JP2008300023A - Information reproducing device and method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an information reproducing device for improving detection performance for a nonlinear reproduction waveform. <P>SOLUTION: An equalizer 105 equalizes a reproduction signal to have desired PR characteristics. A Viterbi detector 106 detects binary data from the equalization reproduction signal equalized by the equalizer 105. A level average value calculator 107 calculates an average value of an input signal of the equalizer 105, for each pattern of detection data on the basis of the input signal of the equalizer 105 and the detection data output by the Viterbi detector 106. An equalization target calculator 109 outputs, as an equalization target value, the average value calculated by the level average value calculator 107 and corresponding to the pattern of detection data to a tap coefficient calculator 108. The tap coefficient calculator 108 calculates a tap coefficient of the equalizer 105 from the equalization target value and the input signal of the equalizer 105. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to an information reproducing apparatus and method, and more particularly to an information reproducing apparatus and method using a PRML (Partial Response Maximum Likelihood) method.

  There is a signal processing technique called a PRML system as a playback system for high-density recorded information in an optical disk device or the like. The PRML system is a system that combines partial response waveform equalization and maximum likelihood detection, and after correcting the reproduction signal by waveform equalization in order to maximize the characteristics of the maximum likelihood detector considering the reproduction channel. Perform maximum likelihood detection.

The PR method can be expressed as PR (h ₀ , h ₁ ,..., H _N−1 ) where _N is the constraint length. In the PR method, an output value of “0h ₀ h ₁ ” is output from a signal read from the head by an equalizer typified by a FIR (Finite Impulse Response) filter with respect to the recording data string a _k “010... 0”. ... h _N-1 0 ... 0 ", waveform equalization using intersymbol interference is performed. As a result, the recording data string becomes multi-valued data after equalization. The greater the constraint length N, the more multivalue.

  In the PR method, the constraint length N and hi (i = 0, 1,..., N−1) are changed in accordance with the characteristics of the reproduction signal due to the difference in recording density and the like. Initially, the PR (1, 1) method with a constraint length N = 2 was used. However, in recent optical disk devices aiming at higher density, the PR (1, 2, 2, 1) method and PR (1) are used. The use of the 2, 2, 2, 1) method has become mainstream.

  FIG. 4 shows a recording waveform and an output waveform in the PR system. For the recording waveform (a), the output waveform of the PR (1, 2, 2, 1) system is as shown in (b), and the output waveform of the PR (1, 2, 2, 2, 1) system Is as shown in (c). In addition, FIGS. 5A and 5B show the PR type eye patterns. FIG. 5A shows a PR (1, 2, 2, 1) type eye pattern, and FIG. 5B shows a PR (1, 2, 2, 2, 1) type eye pattern. In this way, the output waveform of the PR system is multivalued with respect to the binary recording waveform.

Maximum likelihood detector constraint length N, the recording data sequence _{a k-N + 1, a} k-N + 2 of length N-1, _..., with respect to the state at time k-1 based on the pattern of _{a k-1} , Information detection is performed based on the state transition by the input signal sequence y _{k at} time k. Normally, the number of consecutive “1” s and the number of consecutive “0” s are limited to a predetermined length in a recording data string recorded on the information recording medium by appropriate encoding. The encoding method is characterized by a minimum run length d which is the minimum number of consecutive _1s or 0s of the recording data sequence _ak . For example, in the PR (1, 1) method in which the simplest constraint length N is “2” and the minimum run length d = 0, the state transition is as follows: a _k = 0 to state S ₀ and a _k = 1 When assigned to state S _1, as shown in FIG. 6, the transition 2 state 3 value reference value of the combined "0,1,2" to the output of the equalizer. In general, a reference value from which a DC component is removed is used, and thereafter, “0, 1, 2” is expressed as “−1, 0, 1” according to this notation.

FIG. 7 shows a trellis diagram in which the state transition diagram of FIG. 6 is developed on the time axis. In the trellis diagram, the path length at each time is called a branch metric. The branch metric that makes a transition from the state S (k−1) = S _i (i = 0, 1) to the state S (k) = S _j (j = 0, 1) is given by the following Equation 1.
l _k ⁰⁰ = (y _k − (− 1)) ²
l _k ⁰¹ = (y _k −0) ²
l _k ¹⁰ = (y _k −0) ²
l _k ¹¹ = (y _k −1) ² (1)

In the Viterbi algorithm, among the paths input to one state at a certain time k in the trellis diagram, the total length of the paths up to time k−1 and the value of y _k input at time k are used. The length of the path when passing through each path is calculated, and an operation for selecting a smaller path is performed. The minimum value of the path length to reach each state at a certain time is called a path metric, and the PR (1, 1) system path metric is given by the following formula 2.
m _k (S ₀ ) = min {m _k−1 (S ₀ ) + l _k ⁰⁰ , m _n−1 (S ₁ ) + l _k ¹⁰ }
m _k (S ₁ ) = min {m _k−1 (S ₀ ) + l _k ⁰¹ , m _k−1 (S ₁ ) + l _k ¹¹ } (2)

m _k (S ₀ ) indicates the minimum value of the length of the path that transitions to the state S ₀ at time k. That is, the path metric to a transition to a state S ₀ at time k-1, and the sum of the branch metric transition to S ₀ from the state S ₀ at time k at time k-1, time k-1 to the state S Of the sum of the path metric until the transition to ₁ and the branch metric that transitions from state S ₁ to time S ₀ at time k−1, the smallest one is selected. At each time, only the path having the metric of Expression 2 remains as a path having the possibility of becoming the maximum likelihood path, and the path metric is updated. If the path left in this way is called a surviving path, and the surviving path is hit in the past, the probability that the paths merge into one at a certain point increases. The merged path is detected as the maximum likelihood path, and detection data is obtained. Even when the constraint length increases, detection data can be obtained by obtaining the maximum likelihood path based on the same principle only by increasing the number of branches.

By the way, if the distortion of the pit shape on the information recording medium is increased by increasing the information density, nonlinear distortion such as asymmetry occurs in the reproduced waveform. Asymmetry occurs mainly depending on the recorded mark shape. FIG. 8 shows an example of a reproduced waveform with asymmetry. When asymmetry occurs, as shown in FIG. 8, in the reproduced waveform, the interval between the multilevel levels becomes unbalanced in the amplitude direction. If the reproduction waveform values for the longest mark and space are I _11H and I _11L, and the reproduction waveform values for the shortest mark and space are I _2H and I _2L , respectively, the asymmetry is
0.5 * [( _I11H + _I11L )-( _I2H + _I2L )] / ( _{I11H- I11L} )
Defined by

  In the PRML system, nonlinearity cannot be reduced with a linear filter such as an FIR filter generally used as an equalizer. Further, the reference value of the Viterbi detector is not set so as to cope with nonlinear distortion such as asymmetry. The reproduced waveform with asymmetry is non-linear in that the positive and negative peak levels corresponding to the pattern density are not symmetrical with respect to the central reference value, that is, the center of the positive and negative peak levels differs depending on the pattern density. A phenomenon occurs. Therefore, for example, in the PR (1, 1) ML system, the input reference level of the Viterbi detector is “−1, 0, 1”, but in the case of a waveform with asymmetry, the input waveform is “−0. It may be 8, 0.1, 1.2 ". In this case, the errors are different from “−0.2, 0.1, 0.2” with respect to the correct reference value, and complete correction cannot be performed by offset correction. If an error occurs at each time, there is a high possibility that an incorrect path is selected, and there is a problem that the detection accuracy is degraded. Furthermore, since the intersymbol interference increases in the reproduced signal as the recording density increases, a higher-order PRML method may be used to suppress a decrease in detection accuracy. In this case, since the reference value of the Viterbi detector increases and the interval becomes narrow, even with the same asymmetry amount, the deviation becomes relatively large and the reproduction performance deteriorates.

  In order to suppress the above-described deterioration in reproduction performance, an adaptive PRML method has been proposed. The adaptive PRML method is described in Non-Patent Document 1, for example. FIG. 9 shows the configuration of an information reproducing apparatus using the adaptive PRML method. An adaptive PRML information reproducing apparatus includes an optical head 202, an A / D converter 203, a PLL circuit 204, an equalizer 205, a Viterbi detector 206, a level average value calculator 207, a tap coefficient calculator 208, and the like. The computerized target calculator 209 is provided. The optical disk medium 201 is rotated at a constant angular velocity rotation or a constant linear velocity rotation by a spindle motor (not shown). The optical head 202 irradiates the information mark recorded on the optical disc medium 201 with a focused spot by the objective lens. In the optical head 202, the distance between the disk surface and the objective lens, and the radial position of the disk guide groove and the focused spot are accurately controlled by a server circuit (not shown).

  Reflected light from the optical disk medium 201 changes in reflectance or polarization depending on the presence or absence of an information mark. By detecting this with a detector (not shown), a reproduction signal is obtained. The reproduction signal is converted into a digital reproduction signal by the A / D converter 203. The PLL circuit 204 inputs an A / D converted digital reproduction signal, performs phase control so as to synchronize with the input channel frequency, and outputs a resampling signal interpolated from a data string of a fixed sampling rate. The equalizer 205 equalizes the output of the PLL circuit 204 so as to have an externally set PR characteristic. The tap coefficient in the equalizer 205 is controlled by the tap coefficient calculator 208.

  The Viterbi detector 206 receives the output of the equalizer 205, performs maximum likelihood detection using the output of the level average value calculator 207 as a reference value at the time of branch metric calculation in Viterbi detection, and detects a recorded data string. The level average value calculator 207 calculates the average of the output of the equalizer 205 corresponding to the pattern of the recording data sequence based on the output of the equalizer 205 and the detected recording data sequence. The equalization target calculator 209 receives the recording data string, generates an equalization target signal by convolution of the recording data string and the PR method coefficient, and outputs it. The tap coefficient calculator 208 receives the output of the PLL circuit 204 and the output of the equalization target calculator 209 so that the square error between the output of the PLL circuit 204 and the output of the equalization target calculator 209 is minimized. The tap coefficient is calculated, and the tap coefficient in the equalizer 205 is controlled.

In the normal PRML system, the reference value for branch metric calculation in Viterbi detection uses a value similar to the target value of the equalizer calculated by convolution of the PR system coefficient and the recording data string. On the other hand, in the adaptive PRML, the average value of the output of the equalizer 205 corresponding to the recording data string calculated by the level average value calculator 207 to the reference value at the time of branch metric calculation in Viterbi detection, that is, equal The output value of the generator is used. The reference value at the time of branch metric calculation is, for example, “−1, 0, 1” in the normal PR (1, 1) method, whereas “−0.8, 0.1, 1” in the adaptive PRML method. .2 ". As a result, maximum likelihood detection is performed with a reference value adapted to a reproduction signal having nonlinearity, and the reproduction performance can be improved with respect to the normal PRML system.
ODS '02 Technical Digest, pp 269-271, “Adaptive Partial-Response Maximum-Likelihood Detection in Optical Recording Media”

  However, in the adaptive PRML system, the target value of the equalizer 205 remains the same as that of normal PRML, and the reference value at the time of branch metric calculation is calculated based on the output signal from the equalizer 205. Yes. For signals with strong non-linearity, the difference between the target value of the equalizer 205 and the reproduced waveform becomes large, so that the detection performance may deteriorate. In addition, the non-linearity of the reproduction signal of the super-resolution medium is further increased.

  An object of the present invention is to provide an information reproducing apparatus and an information reproducing method capable of improving detection performance for a non-linear reproduced waveform.

  In order to achieve the above object, an information generation apparatus of the present invention receives an equalizer for equalizing a reproduction signal from an information recording medium, and an equalized reproduction signal equalized by the equalizer, A Viterbi detector for detecting a binary signal by detection, a target signal generating means for generating an equalization target value in the equalizer based on a reproduction signal input to the equalizer, and the equalization target value A tap coefficient calculator for calculating a tap coefficient of the equalizer from the reproduced signal.

  The information reproduction method of the present invention is an information reproduction method for equalizing a reproduction signal from an information recording medium and reproducing a binary signal by viterbi detection, wherein the equalization target value at the time of equalization is equalized. The calculation is based on the previous reproduction signal.

  An information reproducing apparatus of the present invention receives an equalizer for equalizing a reproduction signal from an information recording medium, and an equalized reproduction signal equalized by the equalizer, and detects a binary signal by viterbi detection. A Viterbi detector; target signal generating means for calculating an equalization target value in the equalizer using a second-order partial response coefficient; and a tap of the equalizer from the equalization target value and the reproduction signal And a tap coefficient calculator for calculating a coefficient.

  The information reproduction method of the present invention is an information reproduction method for equalizing a reproduction signal from an information recording medium and reproducing a binary signal by viterbi detection. The equalization target value at the time of equalization is set to a secondary value. It is calculated using a partial response coefficient.

  With the information reproducing apparatus and information reproducing method of the present invention, the detection performance can be improved even for non-linear reproduced waveforms.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 shows the configuration of an information reproducing apparatus according to the first embodiment of the present invention. The information reproducing apparatus includes an optical head 102, an A / D converter 103, a PLL circuit 104, an equalizer 105, a Viterbi detector 106, a level average value calculator 107, a tap coefficient calculator 108, and an equalization target calculator. 109. In the information reproducing apparatus shown in FIG. 9, the level average value calculator 207 uses the output signal of the equalizer 205 as an input signal. On the other hand, in the information reproducing apparatus of this embodiment, the output of the PLL circuit 104, that is, the input of the equalizer 105 is used as the input signal of the level average value calculator 107.

  The optical disk medium 101 is rotated at a constant angular velocity rotation or a constant linear velocity rotation by a spindle motor (not shown). The optical head 102 irradiates the information mark recorded on the optical disk medium 101 with a focused spot by the objective lens. The optical head 102 accurately controls the distance between the disk surface and the objective lens, and the disk guide groove and the radial position of the focused spot by a server circuit (not shown). Reflected light from the optical disk medium 101 changes in reflectance or polarization depending on the presence or absence of an information mark. By detecting this with a detector (not shown), a reproduction signal is obtained.

  The A / D converter 103 converts the reproduction signal into a digital reproduction signal. The PLL circuit 104 receives the A / D converted digital reproduction signal, performs phase control so as to synchronize with the input channel frequency, and outputs a resampling signal interpolated from a data string of a fixed sampling rate. The order of the A / D converter 103 and the PLL circuit 104 may be reversed. The equalizer 105 equalizes the output of the PLL circuit 104 so as to have an externally set PR characteristic. The tap coefficient in the equalizer 105 is controlled by the tap coefficient calculator 108. It is also possible to provide a pre-equalizer that corrects the frequency characteristics of the signal in the signal path leading to the equalizer 105.

  The Viterbi detector 106 receives the output of the equalizer 105 and performs maximum likelihood detection to detect a recorded data string. The Viterbi detector 106 uses the output of the level average value calculator 107 as a reference value at the time of branch metric calculation in Viterbi detection. Based on the output of the PLL circuit 104 and the recording data string detected by the Viterbi detector 106, the level average value calculator 107 shows the relationship between the recording data string and the reference value in the Viterbi detector 106 as a recording data string. For each pattern, the output of the corresponding PLL circuit 104 is averaged and calculated. The equalization target calculator 109 receives the recording data string and the output of the level average value calculator 107, and generates an equalization target signal corresponding to the pattern of the recording data string. The tap coefficient calculator 108 minimizes the square error between the output of the PLL circuit 104 and the equalization target calculation signal based on the output of the PLL circuit 104 and the equalization target signal generated by the equalization target calculator 109. Such tap coefficients are calculated and output to the equalizer 105.

  The operation of the level average value calculator 107 will be described. The level average value calculator 107 calculates a reference value at the time of path metric calculation in the Viterbi detector 106 from the output of the PLL circuit 104 and the recording data string. Note that the recording data sequence detected by the Viterbi detector 106 does not necessarily match the recording data sequence recorded on the optical disc medium 101, but will be described here as being the same. The number of reference values calculated by the level average value calculator 107 is the number of state transitions in the Viterbi detector 106, and is determined according to the recording data length M considered in the state transition and the reference run length d of the recording code. The recording data length M is the same as the PR constraint length in PRML.

FIG. 2 shows a state transition diagram in the Viterbi detector 106. If the recording data length M is “5” and the minimum run length d is “1”, there are 16 patterns of recording data, and the state transition diagram of the Viterbi detector 106 is as shown in FIG. In the figure, the subscript “state S” indicates the pattern “a _k-4 , a _k-3 ,..., A _k ” of the recording data in each state. The level average value calculator 107 discriminates the pattern “a _k-4 , a _k-3 ,..., A _k ” of the input recording data string a _k at each time, and outputs the PLL circuit 104 for each pattern. Is averaged. When the output of the PLL circuit 104 with respect to the pattern “a _{k−M + 1} , a _{k−M + 2} ,..., A _k ” of the recording row is expressed as w _k ( _{ak−M + 1} a _{k−M + 2} ... A _k ), for example, recording data The reference value x [00000] for the pattern “000” in the column _ak is obtained by Equation 3.
x [00000] = E [w _k (00000)] (3)
Here, E [w _k (a _{k−M + 1} a _{k−M + 2} ... A _k )] is an operator that means taking a time average of w _k (a _{k−M + 1} a _{k−M + 2} ... A _k ). is there.

  The level average value calculator 107 calculates the average value of the output of the PLL circuit 104 for all the state transitions by the same calculation as in Expression 3. In the normal PRML system, for example, x [00111] and x [11100] coincide with each other, and the reference value is degenerated to 9 types with respect to 16 state transitions. Also in this embodiment, it is possible to reduce the number of reference values by collecting state transitions that take the same reference value in the normal PRML method. For example, the reference value can be reduced to nine ways by calculating the average value without dividing x [00111] and x [11100], or by taking the average of the average values obtained for each pattern. .

The output of the level average value calculator 107 is also input to the equalization target calculator 109. The equalization target calculator 109 generates a target signal z _k of the equalizer 105 for the recording data for obtaining the tap coefficient of the equalizer 105. The equalization target calculator 109 selects the pattern “a _k ” having the length “5” of the recording data string a _k input at each time from the reference values for each pattern calculated by the level average value calculator 107. ₋₄ , a _k−3 ,..., A _k ″, the reference value x [a _k−4 a _k−3 ... A _k ] is output as the equalization target signal Z _k .

The tap coefficient calculator 108 obtains a tap coefficient in the equalizer 105 using the equalization target signal z _k generated by the equalization target calculator 109. Specifically, the tap coefficients of the equalizer 105 and _{c u,} the number of taps as Nt,
C _u (k + 1) = C _u (k) −μ × (Σc _Nt−1−u × w _k−u −z _k ) × w (ku) (0 ≦ u ≦ Nt−1) (4)
Thus, the tap coefficient is obtained. Here, μ in Equation 4 is a coefficient that determines the time for the tap coefficient to converge. The tap coefficient calculator 108 taps so that an error (Σc _Nt−1−u × w _k−u −z _k ) between the output of the equalizer 105 and the equalization target value Z _k is attenuated according to Equation 4. Calculate the coefficient. The equalizer 105 generates an equalized output signal by convolution of the tap coefficient c _u calculated according to Equation 4 and the output w _k of the PLL circuit 104.

  The calculation of the equalization target value by the equalization target calculator 109 is performed sequentially. However, if the change in the signal characteristics is small, the equalization target value obtained at a certain time can be fixed, and thereafter the equalization target value can be a fixed value. In the above description, optimization by the LMS method using the principle of least squares is used for calculating the tap coefficient. However, other generally known optimization methods such as the RLS method can also be used.

Next, an example of detection performance in the present embodiment will be described. Table 1 shows the error rates of this embodiment and the normal adaptive PRML (PR (1, 2, 2, 2, 1) equalization + 16-value adaptive Viterbi detector) system using reproduced waveforms with different asymmetry. Results are shown. In the measurement, the optical disk ROM medium on which signals with different asymmetry were recorded was reproduced with sufficient reproduction power, and only the characteristics depending on the nonlinear distortion due to the waveform asymmetry in the state where the detection system noise and medium noise could be ignored. Evaluated.

  From Table 1, it can be seen that in this embodiment, the error rate can be improved regardless of the characteristics of the asymmetry as compared with the normal adaptive PRML system. This is because the target value in the equalizer 105 is determined as the PR (1, 2, 2, 2, 1) method in the normal adaptive PRML method, but in this embodiment, the signal before equalization is used. This is considered to be because an improvement effect is obtained by performing equalization in accordance with the characteristics of the signal before equalization.

表２を参照すると、何れのアシンメトリについても、基準値１６値としたときの誤り率が、基準値を９値としたときの誤り率よりも小さくなっていることがわかる。すなわち、基準値を１６値とすることで、良好な特性が得られることがわかる。このように、通常のＰＲＭＬ方式では記録データパターンにより同じ基準値となるパターンに対しても、基準値を分けて求める方が良好な性能が得られる。 Next, with respect to the reproduced waveform used above, the number of reference values (the number of target values of the equalizer 105) at the time of path metric calculation of the Viterbi detector 106 is reduced to all 16 values and 9 values. Table 2 shows the result of calculating the error rate for each case.

Referring to Table 2, it can be seen that, for any asymmetry, the error rate when the reference value is 16 is smaller than the error rate when the reference value is 9 values. That is, it can be seen that good characteristics can be obtained by setting the reference value to 16 values. As described above, in the normal PRML system, it is possible to obtain better performance by separately obtaining the reference value for the pattern having the same reference value depending on the recording data pattern.

  The effect will be described. In this embodiment, the equalization target value of the equalizer 105 is obtained based on the output value of the level average value calculator 107 that calculates the level value based on the input signal of the equalizer 105. Thus, by obtaining the equalization target value using the signal before equalization, the equalization target value itself is closer to the signal before equalization than in the normal adaptive PRML system. Thereby, good equalization can be performed by the equalizer 105, and detection performance can be improved. In this embodiment, the reference value in the Viterbi detector 106 is determined based on the output value of the level average value calculator 107. By doing so, the equalization target value and the reference value at the time of branch metric calculation can be matched, and the detection performance of the Viterbi detector 106 can be further improved.

  In this embodiment, the recording data length M considered in the state transition is set to “5” and the minimum run length d is set to “1”. However, when M is other than “5” or d is other than “1”. It is also applicable to cases. However, if M is increased, the circuit amount greatly increases with respect to the improvement in performance. Usually, a sufficient effect is obtained when M is “5” or less.

  FIG. 3 shows the configuration of the information reproducing apparatus of the second embodiment of the present invention. The information reproducing apparatus of this embodiment has a configuration in which the level average value calculator 107 in the information reproducing apparatus of the first embodiment shown in FIG. In this embodiment, the equalization target calculator 109 calculates an equalization target value using a secondary partial response. Further, the Viterbi detector 106 sets a reference value at the time of branch metric calculation using a secondary partial response. The coefficient calculator 110 calculates each coefficient of the secondary partial response from the pre-equalized reproduction signal output from the PLL circuit 104 and the detection data output from the Viterbi detector 106. In the normal PRML system, only the primary component of the partial response is used. By adding a secondary component to this, a non-linear waveform can be expressed.

ここで、記録データ列ａ_ｋは、２値データであるので、本来「０」と「１」との２値をとるが、本実施例では、記録データ列ａ_ｋにおける「０」を−１に置き換えて、式５を計算するものとする。 The coefficient calculator 110 calculates a primary component (h 1 — _i ) and a secondary component (h 2 — _{i, j} ) of the partial response. In ordinary PRML, 1-order component _{h 1_I} is usually that is predetermined, for example, in the case of _{PR (1,2,2,2,1), h 1_1 =} 1, h 1_2 = 2 , H ₁ — ₃ = 2, h ₁ — ₄ = 2, and h 1 — ₅ = 1. h 2 — _{i, j} is a secondary component newly added in the present embodiment. The equalization target calculator 109 calculates the equalization target value z _k using the recording data string a _{k according} to the following equation (5).

Here, since the recording data string _ak is binary data, the binary value of “0” and “1” is originally taken. In this embodiment, “0” in the recording data string _ak is set to −1. In this case, Equation 5 is calculated.

For the Viterbi detector 106 as well, a reference value for calculating the branch metric is calculated using Equation 5. In the present embodiment, the primary component h 1 — _i and the secondary component h 2 — _{i, j} are determined using the coefficient calculator 110 according to the characteristics of the actual reproduction waveform. A predetermined value can also be used. In that case, the coefficient calculator 110 is omitted from the configuration shown in FIG. 3, and the primary component h 1 — _i and secondary component h 2 — _{i, j} that are set in advance in the equalization target calculator 109 and the Viterbi detector 106 are determined. Should be remembered.

Here, in order to realize high detection performance, it is essential to suppress an increase in noise due to equalization. The increase in noise due to equalization becomes more significant as the difference between the reproduction signal before equalization and the equalization target value increases. For this reason, it is necessary to select an equalization target value as close as possible to the reproduction signal before equalization. Selecting an equalization target value as close as possible to the reproduction signal before equalization is equivalent to minimizing the square error between the resampled data w _k before equalization and the equalization target value z _k . That is, at the coefficient calculator _{110, ε = E [(w} k -z k) 2] and to minimize _{h 1_i, h 2_i,} may be obtained and _j.

A vector A _k and a vector H are respectively
A _k = [a _k , a _k−1 , a _k−2 . . . a _k × a _k , a _k × a _k−1 . . a _k × a _k-4 . . . a _k-4 × a _k-4 ]
H = [h _{1_1} , h _{1_2} , h _{1_3} . . . _{h2_1,1} , _{h2_1,2} , _{h2_1,5} . . . h 2 — 5, ₅ ]. In this case, the equalization target value z _k is
z _k = A _k × H ^T (H ^T is a transposed vector of H)
Can be expressed as Therefore,
ε = E [(w _k −z _k ) ² ] = E [(w _k −A _k × H ^T ) ² ]
Thus, obtaining H that minimizes ε results in a well-known ordinary least square error problem.

  Using ROM medium, normal adaptive PRML (PR (1, 2, 2, 2, 1) + adaptive Viterbi) and reference value when calculating path metric based on secondary partial response Comparison of detection performance was performed. As a ROM medium, a phase change type super-resolution layer is formed on a PC (polycarbonate) substrate having a thickness of 0.6 mm, in which pits having a depth of 60 nm are formed with a minimum pit length of 0.1 μm and a track pitch of 0.4 μm. The added medium was used. This ROM medium was reproduced by an optical head having a wavelength of 405 nm and an objective lens numerical aperture of 0.65. The data pattern formed on the ROM medium is assumed to be known, and the error rate is actually measured to compare the detection performance. Under this condition, since the shortest pit length is shorter than the optical cutoff, data cannot be detected with normal playback power (0.5 mW), but when the playback power is set to 3 mW. The data can be detected by the super-resolution effect.

表３を参照すると、本実施例では、通常のＰＲＭＬ方式に比べて、誤り率をほぼ半減できることがわかる。 In a typical adaptive PRML method, with the calculated target value, the primary component is a _{(h 1_1, h 1_2, h} 1_3, h 1_4, h 1_5) = (1,2,2,2,1), 2 -order component For h 2 — _{i, j} = 0, and the target value was calculated from Equation 5. Further, in the adaptive PRML method of this embodiment, the primary component _{(h 1_1, h 1_2, h} 1_3, h 1_4, h 1_5) = (0.8,1.2,1.4,1.2,0 .8) and the secondary components are (h 2 — _1,3 , h 2 _—2,4 , h 2 _—3,4 , h 2 _—3,5 ) = (0.07, _0.1 , _0.13 , 0.15), The target value was calculated from Equation 5. Table 3 shows a comparison of error rates between the normal adaptive PRML method and the reproduction method based on this embodiment.

Referring to Table 3, it can be seen that the error rate can be almost halved in this embodiment as compared with the normal PRML system.

In the above, the second-order component _{h 2_I,} as component _{j, h 2_1,3, h 2_2,4, h} 2_3,4, define only four _{h 2_3,5,} component later were all zero. In Viterbi detection with a constraint length of 5, a maximum of 5 × 5 = 25 types of secondary components can be defined. However, as a result of analyzing various reproduced waveforms, it was confirmed that the number of components that significantly affect the detection performance is almost 5 or less in most cases. Defining an unnecessarily large number of secondary partial response components leads to an increase in circuit scale, so it is desirable to limit the number of secondary components to about five.

  In the present embodiment, a secondary component of the partial response is considered and used for calculating the equalization target value. By using the secondary component, the difference between the equalization target value and the reproduction signal waveform can be reduced even for a reproduction signal having nonlinearity, and the Viterbi detection performance can be improved. Further, by calculating the coefficient of the partial response based on the output of the PLL circuit 104 before equalization, the square error between the sample data before equalization and the equalization target value can be reduced. A high detection performance can be obtained by suppressing an increase in noise.

  In the first embodiment, the average value calculated by the level average value calculator 107 is input to the Viterbi detector 106, and the reference value when the branch metric is calculated in the Viterbi detector and the equalization target value are obtained. Although matched, the reference value at the time of branch metric calculation and the equalization target value do not necessarily need to match. The reference value at the time of branch metric calculation by the Viterbi detector may be determined using the level average value calculator 207 shown in FIG. 9 and the output of the equalizer 105. Also, the reference value at the time of branch metric calculation in the second embodiment does not necessarily need to match the equalization target value. For example, it may be determined based on the average value calculated by the level average value calculator 107 shown in FIG. 1, or may be determined based on the average value calculated by the level average value calculator 207 shown in FIG. May be.

  Although the present invention has been described based on the preferred embodiments, the information reproducing apparatus and information reproducing method of the present invention are not limited to the above embodiments, and various modifications can be made from the configuration of the above embodiments. Further, modifications and changes are also included in the scope of the present invention.

1 is a block diagram showing a configuration of an information reproducing apparatus according to a first embodiment of the present invention. The state transition diagram which shows the state transition in a Viterbi detector. The block diagram which shows the structure of the information reproduction apparatus of 2nd Example of this invention. The wave form diagram which shows the output waveform of PR (1, 2, 2, 1) system and PR (1, 2, 2, 2, 1) system with respect to a recording waveform. The figure which shows the eye pattern of PR (1, 2, 2, 1) system and PR (1, 2, 2, 2, 1) system. The state transition diagram which shows the state transition of the Viterbi detector in PR (1, 1) system. The trellis diagram which shows the state transition of the Viterbi detector in PR (1, 1) system. The wave form diagram which shows the example of a reproduction | regeneration waveform with asymmetry. The block diagram which shows the structure of the information reproduction apparatus using the adaptive PRML system of related technology.

Explanation of symbols

101: optical disk medium 102: optical head 103: A / D converter 104: PLL circuit 105: equalizer 106: Viterbi detector 107: level average value calculator 108: tap coefficient calculator 109: equalization target calculator 110 : Coefficient calculator

Claims (22)

An equalizer for equalizing a reproduction signal from an information recording medium;
A Viterbi detector for inputting an equalized reproduction signal equalized by the equalizer and detecting a binary signal by Viterbi detection;
Target signal generation means for generating an equalization target value in the equalizer based on a reproduction signal input to the equalizer;
An information reproduction apparatus comprising: a tap coefficient calculator that calculates a tap coefficient of the equalizer from the equalization target value and the reproduction signal.   A level at which the target signal generating means calculates a level average value of the reproduction signal with respect to the binary signal pattern based on the reproduction signal input to the equalizer and the binary signal pattern output from the Viterbi detector. The average value calculator, and an equalization target calculator for obtaining the equalization target value from an output of the level average value calculator and a binary signal pattern output from the Viterbi detector, according to claim 1. Information playback device.   The information reproducing apparatus according to claim 2, wherein the level average value calculator calculates a level average value of the reproduction signal for each of the binary signal patterns. An equalizer for equalizing a reproduction signal from an information recording medium;
A Viterbi detector for inputting an equalized reproduction signal equalized by the equalizer and detecting a binary signal by Viterbi detection;
Target signal generation means for calculating an equalization target value in the equalizer using a secondary partial response coefficient;
An information reproduction apparatus comprising: a tap coefficient calculator that calculates a tap coefficient of the equalizer from the equalization target value and the reproduction signal.   The information reproduction apparatus according to claim 4, wherein the target signal generation unit calculates the equalization target value based on a binary signal pattern output from the Viterbi detector and a secondary partial response coefficient.   6. The coefficient calculator for calculating the second-order partial response coefficient based on a reproduction signal input to the equalizer and a binary signal pattern output from the Viterbi detector. Information playback device. The target signal generating means sets the data string recorded on the information recording medium as a _i , the secondary partial response coefficient as h 1 — _i , h 2 — _{i , j} , and the PR constraint length as m, and the equalization target value z _k The

The information reproducing apparatus according to any one of claims 4 to 6, which is obtained by: The information reproducing apparatus according to any one of claims 4 to 7, wherein h _{2_i, j} having a value different from 0 among the h _{2_i, j} is 5 or less.   9. The information reproducing apparatus according to claim 4, wherein the number of equalization target values matches the number of branch metrics for the Viterbi detection.   The information reproducing apparatus according to claim 1, wherein a reference value at the time of branch metric calculation in the Viterbi detector matches the equalization target value.   The information reproducing apparatus according to claim 1, wherein the number of branch metrics in the Viterbi detector is sixteen. An information reproduction method for equalizing a reproduction signal from an information recording medium and reproducing a binary signal by viterbi detection,
An information reproduction method characterized in that an equalization target value at the time of equalization is calculated based on the reproduction signal before equalization.   13. The information reproduction method according to claim 12, wherein a tap coefficient at the time of equalization is determined from the equalization target value and the reproduction signal.   The information reproduction method according to claim 12 or 13, wherein the reproduction signal before equalization with respect to a data string pattern recorded on the information recording medium is averaged to obtain the equalization target value. An information reproduction method for equalizing a reproduction signal from an information recording medium and reproducing a binary signal by viterbi detection,
An information reproduction method, characterized in that an equalization target value at the time of equalization is calculated using a secondary partial response coefficient.   The information reproduction method according to claim 15, wherein the equalization target value is calculated based on a binary signal pattern output from the Viterbi detector and a secondary partial response coefficient.   The information reproduction method according to claim 15 or 16, wherein the second-order partial response coefficient is calculated based on a reproduction signal input to the equalizer and a binary signal pattern output from the Viterbi detector. The data string recorded on the information recording medium is a _i , the secondary partial response coefficients are h 1 — _i , h 2 — _{i , j} , and the PR constraint length is m, and the equalization target value z _k is

The information reproduction method according to any one of claims 15 to 17, which is obtained by: 19. The information reproducing method according to claim 18, wherein h 2 — _{i, j} having a value different from 0 among h 2 — _{i, j} is 5 or less.   20. The information reproducing method according to claim 15, wherein the number of equalization target values matches the number of branch metrics for the Viterbi detection.   21. The information reproducing method according to claim 12, wherein a reference value at the time of branch metric calculation in the Viterbi detection matches the equalization target value.   The information reproducing method according to any one of claims 12 to 21, wherein the number of branch metrics in the Viterbi detection is 16.

Images