JP4189747B2 - Signal processing device - Google Patents

Description

  The present invention relates to a signal processing device, and more particularly to a signal processing device that can compensate for nonlinear distortion of an equalized waveform.

  Conventionally, as a method for detecting a reproduction signal which is digital data reproduced from a recording medium, there has been a peak detection method for specifying a peak of voltage amplitude. However, with the recent improvement in information processing technology, the recording density of recording media has increased, and intersymbol interference has occurred in the reproduced signal. In some cases, accurate detection was not possible with the peak detect method. .

  Therefore, there is a PRML (Partial Response Maximum Likelihood) method that combines a Partial Response (PR) method that actively uses Inter Symbol Interference and a Maximum Likelihood Sequence Detection (ML) method. It came to be used. However, the recording density of the recording medium is further increased, and as a result, the influence of intersymbol interference and the nonlinear distortion due to the medium become conspicuous, and sufficient performance cannot be expected with the conventional PRML system.

  In contrast, methods such as DFE (Decision Feedback Equalization) and FDTS / DF (Fixed Delay Tree Search / Decision Feedback) enable whitening of intersymbol interference and colored noise of noise contained in the playback signal. Well known as. There is also a NPML (Noise Predictive Maximum Likelihood) method that incorporates the DFE concept into a PRML method with good detection accuracy. This method has already been put to practical use in hard disk drive devices. Furthermore, there is a PDNP (Pattern-Dependent Noise Prediction) method that develops this NPML method and reduces the influence of distortion caused by a medium or the like depending on the pattern of a signal to be recorded (see Non-Patent Document 1, for example).

  FIG. 1 is a block diagram illustrating a configuration example of a conventional NPML signal processing unit.

  In FIG. 1, a signal processing unit 1 is a signal processing unit that detects a reproduction signal reproduced from a recording medium. The reproduction signal having the waveform equalized to the equalization reference value input from the input terminal 11 is supplied to the adder 12, added with the output of the noise predictor 13, and supplied to the Viterbi detector 14. The Viterbi detector 14 outputs a binary signal detected from the supplied reproduction signal to the outside of the signal processing unit 1 from the output terminal 15 and supplies it to the partial response processing unit (PR) 16, It is converted into a partial response as shown in 1).

  Here, M is the code interference length. The partial response processing unit 16 converts the input signal into a partial response response represented by the equation (1), and supplies it to the adder 17. The adder 17 receives an input signal from the input terminal 11. The adder 17 subtracts the value of the input signal from the response of the partial response to obtain an error signal and supplies it to the noise predictor 13. The noise predictor 13 predicts the colored noise component included in the input signal from the error signal, and supplies the prediction result to the adder 12.

Here, the output of the adder 12, i.e., a reproduced signal whose waveform is equalized and y _n, when the converted signal to the response signal a partial response of 2 values determined by the Viterbi detector 14 and b ^ _n The input z _n of the Viterbi detector 14 can be expressed as shown in Equation (2).

In Equation (2), N1 is the length of the noise predictor 13 composed of a FIR (Finite Impulse Response) filter, and p _i (1 ≦ i ≦ N1) is a coefficient of the filter. From the equation (2), the signal processing unit 1 in FIG. 1 returns the determination result to the input as compared with the normal Viterbi detection device, and therefore the input to the Viterbi detection device differs for each branch. Therefore, considering the transition from the state j (s _j ) to the state k (s _k ), the input z _n to the Viterbi detector 14 is _expressed by the following equation (3).

Further, by assigning b ^ _n to the Viterbi state, the expression (3) can be expressed as the following expression (4).

In Expression (4), bn is an equalization reference value, and the number of states in the Viterbi detector 14 is 2 ^{M + K} (0 ≦ K ≦ N1).

In the case of PDNP, noise predictors are individually prepared for all branches in the Viterbi detector, and optimum noise predictor coefficients are prepared for each branch. Accordingly, the input z _n of the Viterbi detector in this case is expressed as shown in Expression (5) as opposed to Expression (4) in the case of the NPML method.

  In this manner, the signal processing unit 1 whitens the intersymbol interference of the noise included in the reproduction signal and the colored noise.

Jaekyun Moon, and Jongseung Park, “Pattern-Dependent Noise Prediction in Signal-Dependent Noise”, IEEE J. Select. Areas Commun., Vol. 19, No. 4, pp. 730-743, April 2001

  However, the above method has a problem that the nonlinear distortion caused by the response of the recording / reproducing mechanism such as the reproducing head or the medium cannot be completely removed.

  The present invention has been made in view of such a situation, and makes it possible to compensate for non-linear distortion of an equalized waveform.

The signal processing apparatus according to the present invention includes noise whitening means for predicting and whitening intersymbol interference and colored noise contained in an input signal, distortion removing means for removing nonlinear distortion contained in the input signal, and noise whitening. Detecting means for detecting binary values from the input signal from which the intersymbol interference and the colored noise are whitened by the means and the nonlinear distortion being removed by the distortion removing means, and outputting a detection result. From the detection result of the means, the non-linear distortion included in the input signal is obtained using the correlation of the input signal, and the obtained non-linear distortion is removed from the input signal .

  The distortion removing means can obtain a nonlinear distortion included in the input signal using an FIR filter and remove the obtained nonlinear distortion from the input signal.

  The distortion removing means may obtain nonlinear distortion included in the input signal for each branch in the detecting means, and remove the obtained nonlinear distortion from the input signal.

  In the detection process of the detection means, further comprising a conversion means for converting either one of the data corresponding to the state transition for each branch and the data stored in a predetermined path memory, or both into a partial response response, The distortion removing unit can obtain the nonlinear distortion included in the input signal based on the response of the partial response obtained by the conversion by the converting unit.

  The distortion removing means may obtain nonlinear distortion included in the input signal using coefficients whose values are set independently for each branch.

  The distortion removing means can obtain non-linear distortion included in an input signal whitened by predicting intersymbol interference and colored noise by the noise whitening means.

  The noise whitening means may include prediction means for predicting intersymbol interference and chromatic noise using a coefficient that is set in common for each branch.

  The noise whitening means may comprise prediction means for predicting intersymbol interference and colored noise using coefficients whose values are set independently for each branch.

In the signal processing apparatus of the present invention, the intersymbol interference and the colored noise included in the input signal are predicted and whitened, and binary is detected from the input signal from which the nonlinear distortion included in the input signal is removed, The detection result is output. From the detection result, the nonlinear distortion included in the input signal is obtained using the correlation of the input signal, and the obtained nonlinear distortion is removed from the input signal.

  According to the present invention, a signal can be processed. In particular, the nonlinear distortion of the equalized waveform can be compensated, and the maximum likelihood detection process can be performed more accurately.

  Embodiments of the present invention will be described below with reference to the drawings.

  FIG. 2 is a block diagram illustrating a configuration example of a signal processing unit to which the present invention is applied. In FIG. 2, a signal processing unit 31 is a signal processing device that detects a reproduction signal reproduced from a recording medium, for example, and is input via an input terminal 41 to which a waveform-equalized reproduction signal is input and the input terminal 41. An adder 42 that adds the reproduced signal and the output of the noise predictor 43, a noise predictor 43 that predicts a noise component of the reproduced signal, an adder 44 that adds the output of the filter 48 to the output of the adder 42, A Viterbi detector 45 that performs likelihood detection processing, an output terminal 46 that outputs the output of the Viterbi detector 45 to the outside of the signal processing unit 31, and a partial response processing unit (PR that converts the output of the Viterbi detector 45 into a partial response response 47, a filter unit 48 for obtaining nonlinear distortion from the response of the partial response output from the partial response processing unit (PR) 47, a partial response From the response, and the addition unit 49 for obtaining the error signal.

The adder 42 adds the reproduced signal having the waveform equalized to the equalization reference value input via the input terminal 41 and the output of the noise predictor 43, and supplies the addition result to the adder 44. The adder 42 includes adders 42-1 to 42-2 ^{M + K + 1} corresponding to the state transition of each branch in the Viterbi detector 45, and the input reproduction signal is input to each adder. And the signal supplied from each noise predictor corresponding to the state transition for each branch in the Viterbi detector 45 of the noise predictor 43, and the adder of the adder 44 is added to the corresponding adder. Supply.

The noise prediction unit 43 receives the error signal obtained from the addition unit 49 (adder 49-1 to adder 49-2 ^{M + K + 1} ) and reproduces the waveform equalized to the equalization reference value. The colored noise component is predicted, and the prediction result is supplied to the adding unit 42. The noise prediction unit 43 includes noise predictors 43-1 to 43-2 ^{M + K + 1} corresponding to the state transition of each branch in the Viterbi detector 45. The prediction result is supplied to an adder (adder 42-1 to adder 42-2M ^{+ K + 1} ) corresponding to the noise predictor of the adding unit 42.

The addition unit 44 adds the addition result supplied from the addition unit 42 and the output of the filter unit 48, and supplies the addition result to the Viterbi detector 45. The adding unit 44 includes adders 44-1 to 44-2 ^{M + K + 1} corresponding to the state transition of each branch in the Viterbi detector 45, and each adder corresponds to the adder 44-1. The addition result supplied from the adder (adder 42-1 to adder 42-2 ^{M + K + 1} ) of the adder 42 and the state transition of each branch of the Viterbi detector 45 of the filter unit 48 are changed. The signal supplied from each corresponding FIR (Finite Impulse Response) filter is added, and the addition result is supplied to the Viterbi detector 45.

The Viterbi detector 45 corresponds to the state transition of each branch in the Viterbi detector 45 supplied from each adder (adder 44-1 to adder 44-2M ^{+ K + 1} ) of the adder 44. Based on the addition result, maximum likelihood detection processing (maximum likelihood detection processing by the Viterbi detection method) is performed, and the obtained determination value is output to the outside of the signal processing unit 31 via the output terminal 46. The Viterbi detector 45 supplies the determination result to the partial response processing unit (PR) 47 corresponding to the state transition for each branch in the Viterbi detector 45.

  The partial response unit (PR) 47 converts the determination result stored in the path memory of the Viterbi detector 45 into a partial response response corresponding to the state transition for each branch in the Viterbi detector 45, and filters them. To the unit 48 and the addition unit 49.

The filter unit 48 obtains non-linear distortion included in the reproduction signal waveform-equalized to the equalization reference value with respect to the response of the partial response supplied from the partial response unit (PR) 47, and supplies it to the adding unit 44. Supply. The filter unit 48 includes FIR filters 48-1 to 48-2 ^{M + K + 1} corresponding to the state transition of each branch in the Viterbi detector 45. In each FIR filter, the Viterbi detector 45 calculated nonlinear distortion than the response of the partial response corresponding to the state transitions for each branch, the processing result, the adding unit 44, the FIR filter corresponding adder (adder 44-1 to the adder 44-2 ^{M + K + 1} ).

The adder 49 subtracts the value of the input reproduction signal from the response of the partial response supplied from the partial response unit (PR) 47 and supplies the subtraction result to the noise prediction unit 43 in order to obtain an error signal. To do. The adder unit 49 includes adders 49-1 to 49-2 ^{M + K + 1} corresponding to the state transition of each branch in the Viterbi detector 45. In each adder, the Viterbi detector 45 The value of the input reproduction signal is subtracted from the response of the partial response corresponding to the state transition for each branch, and the subtraction result is a noise predictor (noise predictor) corresponding to the adder of the noise predictor 43. 43-1 to noise predictor 43-2 ^{M + K + 1} ).

In FIG. 2, an adder 42-1 to an adder 42-2 ^{M + K + 1} , a noise predictor 43-1 to a noise predictor 43-2 ^{M + K + 1} , an adder 44-1 to an adder. 44-2 ^{M + K + 1} , FIR filter 48-1 to FIR filter 48-2 ^{M + K + 1} , and adder 49-1 to adder 49-2 ^{M + K + 1} Only the others are shown and the others are omitted. Specifically, the two branches in a certain kth state are represented by the 2k−1, 2k (1 ≦ i ≦ 2 ^{M + K} ) th. In FIG. 2, the adder 42, the noise predictor 43, the adder 44, and the filter 48 for the first and 2 ^{M + K} states are clearly shown.

  FIG. 3 is a block diagram illustrating a configuration example of a noise predictor or FIR filter included in the noise prediction unit 43 and the filter unit 48. In FIG. 3, the FIR filter includes an input terminal 61, delay elements 62-1 to 62-N which are N delay elements for delaying an input signal, and N multiplications for multiplying the delayed signal by a filter coefficient. Multipliers 63-1 to 63-N, adders 64 for adding the multiplication results, and an output terminal 65 for outputting the addition results.

  The delay elements 62-1 to 62-N are each configured by a flip-flop circuit or the like, and each delay element is connected to the input terminal 61 in series. Each delay element holds a signal input for a predetermined period, and then supplies the signal to a multiplier to which the delay element corresponds and a delay element at the next stage. Note that the delay element 62-N, which is the Nth delay element, has no subsequent delay element, and therefore supplies its output only to the multiplier 63-N to which the delay element 62-N corresponds. Here, N represents the number of taps of the FIR filter.

The multipliers 63-1 to 63 -N multiply the supplied signals by preset filter coefficients p _{N to} p ₁ , and supply the multiplication results to the adder 64. The adder 64 adds all the supplied multiplication results, and outputs the addition result to the outside via the output terminal 65 as a filter output.

  Next, the operation of the signal processing unit 31 in FIG. 2 will be described.

A reproduction signal that has been waveform-equalized to an equalization reference value and that has been input from the input terminal 41 is supplied to an adder 42 and an adder 49. The adder 42 (adder 42-1 to adder 42-2 ^{M + K + 1} ) and the noise prediction unit 43 (noise predictor 43-1 to noise predictor 43-2 ^{M + K +) 1} ) and the addition result are supplied to the adding unit 44. The adder 44 (adder 44-1 to adder 44-2 ^{M + K + 1} ) includes the addition result and the filter unit 48 (FIR filter 48-1 to FIR filter 48-2 ^{M + K + 1} ). The output is added, and the addition result is supplied to the Viterbi detector 45. The Viterbi detector 45 performs maximum likelihood detection processing on the addition result, and outputs a determination value that is a determination result to the outside of the signal processing unit 31 via the output terminal 46.

In addition, the determination result stored in the path memory of the Viterbi detector 45 corresponding to the state transition of each branch in the Viterbi detector 45 is supplied to the partial response processing unit (PR) 47. The partial response processing unit (PR) 47 converts each supplied determination result into a partial response response corresponding to the state transition for each branch in the Viterbi detector 45, and outputs the response to the filter unit 48 and the addition unit 49. Supply. The filter unit 48 (FIR filter 48-1 to FIR filter 48-2 ^{M + K + 1} ) performs nonlinear distortion included in the reproduced signal whose waveform is equalized to the equalization reference value with respect to the supplied partial response response. And the processing result is supplied to the adding unit 44 to compensate for the nonlinear distortion included in the reproduction signal waveform equalized to the equalization reference value.

An adder 49 (adders 49-1 to 49-2 ^{M + K + 1} ) is provided for each branch in the Viterbi detector 45 supplied from the partial response processing unit (PR) 47 in order to obtain an error signal. The reproduction signal is subtracted from the response of the partial response corresponding to the state transition, and the subtraction result is supplied to the noise prediction unit 43. The noise predicting unit 43 (noise predictor 43-1 to noise predictor 43-2 ^{M + K + 1} ) is a code included in the reproduction signal for each branch in the Viterbi detector 45 based on the supplied subtraction result. Interference and chromatic noise are predicted, and the prediction result is supplied to the adder 42 to whiten the chromatic noise included in the reproduction signal.

  As described above, the signal processing unit 31 not only whitens the intersymbol interference and the chromatic noise included in the reproduction signal in the noise prediction unit 43, the addition unit 42, and the addition unit 49, but also adds the filter unit 48 and the addition unit. The unit 44 compensates for nonlinear distortion in the equalized waveform. This nonlinear distortion can be expressed by a linear combination of past determination values from the determination value of the Viterbi detector 45 that does not include distortion, using the correlation of signals. Therefore, this distortion can be expressed as in equation (6).

In Expression (6), q _i (1 ≦ i ≦ N2) is a coefficient, and N2 represents the length of the linear combination. As a result, the response of b _n can be made closer to the response of the equalized reproduction signal y _n , and the influence of distortion can be reduced.

That is, in the signal processing unit 31 shown in FIG. 2, when the values of the coefficients of the noise predictor 43-1 to noise predictor 43-2 ^{M + K + 1} of the noise prediction unit 43 are unified, that is, In the case of NPML, the signal processing unit 31 can be expressed by Expression (7) with respect to Expression (4).

Similarly, in the signal processing unit 31 shown in FIG. 2, the values of the coefficients of the noise predictors 43-1 to 43-2 ^{M + K + 1} of the noise predictor 43 are set independently of each other. When it is performed, that is, in the case of PDNP, the signal processing unit 31 can be expressed by Expression (8) with respect to Expression (5).

  As described above, the signal processing unit 31 not only whitens the intersymbol interference and the chromatic noise included in the reproduction signal in the noise prediction unit 43, the addition unit 42, and the addition unit 49, but also includes the filter unit 48 and Since the adder 44 compensates for nonlinear distortion in the equalized waveform, the maximum likelihood detection process can be performed more accurately. That is, the signal processing unit 31 can compensate for the nonlinear distortion of the equalized waveform.

  FIG. 4 is a diagram showing a configuration example of an optical disc reproducing apparatus to which the present invention is applied.

  In FIG. 4, an optical disc playback apparatus 100 is an apparatus that plays back information recorded on an optical disc such as a CD or a DVD. Information recorded on the optical disk 111 of the optical disk reproducing apparatus 100 is reproduced as a reproduction signal by the optical head 112 and amplified to an appropriate amplitude by the RF amplifier 113.

  The amplified reproduction signal is input to a gain control amplifier (GCA) 114 and amplified to a desired amplitude. The amplified reproduction signal is supplied to a low-pass filter (LPF) 115 for preventing anti-aliasing, filtered, and then supplied to an AD converter (Analog to Digital Converter) 117 for quantum. It becomes.

  The output of the low pass filter 115 is also supplied to the gain controller 116. The gain controller 116 determines the gain (amplification degree) of the gain control amplifier 114 based on the supplied reproduction signal, and operates the gain control amplifier 114 as an AGC circuit (Auto Gain Controller). Further, the output of the low-pass filter 115 is also supplied to a PLL (Phase Locked Loop) 118. A reference clock is also supplied to the PLL 118. The PLL 118 synchronizes the supplied reference clock with the output of the low-pass filter 115 to generate a sampling clock for the AD converter 117 and supplies it to the AD converter 117.

  The AD converter 117 quantizes the reproduction signal, which is the output of the low-pass filter, using the sampling clock. The quantized reproduction signal is supplied to the FIR filter 119. The FIR filter 119 performs adaptive equalization processing or equalization processing on the supplied reproduction signal, and supplies the equalized reproduction signal to the signal processing device 31 described with reference to FIG. The signal processing device 31 performs detection processing on the supplied reproduction signal, restores the original signal before being recorded on the optical disc 111, and outputs it.

  As described above, by detecting the reproduction signal reproduced from the optical disk 111 using the signal processing unit 31 described with reference to FIG. 2, the signal processing unit 31 can detect intersymbol interference or colority included in the reproduction signal. Since not only noise is removed but also non-linear distortion in the equalized waveform is removed, the optical disc reproducing apparatus 100 can perform the maximum likelihood detection process more accurately. That is, the optical disc reproducing apparatus 100 can compensate for the nonlinear distortion of the equalized waveform.

FIG. 5 is a graph showing an example of the tolerance of the optical disc playback apparatus 100 with respect to the inclination in the radial direction of the optical disc of the Blue-ray Disc (registered trademark) format (user capacity: 27 gigabytes). In FIG. 5, the horizontal axis indicates the radial skew that is the inclination of the disk in the radial direction, and the vertical axis indicates the bit error rate at that time. Further, a curve 131 in FIG. 5 shows an example of resistance to the radial inclination of the disc when a reproduction signal is detected by the PRML method (PR (1, 2, 2, 1)). Shows an example of tolerance to the radial inclination of the disc when a reproduction signal is detected by the NPML method, and a curve 133 shows a case where a distortion correction filter is applied to the NPML method, that is, a noise predictor. when detecting the reproduction signal using a 43-1 to noise estimator 43-2 M + K ⁺ signal processing unit 31 of FIG. 2 where the ^first coefficients in correspondence with the NPML scheme in common, in the radial direction of the disk An example of resistance to tilt is shown.

  In FIG. 5, when the reproduction signal is detected by using the signal processing unit 31 of FIG. 2 corresponding to the NPML method, the horizontal spread of the curve 133 is the largest, and the reproduction signal is detected by the PRML method or the NPML method. Compared to the case, it is shown that the tolerance to the inclination in the radial direction of the disk is the greatest.

Similarly, FIG. 6 is a graph showing another example of the tolerance of the optical disc playback apparatus 100 with respect to the inclination in the radial direction of an optical disc in the Blue-ray Disc (registered trademark) format (user capacity: 27 gigabytes). In FIG. 6, the horizontal axis indicates the radial skew that is the inclination of the disk in the radial direction, and the vertical axis indicates the bit error rate at that time. A curve 141 in FIG. 6 shows an example of resistance to the radial inclination of the disc when a reproduction signal is detected by the PRML method (PR (1,2,2,1)). Shows an example of tolerance to the radial inclination of the disc when a reproduction signal is detected by the PDNP method, and a curve 143 shows a case where a distortion correction filter is applied to the PDNP method, that is, a noise predictor. 43-1 to noise predictor 43-2 When the reproduction signal is detected by using the signal processing unit 31 of FIG. 2 corresponding to the PDNP method by setting the coefficients of ^{M + K + 1} independently of each other. An example of resistance to the radial tilt of the disk is shown.

  In FIG. 6, when the reproduction signal is detected using the signal processing unit 31 of FIG. 2 corresponding to the PDNP method, the horizontal spread of the curve 143 is the largest, and the reproduction signal is detected by the PRML method or the PDNP method. Compared to the case, it is shown that the tolerance to the inclination in the radial direction of the disk is the greatest.

  As described above, the optical disc reproducing apparatus 100 (signal processing unit 31) compensates for the non-linear distortion of the equalized waveform using the FIR filter that is a distortion correction filter, as shown in FIGS. The tolerance to the radial tilt of the disk can be increased.

Note that the filter coefficient of the filter unit 48 (FIR filter 48-1 to FIR filter 48-2 ^{M + K + 1} ) of the signal processing unit 31 in FIG. 2 and the noise prediction unit 43 (noise predictor 43-1 to noise prediction). The coefficient of the unit 43-2 ^{M + K + 1} ) may be a predetermined value, but can be adaptively equalized using an LMS (Least Mean Square) algorithm.

Next, how to obtain the error signal will be described. The LMS algorithm is the most representative of many adaptive equalization algorithms, and is widely known, so the description thereof is omitted here. First, the case of the signal processing device 31 corresponding to NPML will be described. The error signal for the FIR filter of the noise predictor is the difference between the signal obtained by converting the determination result of the Viterbi detector into a response of the partial response and the input signal of the signal processing device 31, and the output of the noise predictor using the calculation result as an input. And the difference. Then, the FIR filter coefficient p _i (1 ≦ i ≦ N1) may be determined using a normal LMS algorithm.

The coefficients of the FIR filter that compensates for the nonlinear distortion of the equalized waveform are described in Jaekyun Moon, and Jongseung Park, “Pattern-Dependent Noise Prediction in Signal-Dependent Noise”, IEEE J. Select. Areas Commun., Vol. 19, The LMS algorithm as described in No.4, pp.730-743, April 2001 is used. The binary judgment result of the Viterbi detector is stored in a memory or the like for the length of the Viterbi detector state, and the state transition of the current time is detected from the judgment result of the current time and the past judgment result every time, The coefficient q _i of the FIR filter of the branch corresponding to the transition is updated. The error signal at this time is a signal obtained by converting the determination result into a partial response response, and is input to the FIR filter having the same coefficient as the coefficient of the FIR filter of the branch corresponding to the detected state transition, and the output is described above. It is assumed that it is subtracted from the error signal for determining the output of the noise predictor.

  FIG. 7 is a block diagram showing a configuration example of a signal processing unit to which the present invention is applied in such a case. The same components as those of the signal processing unit 31 shown in FIG. 2 are denoted by the same reference numerals as those in FIG.

In FIG. 7, the output of the addition unit 49 of the signal processing unit 151 is supplied to a noise prediction unit 163 including noise predictors 163-1 to 163-2 ^{M + K + 1} whose coefficients can be updated. That is, as in the case of FIG. 2, the outputs of the adders 49-1 to 49-2 ^{M + K + 1} are the noise predictors 163-1 to 163-2 ^{M + K + 1, respectively.} To be supplied. The noise predictors 163-1 to 163-2 ^{M + K + 1} are the noise predictors 43-1 to 43-2 ^{M + K + of} FIG. 2 except that the coefficients can be updated. ¹ , and their outputs are supplied to adders 42-1 through 42-2 ^{M + K + 1} of the adder 42, respectively.

The output of the partial response processing unit (PR) 47 (partial response response) is supplied to a filter unit 168 including FIR filters 168-1 to F168 filters 168-2 ^{M + K + 1} whose coefficients can be updated. . That is, as in the case of FIG. 2, the response of the partial response corresponding to the state transition for each branch in the Viterbi detector 45 is supplied to the FIR filters 168-1 to 168-2 ^{M + K + 1} , respectively. Is done. FIR filters 168-1 through 168-2 ^{M + K + 1} are the same as FIR filters 48-1 through 48-2 ^{M + K + 1} in FIG. 2 except that the coefficients can be updated. These outputs are supplied to adders 44-1 to 44-2M ^{+ K + 1} of the adder 44, respectively.

The signal processing unit 151 is a circuit that determines coefficients of the noise predictors 163-1 to 163-2 ^{M + K + 1} and the FIR filters 168-1 to 168-2 ^{M + K + 1.} A delay processing unit 171 that delays the input reproduction signal, a partial response processing unit 172 that converts the output (determination value) of the Viterbi detector 45 into a partial response response, and a reproduction signal delayed from the partial response response An adder 173 that subtracts, an adder 174 that subtracts the output of the noise predictor 175 from the subtraction result in the adder 173, an adder 176 that subtracts the output of the FIR filter 177 from the subtraction result of the noise predictor 175 and adder 174. The filter coefficient is determined by the subtraction result of the FIR filter 177 and the adder 176 that performs filtering on the response of the partial response When a known signal pattern is used for training in determining a filter coefficient, a switch circuit 178 for selecting an FIR filter to be used, an input terminal 181 for inputting this signal, a judgment value supplied from the Viterbi detector 45 or a known signal The circuit includes a switch circuit 182 that selects one of the patterns, and a pattern detector 183 that detects a state transition pattern based on a signal selected by the switch circuit 182.

  In the signal processing unit 151, the input reproduction signal is supplied to the addition unit 42 and the addition unit 49 and is also supplied to the delay processing unit 171, and is delayed so as to have the same timing as the output of the Viterbi detector 45. The Further, the binary determination value output from the Viterbi detector 45 is output from the output terminal 46 and also supplied to the partial response processing unit (PR) 172 and the switch circuit 182. The partial response processing unit (PR) 172 converts the supplied determination value into a partial response response, and supplies it to the adder 173 and the FIR filter 177.

The adder 173 subtracts the reproduction signal delayed by the delay processing unit 171 from the response of the partial response, and supplies the subtraction result to the adder 174 and the noise predictor 175. The noise predictor 175 supplies the calculation result to the adder 174. The adder 174 subtracts the output of the noise predictor 175 from the subtraction result in the adder 173, and uses the subtraction result as an error signal of each noise predictor constituting the noise predictor 163. The noise predictor 163-2 is supplied to ^{M + K + 1} . The error signal is also supplied to the adder 176.

  The FIR filter 177 calculates nonlinear distortion of the equalized waveform from the response of the partial response supplied from the partial response processing unit (PR) 172 and supplies the calculation result to the adder 176. The adder 176 subtracts the output of the FIR filter 177 from the subtraction result of the adder 174, and supplies the subtraction result to the switch circuit 178 as an error signal of each FIR filter constituting the filter unit 168.

  By the way, the determination value is supplied from the Viterbi detector 45 to the switch circuit 182, and further, a signal pattern can be supplied from the outside via the input terminal 181. The switch circuit 182 switches the switch so as to supply a known signal pattern input via the input terminal 181 to the pattern detector 183 when using a known signal pattern in training for determining the filter coefficient. . When training with the determination result of the Viterbi detector 45, the switch is switched so that the determination value supplied from the Viterbi detector 45 is supplied to the pattern detector 183.

The pattern detector 183 creates a signal for selecting a branch to which the LMS is applied based on the known signal pattern or determination value supplied in this way, and supplies it to the switch circuit 178 and the FIR filter 177. The FIR filter 177 selects the FIR filter coefficient based on the signal supplied from the pattern detector 183 in this way. Further, the switch circuit 178 selects the FIR filter of the branch to which the LMS is applied from the FIR filter 168-1 to the FIR filter 168-2 ^{M + K + 1} based on the signal supplied from the pattern detector 183 in this way. Then, the FIR filter error signal is controlled to be supplied to the selected FIR filter by switching the switch.

Next, the case of the signal processing device 31 corresponding to PDNP will be described. The error signal for the FIR filter of the noise predictor is the difference between the signal obtained by converting the determination result of the Viterbi detector into the response of the partial response and the input signal of the signal processing device 31 and the calculation thereof, as in the case of corresponding to the NPML. This is the difference from the output of the noise predictor with the result as input. At this time, the FIR filter coefficient p _i (1 ≦ i ≦ N1) of the noise predictor is determined at each time and the current time as in the case of determining the coefficient of the FIR filter that compensates for nonlinear distortion of the equalized waveform. The state transition at the current time is detected from the past determination results, and the coefficient of the noise predictor of the branch corresponding to the transition is used. And the coefficient of the noise predictor which the branch corresponding to the detected state transition has is updated.

  The coefficient of the FIR filter that compensates for the non-linear distortion of the equalized waveform is calculated in the same manner as in the case of corresponding to NPML.

  FIG. 8 is a block diagram showing a configuration example of a signal processing unit to which the present invention is applied in such a case. The same components as those of the signal processing unit 151 shown in FIG. 7 are denoted by the same reference numerals as those in FIG.

  8, the configuration of the signal processing unit 201 is basically the same as that of the signal processing unit 151 in FIG. 7, except that the output of the adder 174 is supplied to the switch circuit 221 and the output to the adder 174. The signal processing apparatus 151 is different from the signal processing device 151 in that the noise predictor coefficient to be used is the noise predictor coefficient of the branch corresponding to the detected pattern.

  The pattern detector 183 generates a signal for selecting a branch to which the LMS is applied based on a known signal pattern or determination value supplied in the same manner as in FIG. 7, and generates a signal that selects the switch circuit 178 and the FIR filter 177. To the noise predictor 215 and the switch circuit 221.

  The noise predictor 215 sets a coefficient based on the signal supplied from the pattern detector 183, predicts a colored noise component from the subtraction result of the adder 173 using the coefficient, and adds the prediction result to the adder 174. To supply.

The error signal of the noise predictor that is the subtraction result output from the adder 174 is supplied to the switch circuit 221. Similarly to the error signal of the FIR filter, the switch circuit 221 selects the noise predictors 163-1 to 163 as the branch noise predictors to which the LMS is applied based on the signal supplied from the pattern detector 183. By selecting from −2 ^{M + K + 1} and switching the switch, the error signal of the noise predictor is controlled to be supplied to the selected noise predictor.

  The method for updating the noise predictor and FIR filter coefficients may be other than those described above, and any algorithm may be used.

  In the above description, the optical disk reproducing apparatus has been described. However, the present invention is not limited to this, and the signal processing unit 31 in FIG. 2, the signal processing unit 151 in FIG. 7, and the signal processing unit 201 in FIG. The present invention may be applied to any apparatus such as a recording / reproducing apparatus or a communication apparatus.

  Further, in the above description, the method for compensating nonlinear distortion using the determination value output by the Viterbi detector 45 has been described. However, the present invention is not limited to this, and the Viterbi detector 45 is stored in a built-in predetermined path memory. Of course, the data may be supplied to the partial response processing unit 47. That is, the signal processing unit 31 may compensate for the nonlinear distortion by using one or both of data corresponding to the state transition for each branch and data stored in a predetermined path memory.

  Further, in this specification, the system represents the entire apparatus constituted by a plurality of apparatuses.

It is a block diagram which shows the structural example of the conventional signal processing part. It is a block diagram which shows the structural example of the signal processing part to which this invention is applied. FIG. 3 is a block diagram illustrating a configuration example of an FIR filter in FIG. 2. It is a figure which shows the structural example of the optical disk reproducing device to which this invention is applied. It is a figure which shows the example of the tolerance with respect to the inclination of the radial direction of a disk. It is a figure which shows the other example of tolerance with respect to the inclination of the radial direction of a disk. It is a figure which shows the other structural example of the signal processing part to which this invention is applied. It is a figure which shows the further another structural example of the signal processing part to which this invention is applied.

Explanation of symbols

  1 signal processing unit, 31 signal processing unit, 42 addition unit, 43 noise prediction unit, 44 addition unit, 45 Viterbi detector, 47 partial response processing unit, 48 filter unit, 49 addition unit, 62-1 to 62-N delay Element, 63-1 to 63-N multiplier, 64 adder, 100 optical disk playback device, 111 optical disk, 112 optical head, 113 RF amplifier, 114 GCA, 115 LPF, 116 gain controller, 117 AD converter, 118 PLL , 119 FIR filter, 163 noise prediction unit, 168 filter unit, 171 delay processing unit, 172 partial response processing unit, 173 adder, 174 adder, 175 noise predictor, 176 adder, 177 FIR filter, 178 switch circuit, 182 Switch circuit, 183 pattern detector

Claims (8)

In a signal processing device that detects a binary value from a waveform-equalized input signal and outputs a detection result,
Noise whitening means for predicting and whitening intersymbol interference and colored noise included in the input signal;
Distortion removing means for removing nonlinear distortion included in the input signal;
Detecting means for detecting binary values from the input signal from which the intersymbol interference and colored noise are whitened by the noise whitening means, and from which the nonlinear distortion is removed by the distortion removing means, and outputting a detection result; Prepared ,
The distortion removing unit obtains a nonlinear distortion included in the input signal from the detection result of the detecting unit using the correlation of the input signal, and removes the obtained nonlinear distortion from the input signal.
Signal processing apparatus characterized by. The signal processing apparatus according to claim 1, wherein the distortion removing unit obtains nonlinear distortion included in the input signal using an FIR filter, and removes the obtained nonlinear distortion from the input signal. 2. The signal processing according to claim 1, wherein the distortion removing unit obtains a nonlinear distortion included in the input signal for each branch in the detection unit, and removes the obtained nonlinear distortion from the input signal. apparatus. In the detection processing of the detection means, further comprising conversion means for converting either one or both of data corresponding to the state transition for each branch and data stored in a predetermined path memory into a partial response response,
The signal processing apparatus according to claim 3, wherein the distortion removing unit obtains a nonlinear distortion included in the input signal based on a response of the partial response obtained by being converted by the converting unit. The signal processing apparatus according to claim 3, wherein the distortion removing unit obtains a nonlinear distortion included in the input signal by using coefficients whose values are set independently for each branch. The distortion removing unit obtains a non-linear distortion included in the input signal that is whitened by predicting the intersymbol interference and the colored noise by the noise whitening unit. Signal processing equipment. The noise whitening means includes prediction means for predicting the intersymbol interference and the chromatic noise using a coefficient that is set in common with each other for each branch. 4. The signal processing device according to 3. The noise whitening means includes prediction means for predicting the code-sense interference and the colored noise using coefficients whose values are set independently for each branch. A signal processing device according to 1.

Images