficients for a polarization state at a certain point in time and for updating performed for each convergence operation at the timing of a change in the polarization state.

[0041] The output of the first digital filter 13 is supplied as a compensated reception signal to the Figure 2 The decoder 8 shown. The input signal may be both a horizontally polarized signal and a vertically polarized signal, or may be either one. In the following description, as a common structure of coherent optical communication, the case of transmission using both horizontally polarized signals and vertically polarized signals is described, but this embodiment is also applicable to the case of only one signal.

[0042] The tap coefficients of the adaptive filter 9 can usually be calculated as a Wiener solution by creating a matrix equation. However, this requires very complex calculations, so a simple method for calculating the coefficients is usually a successive update algorithm. This is one method for convergence. The algorithm for calculating the tap coefficients of the adaptive filter 9 is not limited to CMA, and various successive update algorithms can also be used, such as RDE (Radius Directed Equalization), which is another blind equalization method. Furthermore, a successive update algorithm such as RLS (Recursive Least-Squares) or LMS (Least Mean Square) can also be used. In this successive update algorithm, a known signal such as a training signal or pilot signal is inserted into the optical signal on the transmitting side. The tap coefficients are then updated at each step size to minimize the error between the transmitted known signal and the true value of the known signal (the value set on the transmitting side) (e.g., the amplitude difference in the IQ plane, the sum of the squares of the IQ amplitudes). In CMA, the tap coefficients are also updated so as to minimize the error between the output of the digital filter and the original expected value (the “expected value” can be easily estimated as the desired value of the amplitude in the case of a constant envelope).

[0043] The second digital filter 10 is connected in parallel with the first digital filter 13. The second digital filter 10 also compensates the input signal in the same manner as the first digital filter 13. In addition, the second tap coefficient updater 11 also operates in the same manner as the first tap coefficient updater 14, adaptively updating the tap coefficients of the second digital filter 10 through CMA and according to the polarization state of the input signal. The tap coefficients updated in the convergence operation of the CMA are set in the second digital filter 10 at each update. By repeating this update, the tap coefficients converge to a specified value. However, the output of the second digital filter 10 is different from the output of the first digital filter 13 and is not supplied to the decoder 8 as a compensated received signal. The output of the second digital filter 10 is only used to calculate the tap coefficients of the first digital filter 13.

[0044] The tap coefficient control circuit 12 sets the tap coefficients converged by the second tap coefficient updater 11 as the initial values ​​for the first digital filter 13, which is updated by the first tap coefficient updater 14, during the convergence operation. At this time, the tap coefficients converged by the second tap coefficient updater 11, i.e., the tap coefficients set to the initial values, are determined to be valid or invalid for each tap. The tap coefficients of the first digital filter corresponding to the tap coefficients determined to be invalid are forcibly set to zero. Furthermore, during the convergence operation of the first tap coefficient updater 14, the tap coefficients remain set to zero until the next determination result is obtained. For taps whose tap coefficients are set to zero, no multiplication is performed in the first digital filter 13.

[0045] Figure 4 This is a diagram showing the structure of the first digital filter. The first digital filter 13 includes butterfly-shaped FIR (Finite Impulse Response) filters FIR_A, FIR_B, FIR_C, and FIR_D. Each FIR filter has N taps. However, the number of taps in the FIR filters may differ. Furthermore, the second digital filter 10 can be configured similarly to the first digital filter 13.

[0046] Figure 5 The figure shows the details of each FIR filter. FIR_A is a filter for horizontal polarization signal, and has a tap coefficient W HH_1 ~W HH_N FIR_B is a filter for the influence from vertical polarization signal to horizontal polarization signal, with tap coefficient W VH_1 ~W VH_N FIR_C is a filter for the influence from horizontal polarization signal to vertical polarization signal, with tap coefficient W HV_1 ~W HV_N FIR_D is a filter for vertically polarized signals with a tap coefficient of W. VV_1 ~W VV_N N is the number of taps. Each tap coefficient is sequentially multiplied by the delayed input signal, and their sum is output as the filter result. Z^(-1) represents the successive delays.

[0047] The first digital filter 13 outputs the sum of the filtering results of FIR_A for the horizontal polarization signal and FIR_B for the vertical polarization signal as the compensation output for the horizontal polarization signal, and outputs the sum of the filtering results of FIR_C for the horizontal polarization signal and FIR_D for the vertical polarization signal as the compensation output for the vertical polarization signal. Furthermore, the first digital filter 13 is not limited to a butterfly-type structure and may also be a structure without FIR_B and FIR_C.

[0048] The first digital filter 13 and the first tap coefficient updater 14 constitute the adaptive filter 9. At this time, the convergence operation of the tap coefficients of FIR_A, FIR_B, FIR_C, and FIR_D is expressed as follows.

[0049] W HH (n+1)=W HH (n)+μe H (n)H out (n) Hin ＊ (n)

[0050] W VH (n+1)=W VH (n)+μe V (n)H out (n) Vin ＊ (n)

[0051] W HV (n+1)=W HV (n)+μe H (n)V out (n) Hin ＊ (n)

[0052] W VV (n+1)=W VV (n)+μe V (n)V out (n) Vin ＊ (n)

[0053] Here, n is a value indicating the update order in the successive update algorithm. HH (n) represents the tap coefficient W in case of updating order n HH_1 ~W HH_N .Tap coefficient W VH (n) represents the tap coefficient W in case of updating order n VH_1 ~W VH_N .Tap coefficient W HV (n) represents the tap coefficient W in case of updating order n HV_1 ~W HV_N .Tap coefficient WVV (n) represents the tap coefficient W in case of updating order n VV_1 ~W VV_N . μ represents the step size of the update algorithm. e H (n) represents the error between the filter output and the expected value in horizontal polarization. V (n) represents the error between the filter output and the expected value in vertical polarization. out (n) represents the filter output in horizontal polarization. Hin(n) represents the filter input in horizontal polarization. V out (n) represents the filter output in vertical polarization. Vin(n) represents the filter input in vertical polarization. * represents conjugate or complex conjugate. Figure 4 、 5 In FIG, a line for inputting an input signal to the first tap coefficient updater 14 is omitted. In addition, the signal and the tap coefficient are represented by complex numbers.

[0054] The above equation is only one example of a successive update algorithm, and the equations representing successive update algorithms are not limited to the above. Any equation that updates the value using a step size will suffice. This step size determines the tracking and noise immunity of the digital signal processing during the adaptive control of the tap coefficients. A larger step size improves the tracking of the digital signal processing and enhances the reception robustness against high-speed polarization state fluctuations. However, during low-speed polarization state fluctuations, the transmission characteristics deteriorate due to the influence of noise.

[0055] By using the above-mentioned successive update algorithm, the tap coefficients are updated in sequence according to the update order n, and the tap coefficients eventually converge. The convergence condition is determined by the number of update orders n or the error between the filter output and the desired value. Then, through the convergence of the above-mentioned algorithm, the tap coefficients W of each FIR filter are obtained. HH_1 ~W HH_N 、W VH_1 ~W VH_N 、W HV_1 ~W HV_N , and W VV_1 ~W VV_N .

[0056] In the second digital filter 10 and the second tap coefficient updater 11, the tap coefficients W of each FIR filter are obtained in the same manner as above. HH_1 ~W HH_N 、W VH_1 ~W VH_N 、W HV_1 ~W HV_N , and W VV_1 ~W VV_NFurthermore, the intervals between the update orders n of the first tap coefficient updater 14 and the second tap coefficient updater 11 may differ from each other and do not need to coincide with the symbol period (the period at which data values ​​are changed or updated). Furthermore, the convergence conditions of the first tap coefficient updater 14 and the second tap coefficient updater 11 may differ from each other. Furthermore, the symbol period differs from the timing Z^(-1) of the sequential delays of the FIR filter.

[0057] Figure 6 is a diagram showing an example of an adaptive equalizer according to Embodiment 1. In this example, Figure 3 The first digital filter 13 and the second digital filter 10 of the adaptive equalizer 7 and the wiring therebetween are shown in detail.

[0058] The operation of the adaptive equalizer of Embodiment 1 will be described below. The tap coefficients obtained by the CMA of the second digital filter 10 and the second tap coefficient updater 11 are supplied to the tap coefficient control circuit 12. These tap coefficients are W HH_1 ~W HH_N 、W VH_1 ~W VH_N 、W HV_1 ~W HV_N , and W VV_1 ~W VV_N .

[0059] In the tap coefficient control circuit 12, based on an algorithm described later, all of the above taps are determined to be valid or invalid. Taps determined to be invalid are changed to zero and set as the initial values ​​for the CMA convergence operation of the first tap coefficient updater 14 in the taps of the first digital filter 13. At this time, tap coefficients that contribute significantly to the convergence of the tap coefficients are determined to be valid, while other tap coefficients or tap coefficients that contribute less significantly to the convergence of the tap coefficients are determined to be invalid.

[0060] After the initial values ​​are set, the tap coefficients converge in the first digital filter 13 and the first tap coefficient updater 14 using CMA. In the first embodiment, the values ​​of the first digital filter 13 are also output during the tap coefficient convergence operation, that is, during convergence. Furthermore, taps set to zero by the tap coefficient control circuit 12 remain set to zero, regardless of the update result of the first tap coefficient updater 14, until the next convergence result is obtained. Tap coefficients set to zero are adjusted so that they are not used in calculations by the first tap coefficient updater 14. Furthermore, a circuit design is employed that minimizes power consumption even when used.

[0061] For example, the output of FIR_A of the first digital filter 13 is expressed by the following equation at the time of n-th update.

[0062] H out (n) = H in_1 (n)·W HH_1 (n)+H in_2 (n)·W HH_2 (n)+H in_3 (n)·W HH_3 (n)+····H in_N (n)·W HH_N (n)

[0063] Here, H in_1 (n)~H in_N (n) is the horizontal polarization signal H in (n) The signal after sequential delay in FIR_A. In the initial value, n = 0 times.

[0064] Here, it is considered that the tap coefficients W are set from the tap coefficient control circuit 12 to the first digital filter 13. HH_1 (0)~W HH_N (0) is the initial value. At this time, it is assumed that the real part and the imaginary part of the second tap coefficient of FIR_A are both set to zero, that is, W HH_2 When (0)=0+j·0, it becomes

[0065] H out (0)=H in_1 (0)·W HH_1 (0)+H in_2 (0)·W HH_2 (0)+H in_3 (0)·W HH_3 (0)+····H in_N (0)·W HH_N (0)

[0066] =H in_1 (0)·W HH_1 (0)+0+H in_3 (0)·W HH_3 (0)+····H in_N (0)·W HH_N (0).

[0067] Then, calculate W using the CMA algorithm HH_1 (1)~W HH_N (1). In this case, even if a finite value is calculated as W HH_2 The value of (1) is also set to W HH_2The tap coefficients of FIR_A are calculated using (n) = 0 + j·0. That is, in the second tap, the results of both the real and imaginary multiplications are initially set to zero. This continues until the next initial value is set by the tap coefficient control circuit 12. The tap coefficients are set to zero independently for each of the four FIR filters, the real part, and the imaginary part.

[0068] Next, the tap coefficient validity / invalidity determination algorithm of the tap coefficient control circuit 12 will be described. Figure 7 This is a flowchart of the tap coefficient validity / invalidity determination algorithm of the tap coefficient control circuit.

[0069] The tap coefficient validity / invalidity determination algorithm is an algorithm for determining whether a calculated tap coefficient is valid or invalid (hereinafter referred to as validity / invalidity determination). Furthermore, the tap coefficient control circuit 12 performs the tap coefficient validity / invalidity determination each time the tap coefficient successive update algorithm completes convergence via the second digital filter 10 and the second tap coefficient updater 11 and the tap coefficient is determined.

[0070] The following describes the operation of each step. First, in step 1, all tap coefficients W of the second digital filter 10 are obtained by the second digital filter 10 and the second tap coefficient updater 11 based on the successive update algorithm. HH_1 ~W HH_N 、W VH_1 ~W VH_N 、W HV_1 ~W HV_N , and W VV_1 ~W VV_N These values ​​are obtained after the convergence of the successive update algorithm is complete. Each tap coefficient is expressed as a coordinate on the IQ plane (I value + jQ value). I + jQ is a so-called complex number.

[0071] Next, in step 2, for all the tap coefficients of the second digital filter 10 obtained in step 1, the I value and Q value on the IQ plane are aggregated and arranged in descending order of absolute value. In this case, it is also possible to perform valid / invalid judgment only for the I value and valid / invalid judgment only for the Q value. However, in experimental verification, the valid / invalid judgment performed by aggregating the I value and the Q value obtained higher performance than the judgment performed separately. In addition, as Figure 6 As shown, when processing both the horizontal polarization signal and the vertical polarization signal, the I value and Q value of the four FIR filters can be summarized to perform a valid / invalid judgment. In this case, the judgment performed by summarizing the four FIR filters is also higher than the judgment performed according to each filter.

[0072] Next, in step 3 (first determination), for the tap coefficients arranged in order of absolute value, the tap coefficients of the upper specified number M or more are provisionally determined to be valid, and the tap coefficients of the lower specified number M are provisionally determined to be invalid. Here, the larger the absolute value of the tap coefficient, the greater the degree to which it contributes to the calculation (convergence operation) of the tap coefficient updater 14. Therefore, the tap coefficients updated by the second tap coefficient updater 11 are arranged in descending order of the degree to which they contribute to the calculation (convergence operation) of the first tap coefficient updater 14, and the tap coefficients of the upper specified number or more are determined to be valid, and the tap coefficients of the lower specified number are determined to be invalid.

[0073] In step 4-1 (second judgment), if the tap coefficient provisionally judged valid in step 3 was valid A or valid B in the previous second judgment, it is finally judged valid A (Final validity "A" judgment). On the other hand, if it was invalid in the previous second judgment, the process proceeds to step 5-1.

[0074] In step 4-2 (second judgment), if the tap coefficient provisionally judged invalid in step 3 was also invalid in the previous second judgment, it is finally judged invalid (final invalidity judgment). On the other hand, if it was valid in the previous second judgment, the process proceeds to step 5-2.

[0075] In step 5-1 (second judgment), for a tap coefficient that was invalid in the previous second judgment in step 4-1, if its absolute value is greater than threshold value T1, it is finally judged as valid B (final validity B judgment). On the other hand, if its absolute value is less than threshold value T1, it is finally judged as invalid (final invalid judgment).

[0076] Next, in step 5-2 (second determination), for the tap coefficients that were valid A or valid B in the previous second determination in step 4-2, if their absolute value is greater than threshold value T2, they are finally determined to be valid A (final valid A determination). On the other hand, if their absolute value is less than threshold value T2, they are finally determined to be invalid (final invalid determination). The threshold value T1 in step 5-1 and the threshold value T2 in step 5-2 may be different or the same. The above algorithm performs valid / invalid determinations on all tap coefficients.

[0077] In step 6-1, the tap coefficient control circuit 12 does not use the tap coefficients finally determined to be valid A. Instead, it sets the last update result of the first tap coefficient updater 14 as the initial value of the tap coefficients of the first digital filter 13 updated by the first tap coefficient updater 14. Since the last determination was valid, the set coefficients are not zero. Therefore, there is no need to set the new result of the second tap coefficient update as the initial value.

[0078] In step 6-2, the tap coefficient control circuit 12 sets the tap coefficient finally determined to be valid B as the initial value of the tap coefficient of the first digital filter 13 updated by the first tap coefficient updater 14. Since the coefficient was previously determined to be invalid, the set coefficient was zero. Therefore, since the setting is newly performed this time, it is necessary to set the initial value.

[0079] In step 6-3, the tap coefficient control circuit 12 sets the initial value of the tap coefficient of the first digital filter 13 corresponding to the tap coefficient determined to be invalid to zero. This setting of zero is maintained until the next determination is made and a new initial value is set. Furthermore, if the tap coefficient is set to zero, the multiplication of the tap coefficient and the addition of the multiplication results are not performed. This can be achieved by generating an equation in which the filter calculation formula is pre-set to zero and inputting the updated tap coefficient into this equation.

[0080] As described above, the tap coefficient validity / invalidity determination algorithm uses the previous determination result and an arbitrarily set threshold value to determine whether the tap coefficient is valid or invalid. However, the determination method is not limited to the above method. Various determination algorithms can be constructed by combining past determination results with multiple threshold values.

[0081] Figure 8 This figure shows the operating timing of the adaptive equalizer in Embodiment 1. (a) shows the symbol timing of the received signal. (b) shows the operating timing of the second tap coefficient updater 11. (c) shows the timing of the valid / invalid determination of the tap coefficient control circuit 12. (d) shows the operating timing of the first tap coefficient updater 14. The symbol here refers to the period at which the data of the received signal changes or is updated.

[0082] As shown in (a) and (b), the second tap coefficient updater 11 performs convergence operation 2-1 from symbols p1 to p2 of the received signal. Similarly, the second tap coefficient updater 11 performs convergence operation 2-2 from symbols p2 to p3, and the second tap coefficient updater 11 performs convergence operation 2-3 from symbols p3 to p4. The tap coefficients of the second digital filter 10 obtained in each of these convergence operations are determined to be valid or invalid by the tap coefficient control circuit 12, as shown in (c). The results of each determination are reflected in the convergence operation of the first tap coefficient updater 14, as shown in (d). The number of symbols from symbols p1 to p2, the number of symbols from symbols p2 to p3, and the number of symbols from symbols p3 to p4 may be the same or different. Furthermore, the convergence operations 1-1, 1-2, and 1-3 of the first tap coefficient updater 14 shown in (d) are not necessarily the same depending on the convergence status.

[0083] For example, the tap coefficients determined in convergence operation 2-1 of the second tap coefficient updater 11 are determined as decision 2 by the tap coefficient control circuit 12. The tap coefficients determined as decision 2 are reflected in convergence operation 1-2 of the first tap coefficient updater 14. Specifically, the determined tap coefficients are set as initial values ​​for convergence operation 1-2, and convergence operation 1-2 is performed. After convergence operation 1-2 of the first tap coefficient updater 14 converges, the convergence result is continuously used until the next convergence operation 1-3 begins.

[0084] In the above-described operation, if the tap coefficients obtained in convergence operation 2-1 are the same as the tap coefficients obtained in the previous convergence operation, there is no need to perform control (including valid / invalid determination) by the tap coefficient control circuit 12, and the determination result (determination 1) used in convergence operation 1-1 by the first tap coefficient updater 14 can be continued to be used until the next determination (determination 3) is made (not shown). Furthermore, if determination 2 is the same as determination 1, no further control is performed, and the determination result (determination 1) used in convergence operation 1-1 by the first tap coefficient updater 14 is continued to be used until the next determination (determination 3) is made (not shown).

[0085] Furthermore, as shown from the decision 3 to the convergence operation 1-3, the validity / invalidity decision result of the tap coefficient control circuit 12 does not need to be reflected immediately. A slight delay may occur due to circuit delay and the like.

[0086] In addition, the symbols used by the second tap coefficient updater 11 and the first tap coefficient updater 14 do not necessarily need to refer to the same symbol. Figure 8For example, the symbols used in the coefficient update are the 1st symbol, the 11th symbol, and the 21st symbol, but the symbols used by the 1st tap coefficient updater 14 are the 2nd symbol, the 12th symbol, and the 22nd symbol.

[0087] As described above, in this embodiment, the tap coefficients converged by the second tap coefficient updater 11 are determined to be valid or invalid. The tap coefficients of the first digital filter 13 corresponding to the tap coefficients determined to be invalid are set to zero and not used in the calculations of the first tap coefficient updater 14 until the next determination result is obtained. This reduces power consumption. Furthermore, the tap coefficients updated by the second tap coefficient updater 11 are arranged in descending order of their contribution to the calculations of the first tap coefficient updater 14. Tap coefficients above a specified number are determined to be valid, while tap coefficients below the specified number are determined to be invalid. This prevents calculations from being performed for taps whose values ​​are determined to contribute little to the equalization process. This allows for high equalization performance while achieving low power consumption.

[0088] Implementation method 2.

[0089] Figure 9 This is a diagram showing the structure of the adaptive equalizer according to the second embodiment. Figure 3 The adaptive equalizer 7 of the illustrated first embodiment is different in that a third digital filter 15 having the same configuration as the first digital filter 13 is connected in parallel with the first digital filter 13 .

[0090] In the first embodiment, the output of the first digital filter 13 is supplied to the first tap coefficient updater 14, and is also supplied to the first tap coefficient updater 14 as the output of the adaptive filter 9. Figure 2 In contrast, in this embodiment, the output of the third digital filter 15 becomes the output of the adaptive filter 9 , and the output of the first digital filter 13 is supplied only to the first tap coefficient updater 14 .

[0091] At this time, the tap coefficients of the third digital filter 15 are set to the same values ​​as those of the first digital filter 13. However, the first tap coefficient updater 14 maintains the results of the previous convergence operation during the convergence operation and sets a new convergence result after the convergence operation is completed. As a result, the tap coefficients of the third digital filter 15 do not change during the convergence operation of the first digital filter 13 and the first tap coefficient updater 14, thereby obtaining a stable compensation output. Other than this, the structure and operation of the second digital filter 10, the second tap coefficient updater 11, and the tap coefficient control circuit 12 are identical to those of the first embodiment.

[0092] Figure 10 : is a diagram showing an example of an adaptive equalizer according to Embodiment 2. In this example, Figure 9 The first digital filter 13, the second digital filter 10, the third digital filter 15 of the adaptive equalizer 7 and the wiring therebetween are shown in detail.

[0093] The operation of the adaptive equalizer of the second embodiment will be described below. The tap coefficients obtained by the CMA of the second digital filter 10 and the second tap coefficient updater 11 are supplied to the tap coefficient control circuit 12. These tap coefficients are W HH_1 ~W HH_N 、W VH_1 ~W VH_N 、W HV_1 ~W HV_N , and W VV_1 ~W VV_N .

[0094] The tap coefficient control circuit 12 determines whether all the taps are valid or invalid based on the algorithm described above. The tap coefficient control circuit 12 then changes the taps determined to be invalid to zero and sets the zero value in the taps of the first digital filter 13 as the initial value for the CMA operation of the first tap coefficient updater 14.

[0095] After setting, the tap coefficients converge via CMA in the first digital filter 13 and the first tap coefficient updater 14. Taps set to zero by the tap coefficient control circuit 12 continue to be set to zero, regardless of the update result of the first tap coefficient updater 14, until the next convergence result is obtained. For taps set to zero, multiplication and addition are not performed. Furthermore, a circuit design is employed that minimizes power consumption, even if these operations are performed.

[0096] Furthermore, in this configuration, while the tap coefficients are converging, the third digital filter 15 continues to operate according to the tap coefficients obtained in the previous convergence operation, and its output becomes the output of the adaptive equalizer 7. After the convergence operation of the first digital filter 13 and the first tap coefficient updater 14 is completed, the tap coefficients thus obtained are set in the third digital filter 15. Thereafter, the third digital filter 15 continues to operate according to the previous tap coefficients until the next convergence result (tap coefficient) is set by the first tap coefficient updater 14.

[0097] As described above, the adaptive equalizer 7 of Embodiment 2 does not calculate taps with small absolute values ​​of tap coefficients. This allows for high equalization performance while also reducing power consumption. Furthermore, the tap coefficients of the third digital filter 15, through which the main signal passes, do not fluctuate during convergence, resulting in a stable compensation output.

[0098] In addition, the third digital filter becomes a digital filter that is independent of convergence. Therefore, compared with the first digital filter and the second digital filter used in coefficient updating, the bit resolution of the filter coefficients for all main signal inputs can be reduced to a level that does not lose convergence errors and does not cause signal degradation, and low power consumption can also be achieved.

[0099] Furthermore, in the aforementioned successive update algorithm, a step size is defined as an indicator of the tap coefficient update amplitude. The step size of the successive update algorithm of the first tap coefficient updater 14 can also be changed based on the ratio of the number of tap coefficients determined to be invalid to the total number of tap coefficients updated by the second tap coefficient updater 11. For example, if the number of tap coefficients determined to be invalid is half the total number of tap coefficients, the error between the filter output and the desired value is also halved, so the step size needs to be doubled to maintain the tracking speed. Therefore, as the ratio of the number of tap coefficients determined to be invalid to the total number of tap coefficients increases, the error added to the tap coefficients becomes smaller, and therefore the step size needs to be increased accordingly.

[0100] Furthermore, as described above, when a tap coefficient determined to be invalid is set to zero, multiplication is not performed until the next determination is made in the first digital filter 13. Specific methods for this include not feeding back the number of the tap coefficient determined to be invalid from the first tap coefficient updater 14, setting the feedback value to zero, and setting the value to zero after the feedback.

[0101] Furthermore, the coefficient update timing of the first tap coefficient updater 14 and the second tap coefficient updater 11 can also be performed every symbol. In this case, the coefficient update can also be performed every symbol, and the validity / invalidity determination can also be performed every several tens of symbols.

[0102] Description of Reference Numerals

[0103] 4 optical receiver, 5 AD converter, 7 adaptive equalizer, 9 adaptive filter, 10 second digital filter, 11 second tap coefficient updater, 12 tap coefficient control circuit, 13 first digital filter, 14 first tap coefficient updater, 15 third digital filter.

Claims

1. An adaptive equalizer, characterized in that: The adaptive equalizer has: An adaptive filter comprising a first digital filter and a first tap coefficient updater, wherein the first digital filter compensates for distortion of an input signal, and the first tap coefficient updater adaptively updates the tap coefficients of the first digital filter through a convergence action according to a polarization state of the input signal; a second digital filter for compensating for distortion of the input signal; a second tap coefficient updater adapted to adaptively update the tap coefficients of the second digital filter through a convergence action according to the polarization state of the input signal; as well as A tap coefficient control circuit sets the tap coefficients converged in the second tap coefficient updater as the initial values ​​of the tap coefficients of the first digital filter updated by the first tap coefficient updater, arranges the tap coefficients converged in the second tap coefficient updater in descending order according to the degree of their contribution to the convergence action of the tap coefficient update of the first tap coefficient updater, judges the tap coefficients above the upper specified number as valid, judges the tap coefficients less than the specified number as invalid, sets the tap coefficients of the first digital filter corresponding to the tap coefficients judged to be invalid to zero and does not use them for calculation by the first tap coefficient updater until the next judgment result comes out.

2. The adaptive equalizer according to claim 1, wherein As a first determination, the tap coefficient control circuit aggregates the I values ​​and Q values ​​on the IQ plane for the plurality of tap coefficients converged in the second tap coefficient updater and arranges them in descending order of absolute value, determines that the tap coefficients of the upper order are greater than the specified number as valid, and determines that the tap coefficients of the upper order are less than the specified number as invalid. The tap coefficient having a larger absolute value contributes more to the convergence of the tap coefficient update operation of the first tap coefficient updater.

3. The adaptive equalizer according to claim 2, wherein: As a second judgment of judging the tap coefficient that has undergone the first judgment as any one of valid A, valid B, and invalid, the tap coefficient control circuit judges the tap coefficient that has undergone the first judgment as valid as valid A if it was judged as valid A or valid B in the previous second judgment; judges it as valid B if the tap coefficient is greater than the first threshold and was judged as invalid in the previous second judgment; and judges it as invalid if the tap coefficient is less than the first threshold and was judged as invalid in the previous second judgment. Moreover, for the tap coefficient that has been judged as invalid in the first judgment, it is judged as invalid if it was also judged as invalid in the previous second judgment; judges it as valid A if the tap coefficient is greater than the second threshold and was judged as valid A or valid B in the previous second judgment; and judges it as invalid if the tap coefficient is less than the second threshold and was judged as valid A or valid B in the previous second judgment.

4. The adaptive equalizer according to claim 3, wherein: The tap coefficient control circuit replaces the tap coefficient determined to be the valid A, sets the last update result of the first tap coefficient updater as the initial value of the tap coefficient of the first digital filter updated by the first tap coefficient updater, and sets the tap coefficient determined to be the valid B as the initial value of the tap coefficient of the first digital filter updated by the first tap coefficient updater.

5. The adaptive equalizer according to any one of claims 1 to 4, characterized in that The first tap coefficient updater updates the tap coefficients of the first digital filter by a successive update algorithm. The second tap coefficient updater updates the tap coefficients of the second digital filter using a successive update algorithm.

6. The adaptive equalizer according to claim 5, wherein: The adaptive filter further includes a third digital filter that compensates for distortion of the input signal using tap coefficients obtained by convergence through an updating operation of the first tap coefficient updater.

7. The adaptive equalizer according to claim 5, wherein: The step size of the successive updating algorithm of the first tap coefficient updater is changed according to the ratio of the number of tap coefficients determined to be invalid to the total number of tap coefficients updated by the second tap coefficient updater.

8. The adaptive equalizer according to claim 7, wherein: The step size increases as the ratio becomes larger.

9. An adaptive equalization method, characterized in that: The adaptive equalization method comprises the following steps: The first digital filter compensates for the distortion of the input signal; A first tap coefficient updater adaptively updates the tap coefficients of the first digital filter through a convergence action according to the polarization state of the input signal; The second digital filter compensates for the distortion of the input signal; A second tap coefficient updater adaptively updates the tap coefficients of the second digital filter through a convergence action according to the polarization state of the input signal; as well as The tap coefficient control circuit sets the tap coefficient converged by the second tap coefficient updater as the initial value of the tap coefficient of the first digital filter updated by the first tap coefficient updater. The tap coefficient control circuit arranges the tap coefficients that converge in the second tap coefficient updater in descending order according to the degree of convergence action that helps the tap coefficient update of the first tap coefficient updater, and judges the tap coefficients above the upper specified number as valid, and judges the tap coefficients less than the specified number as invalid, and sets the tap coefficients of the first digital filter corresponding to the tap coefficients judged to be invalid to zero and does not use them for calculation of the first tap coefficient updater until the next judgment result comes out.

10. An optical communication system, characterized in that: The optical communication system comprises: an optical receiver, which receives an optical signal and converts the received optical signal into an electrical signal; an AD converter that converts a signal output from the optical receiver into a digital signal; and The adaptive equalizer according to any one of claims 1 to 4, wherein the equalization process is performed to compensate for distortion of the signal output from the AD converter caused by polarization variation.

Images