WO2012140997A1

WO2012140997A1 - Information reproduction device and information reproduction method

Info

Publication number: WO2012140997A1
Application number: PCT/JP2012/057552
Authority: WO
Inventors: 悠介中村; 一石原
Original assignee: 日立コンシューマエレクトロニクス株式会社
Priority date: 2011-04-15
Filing date: 2012-03-23
Publication date: 2012-10-18
Also published as: JP2012226788A

Abstract

Provided is a method to calculate, in an information reproduction device, the effective reliability of even an RLL modulated data string. The device comprises the following: a maximum likelihood path determination unit that outputs a maximum likelihood path, which leads to the most probable decoding results; a conflict path determination unit that outputs a conflict path for which the decoding results differ from those of the maximum likelihood path; a likelihood calculation unit that outputs a likelihood that is the difference between the maximum likelihood path and the conflict path; an inter-signal distance calculation unit that outputs an inter-signal distance which is the distance between a maximum likelihood waveform generated from maximum likelihood path decoding results and a conflict waveform generated from conflict path decoding results; and a reliability output unit that outputs reliability on the basis of a normalization likelihood in which the likelihood is normalized by the inter-signal distance.

Description

Information reproducing apparatus and information reproducing method

The present invention relates to an information reproducing apparatus and information reproducing method for reproducing information from an information recording medium.

Currently, in order to improve the capacity of optical discs, the BD (Blu-ray Disc (trademark)) standard increases the density of two layers of one layer 25 GB, and the BDXL (trademark) standard increases the density of one layer 32 GB to four layers. It is planned.

For further increases in density and capacity in the future, for example, the technique described in paragraph [0007] of Patent Document 1 may be mentioned. This document includes “SOVA (Soft-Output Viterbi Algorithm) as a promising method. In addition to the binary data (0, 1) that has been subjected to maximum likelihood decoding in the hard decision output type PRML detection unit 210, This is a method of outputting the reliability (information on the probability of the decoding result) and obtaining the decoding result as an analog value.If the encoding is performed in advance (outer encoding) on the writing side, the analog value is read on the reading side. It is known that soft decision decoding has improved error rate characteristics compared to hard decision decoding because an analog value can be used as reliability information. Is described.

In order to obtain the reliability in this SOVA, as described in paragraph [0018] of Patent Document 1, for example, in the soft decision output type PRML detector, the second most likely decoding sequence (2nd sequence) ), A difference between the likelihoods of both series is obtained, and a means for updating this is provided.

Japanese Patent Laid-Open No. 11-355151

However, when this technology is applied to an optical disk or the like, there is a problem in that RLL (Run Length Limited) modulation is performed. By the RLL modulation, the distance between the maximum likelihood decoding sequence and the 2nd sequence changes depending on the pattern of the data sequence. As in Patent Document 1, the method of merely obtaining “the difference in likelihood for both sequences” The likelihood cannot be calculated correctly.

Therefore, an object of the present invention is to provide an information reproducing apparatus and an information reproducing method for calculating an effective reliability even for a data string subjected to RLL modulation.

The above problems are solved by the invention described in the claims. For example, in the information reproducing apparatus, a maximum likelihood path determination unit that outputs a maximum likelihood path that is the most probable decoding result, and a competitive path determination unit that outputs a competitive path that is a decoding result different from the maximum likelihood path. A likelihood calculator that outputs a likelihood that is a difference between the maximum likelihood path and the competitive path, a maximum likelihood waveform generated from the decoding result of the maximum likelihood path, and a decoding result of the competitive path. An inter-signal distance calculation unit that outputs an inter-signal distance that is a distance from a competing waveform, and a reliability output unit that outputs a reliability based on a normalized likelihood obtained by normalizing the likelihood based on the inter-signal distance; It is set as the structure which has these.

According to the present invention, it is possible to calculate an effective reliability even for a data string subjected to RLL modulation, and there is an effect of improving the correction capability in soft decision decoding.

The block diagram of the recording / reproducing apparatus in a 1st Example. The state transition diagram of the Viterbi algorithm using PR (1, 2, 2, 2, 1). The figure showing the example of a maximum likelihood waveform, an error waveform, and a reproduction | regeneration waveform. The figure showing the example of a maximum likelihood waveform and an error waveform. The figure showing the example of a maximum likelihood waveform and an error waveform. The figure showing the likelihood distribution for every pattern, and the synthetic | combination likelihood distribution. The figure showing the normalization likelihood distribution for every pattern, and the synthetic | combination normalization likelihood distribution. The block diagram of the soft output decoding part in a 1st Example. The memory block diagram in a 1st Example. The trellis diagram explaining the soft output decoding in the 1st Example. The figure showing the waveform explaining the soft output decoding in a 1st Example. The flowchart showing the likelihood update procedure in a 1st Example. The block diagram of the soft output decoding part in a 2nd Example. The memory block diagram in a 2nd Example. The flowchart showing the likelihood update procedure in a 2nd Example. The block diagram of the soft output decoding part in a 3rd Example. The flowchart showing the likelihood update procedure in a 3rd Example. The figure showing the likelihood distribution for every pattern, and the synthetic | combination likelihood distribution. The block diagram of the soft output decoding part in a 4th Example. The trellis diagram explaining the soft output decoding in the 4th Example. The trellis diagram explaining the soft output decoding in the 4th Example. The state transition diagram of the Viterbi algorithm using PR (a, b; c, d). The trellis diagram explaining the soft output decoding in a 5th Example.

Hereinafter, examples of the present invention will be described.

The present embodiment provides an information reproducing apparatus and method capable of outputting an appropriate reliability in the PRML (Partial Response Maximum Likelihood) method for optical disc reproduction.

First, the outline of the recording / reproducing apparatus and method of this embodiment will be described with reference to FIG.

During the recording operation, user data is received from the host 101, the error correction encoding unit 102 performs error correction encoding using an LDPC (Low Density Parity Check) code, and the modulation unit 103 is used in, for example, BD ( 1, 7) A process of modulating 2-bit data into 3-bit data is executed according to the RLL modulation method. The (1, 7) RLL modulation method is modulated data in accordance with the run length restriction of RLL (1, 7) in which the number of consecutive 0s in the modulated bits is a minimum of 1 and a maximum of 7 ranges. Modulation method. An LDD (Laser Diode Driver) 104 performs laser intensity modulation necessary for recording modulation data on the disk, and a pickup 105 emits a laser and records it on the disk 106.

Next, at the time of playback operation, the recorded data is read from the disk 106 via the pickup 105, the AFE (Analog Front End) 108 executes amplitude adjustment and filtering, and the ADC (Analog-Digital Converter) 109 performs digitization processing. The waveform equalizing unit 110 equalizes the digitized reproduction signal into a waveform so as to have PR characteristics such as PR (1, 2, 2, 2, 1) used in the soft output decoding unit 111, and outputs a soft output decoding unit. 111 obtains a soft output decoded value by a method described later. The soft value demodulator 112 demodulates the (1, 7) RLL modulation of the soft output decoded value, and the error correction unit 113 transmits the data to the host 101 after performing error correction of the LDPC code by sum-product decoding or the like. .

Here, the concept of the soft output decoding unit 111 will be described with reference to FIGS.

The basis of the soft output decoding unit 111 is Viterbi decoding. For example, BDXL uses a PR characteristic of PR (1, 2, 2, 2, 1), and (1, 7) RLL modulation as shown in FIG. Decoding is performed by state transition reflecting the above. The Euclidean distance ED (T, W) between the maximum likelihood waveform (T in FIG. 3) and the waveform equalization unit 110 output waveform (W in FIG. 3) in Viterbi decoding is (Equation 1), and the error waveform (F in FIG. 3). The Euclidean distance ED (F, W) of the waveform (W) output waveform (W) is expressed by (Expression 2), and the likelihood Δ that is the difference between them is expressed by (Expression 3). t _{_n,} _f _n, _w _n represents T at time n, F, the amplitude value of the W.

Also, since Sum-product decoding in the error correction unit 113 is a log domain decoding method, the reliability that is the output of the soft output decoding unit 111 needs to be a log likelihood ratio (LLR). This LLR can be approximated by (Equation 4) if the probability density function of the likelihood Δ is close to a normal distribution. μ represents an average value of the distribution, and σ represents a standard deviation.

In the process of calculating the reliability, in the data subjected to (1, 7) RLL modulation, the decoded values of T and F become patterns that can actually exist, that is, the shortest run length is 2T and 1T exists. The point of not allowing is a problem. For example, in the example of FIG. 4, the decoded value of T is [000011], the decoded value of F is [0001111], and the number of error bits is 1. In the example of FIG. 5, the decoded value of T is [00011000], The decoded value of F is [00110000], and the number of error bits is 2. This is because the Viterbi algorithm operates so as to satisfy the rules of RLL modulation, and if T in FIG. 5 is incorrect, 1T shift is not allowed and the 2T portion slips. For this reason, in RLL-modulated data, the Euclidean distance between T and F (formula 5) changes depending on the decoding result.

This effect will be described with reference to FIG. The figure shows an image in which the likelihood Δ is histogrammed for the RLL-modulated data sequence with the horizontal axis representing the likelihood Δ and the vertical axis representing the occurrence frequency. The pattern distribution of FIG. 4 is a first distribution, and the pattern distribution of FIG. 5 is a second distribution. If each distribution follows a normal distribution, the average value μ of the distribution is a time when W = T and is expressed by (Equation 6).

This value is different between the first distribution and the second distribution because the Euclidean distance between T and F varies depending on the pattern, and as a result, the synthesized overall distribution deviates from the normal distribution. For this reason, it becomes difficult to apply the LLR calculation formula shown in (Formula 4).

Therefore, in order to solve this problem, the present embodiment is characterized in that the likelihood Δ is normalized by the Euclidean distance between T and F as shown in (Expression 7).

FIG. 7 shows an image in which the normalized likelihood Δ ′ is histogrammed with the normalized likelihood Δ ′ on the horizontal axis and the occurrence frequency on the vertical axis. By using the normalized likelihood, the average values match and the overall distribution is close to the normal distribution, so that the LLR can be calculated correctly.

Details of the soft output decoding unit 111 using this concept will be described below with reference to FIGS. First, the configuration of the soft output decoding unit 111 is shown in FIG. Although the PR characteristic to be used is described as PR (1, 2, 2, 2, 1), it is not limited to this.

First, the BM calculation unit 801 calculates a branch metric which is the square of the difference between the reference values REF00000 to REF11111 in FIG. 2 and the equalized waveform input to the soft output decoding unit 111. Thereafter, the ACS calculation unit 802 adds a branch metric to the path metric for each of the states S0000 to S1111 in FIG. 2, compares the addition results at the point where the paths in FIG. 2 merge, and selects a smaller path. The difference of the addition results of the two joined paths is stored in the likelihood candidate memory 803 as the likelihood candidate Δ. Further, the path selection result of the ACS calculation unit 802 is stored in the path memory 804.

If the maximum likelihood path is uncertain when calculating the likelihood, the reliability of the likelihood also decreases, so after determining the maximum likelihood path, it competes with the maximum likelihood path for likelihood calculation based on the result. Confirm the path. The first maximum likelihood path determination is performed by the prior maximum likelihood path determination unit 805, the maximum likelihood path determination based on the result is performed by the maximum likelihood path determination unit 806, and the competitive path determination is performed by the competitive path determination unit 807. FIG. 9 shows the configurations of the path memory 804 and the likelihood candidate memory 803 when the path memory length for the prior maximum likelihood path determination is L1 and the path memory length for the maximum likelihood path determination is L2. The horizontal axis indicates time, and the time n is the base point. FIG. 9 also shows examples of a maximum likelihood decoded value memory 808, a competitive decoded value memory 809, and a likelihood memory 813, which will be described later.

The path determination method will be described with reference to FIG. The trellis diagram of FIG. 10 is a development of the state transition diagram of FIG. First, the maximum likelihood path determination unit 805 determines the maximum likelihood path by tracing back the path selection result of the path memory 804 from time n + L1 to time n. The state at time n is set as a base state. Next, the maximum likelihood path determination unit 806 determines the maximum likelihood path by tracing back as usual from the time n to the time n−L2 with the base state at the time n as the start point. Further, the contention path determination unit 807 selects a path different from that stored in the path memory 804 only with respect to the transition to the time n-1, starting from the base point state at the time n, and then from the time n-1 to the time n. -Determine the contention path by tracing back to L2 as usual. However, if there is only one path connected to the base point state, the base point state is moved from time n to time n-1, and only the transition to time n-2 is stored in the path memory 804. Choose a different path. Similarly, it is assumed that the number of paths connected to the base state is traced back to one. The decoding result by the maximum likelihood path in the maximum likelihood path determination unit 806 is stored in the maximum likelihood decoding value memory 808, and the decoding result by the maximum likelihood path in the competition path determination unit 807 is stored in the competitive decoding value memory 809.

In this embodiment, it is necessary to normalize the likelihood by the Euclidean distance between the maximum likelihood waveform and the competitive waveform as described above. FIG. 11 shows the maximum likelihood waveform and the competition waveform corresponding to the maximum likelihood path and the competition path in FIG. The inter-signal distance calculation unit 810 generates a maximum likelihood waveform and a competing waveform by convolving PR characteristics such as PR (1, 2, 2, 2, 1) with the maximum likelihood decoded value and the competitive decoded value, and inter-waveform Euclidean distance. Is calculated.

The following likelihood calculation method will be described with reference to the flowchart of FIG.

First, the normalized likelihood candidate calculation unit 811 acquires the likelihood candidate Δ (n) at time n from the likelihood candidate memory 803 (S1201). The likelihood candidate Δ (n) is divided by the inter-waveform Euclidean distance output from the inter-signal distance calculation unit 810 to calculate a normalized likelihood candidate Δ (n) ′ (S1202). The likelihood updating unit 812 acquires the kth (1 ≦ k ≦ L2) maximum likelihood decoded value bm _k from the maximum likelihood decoded value memory 808 (S1203), and the kth (1 ≦ k ≦ 1) from the competitive decoded value memory 809. The contention decoding value bc _k of L2) is acquired (S1204). Then, the maximum likelihood decoded value bm _k and the competitive decoded value bc _k are compared (S1205). When the comparison result is match in S1205, k-th likelihood delta _k stored in the likelihood memory 813 is held as it is (S1206). If the comparison result shows non-coincidence, it compares the k-th likelihood stored in the likelihood memory 813 delta _k and the normalized likelihood candidate delta a (n) '(S1207). Comparison result in S1207, replacing 'a k-th likelihood delta _k which is stored in the likelihood memory 813 is smaller delta (n)' normalized likelihood candidate delta (n) in (S1208) (e.g., FIG. 10 Δ ₅ ). As a result of the comparison, k-th likelihood delta _k the normalized likelihood candidate delta (n) 'is stored in the likelihood memory 813 is larger is held as it is (S1206) (e.g., delta ₈ in FIG. 10). The processes from S1203 to S1208 are performed from k = 1 to L2, and it is confirmed whether S1208 (likelihood replacement) is executed in the process (S1209). If S1208 is executed to output the L2-th likelihood delta _L2 of the likelihood memory 813 as the likelihood (S1210), as a likelihood because the likelihood is not updated even once if it was not performed Output (S1211). This is equivalent to giving an average value of the distribution shown in FIG. 7 because the normalized likelihood is used.

Using the likelihood Δ calculated above, the LLR calculation unit 814 calculates the LLR according to the LLR calculation formula of (Expression 4). In addition, although the average value μ and the standard deviation σ in (Expression 4) may be actually measured, values set in advance may be used. Finally, the multiplication unit 815 outputs the result of multiplying the L2nd maximum likelihood decoded value of the maximum likelihood decoded value memory 808 by the LLR to the soft value demodulating unit 112 as the reliability.

According to the above circuit configuration and processing procedure, it is possible to calculate an effective reliability even for a data string subjected to RLL modulation, and it is possible to improve the correction capability in soft decision decoding.

In this embodiment, the Euclidean distance is used to calculate the inter-waveform distance, but it may be read as an absolute value. Although (1, 7) RLL has been described as an example, the present invention is not limited to this, and can be applied to any modulation scheme. The same applies to the following embodiments.

In this embodiment, unlike Embodiment 1, a likelihood that is not normalized is used for updating the likelihood. FIG. 13 shows the configuration of the soft output decoding unit 111 in this embodiment. The parts of the likelihood updating unit 812 and the likelihood memory 813 in the first embodiment (FIG. 8) are replaced with a likelihood updating unit 1301, a comparison likelihood memory 1302, and an output likelihood memory 1303 in FIG. The configuration of the comparison likelihood memory 1302 and the output likelihood memory 1303 is shown in FIG.

Here, the operation of the likelihood updating unit 1301 in FIG. 13 will be described using the flowchart in FIG.

The operations up to S1205 are the same as those in the first embodiment. If the comparison results in S1205 match, the kth likelihoods Δ _k and Δ _k ′ stored in the comparison likelihood memory 1302 and the output likelihood memory 1303 are held as they are (S1501). If the comparison result shows non-coincidence, to compare the k-th likelihood stored in the comparison likelihood memory 1302 delta _k and likelihood candidate Δ (n) (S1502). S1502 comparison result of replacing the k-th likelihood delta _k of likelihood candidate delta (n) is stored if the comparison likelihood memory 1302 less likelihood candidate Δ (n) (S1503), a further output The kth likelihood Δ _k ′ stored in the likelihood memory 1303 is replaced with a normalized likelihood candidate Δ (n) ′ (S1504). If the likelihood candidate Δ (n) is large as a result of the comparison in S1502, the kth likelihoods Δ _k and Δ _k ′ stored in the comparison likelihood memory 1302 and the output likelihood memory 1303 are retained as they are ( S1501). The processing from S1203 to S1504 is performed from k = 1 to L2, and it is confirmed whether S1503 and S1504 (replacement of likelihood) are executed in the process (S1505). S1503, if the S1504 is executed, and outputs L2-th likelihood delta _L2 'of the output likelihood memory 1303 as the likelihood (S1506), if not executed likelihood is not updated even once Therefore, the likelihood is output as 1 (S1507). This is equivalent to giving an average value of the distribution shown in FIG. 7 because the normalized likelihood is used.

According to the above circuit configuration and processing procedure, the likelihood update comparison can be performed before normalization, and the actual output can use the likelihood after normalization, thereby improving the correction capability in soft decision decoding. Is possible.

In this embodiment, unlike

Embodiments

1 and 2, likelihood that is not normalized is used for updating and output of likelihood. FIG. 16 shows the configuration of the soft output decoding unit 111 in this embodiment. The likelihood updating unit 812 of the first embodiment (FIG. 8) is replaced with the likelihood updating unit 1601 in FIG. 16, and the inter-signal distance calculation unit 810 and the normalized likelihood candidate calculation unit 811 in FIG. 8 are deleted. .

Here, the operation of the likelihood update unit 1601 in FIG. 16 will be described using the flowchart of FIG.

The operations up to S1205 are the same as those in the first embodiment. After S1205 comparison result matches, k-th likelihood delta _k stored in the likelihood memory 813 is held as it is (S1701). If the comparison result shows non-coincidence, it compares the k-th likelihood stored in the likelihood memory 813 delta _k and likelihood candidate delta a (n) (S1702). S1702 comparison result of replacing the k-th likelihood delta _k of likelihood candidate delta (n) is stored in the likelihood memory 813 is smaller likelihood candidate Δ (n) (S1703), the likelihood candidate delta ( n) is the k-th likelihood delta _k stored in the likelihood memory 813 is larger is held as it is (S1701). The processes from S1203 to S1703 are performed from k = 1 to L2, and it is confirmed whether S1703 (replacement of likelihood) is executed in the process (S1704). If S1703 is executed, and outputs the L2-th likelihood delta _L2 of the likelihood memory 813 as the likelihood (S1705). If it is not executed, the likelihood has never been updated. Therefore, the likelihood having the smallest Euclidean distance ED (T, F) between the maximum likelihood waveform and the competing waveform is output as the likelihood (S1706). This is equivalent to giving an average value (μ _{2 in} FIG. 18) of the distribution of the most likely path in the distribution because the likelihood has a distribution as shown in FIG. 18 without normalization. . Further, as the likelihood output in S1706, the average value (μ _all ) of the entire distribution in FIG. 18 may be used.

According to the above circuit configuration and processing procedure, an appropriate likelihood can be output even when a likelihood that is not normalized is used, and the correction capability in soft decision decoding can be improved.

This embodiment is different from the embodiment 1-3 in the operation of the competitive path determination. In the first embodiment, after the maximum likelihood path determination is performed, the maximum likelihood path and the competitive path are determined, whereas in the present embodiment, the maximum likelihood path determination is not performed.

FIG. 19 shows the configuration of the soft output decoding unit 111 in this embodiment. The parts of the maximum likelihood path determination unit 806 and the competitive path determination unit 807 in the first embodiment (FIG. 8) are replaced with a maximum likelihood path determination unit 1901 and a competitive path determination unit 1902 in FIG. 19, and the likelihood candidate memory 803 in FIG. The prior maximum likelihood path determination unit 805 is deleted.

Referring to FIG. 20, this path determination method will be described by taking the vicinity of time n as an example. The trellis diagram of FIG. 20 is a development of the state transition diagram of FIG. First, a state with the smallest path metric at time n is set as a base state. Next, the maximum likelihood path determination unit 1901 establishes the maximum likelihood path by tracing back as usual from the time n to the time n−L2 with the base state at the time n as the start point. Further, the contention path determination unit 1902 selects a path different from that stored in the path memory 804 only for the transition to time n−1, starting from the base point state at time n, and then from time n−1 to time n -Determine the contention path by tracing back to L2 as usual.

Alternatively, the following method may be adopted for determining the competitive path.

Referring to FIG. 21, this path determination method will be described by taking the vicinity of time n as an example. The trellis diagram of FIG. 21 is a development of the state transition diagram of FIG. Maximum likelihood path determination section 1901 determines the maximum likelihood path by tracing back from time n to time n−L2 as usual, starting from the state that is the minimum path metric at time n. Also, the contention path determination unit 1902 establishes the contention path by tracing back from time n to time n−L2 as usual, starting from the state where the path metric at the time n is the second smallest path metric. To do. In this case, the likelihood candidate Δ (n) is a difference between the minimum path metric and the next path metric as shown in FIG.

According to the above circuit configuration and processing procedure, it is possible to calculate an effective reliability even for a data string subjected to RLL modulation, and further reduce the amount of memory.

Further, although the present embodiment has been described with respect to the first embodiment, the present embodiment can be similarly applied to other embodiments.

In the above embodiment, one dimension has been described, but in this embodiment, a two dimension example will be described. The difference from the first embodiment is the method of calculating the likelihood candidate Δ.

The method for calculating the likelihood candidate Δ will be described with reference to FIGS. 22 and 23 by taking the vicinity of the nth pixel as an example. FIG. 22 shows a state transition diagram of a two-dimensional PR characteristic of a 2 × 2 matrix as shown in (Equation 8). a, b, c, and d are arbitrary real numbers. Note that the two-dimensional PR characteristics to be used are not limited to this, and can be similarly extended by using an arbitrary matrix.

First, the BM calculation unit 801 calculates a branch metric that is the square of the difference between the reference values REF [00; 00] to REF [11; 11] in FIG. 22 and the equalized waveform input to the soft output decoding unit 505. . After that, the ACS calculation unit 802 adds the branch metric to the path metric for each of the states S [0; 0] to S [1; 1] in FIG. 22, and compares the addition results at the point where the paths in FIG. However, since the four paths merge as shown in the figure, the path with the smallest addition result is selected from the four paths.

The trellis diagram of FIG. 23 is an expansion of the state transition diagram of FIG. 22 and considers the merging of paths at S [0; 0] of the nth pixel. The ACS calculation unit 802 calculates the addition result difference (Δ ₁ , Δ ₂ , Δ _{3 in} FIG. 23) between the selected path and the other paths, and the smallest _{one of} these Δ ₁ , Δ ₂ , Δ ₃ is calculated. It is determined that the path is most likely to be erroneous, and is stored in the likelihood candidate memory 803 as a likelihood candidate Δ. Delta ₁ in FIG. 23 is a minimum.

The competitive path determination unit 807 selects a path different from that stored in the path memory 804 only for the transition to the (n-1) th pixel, starting from the base point state at the nth pixel. Is the path used when the likelihood candidate Δ is calculated. The A path in FIG. 23 delta ₁ are competing paths.

Further, the likelihood updating unit 812 acquires the kth (1 ≦ k ≦ L2) maximum likelihood decoded value bm _k from the maximum likelihood decoded value memory 808 (S1203), and the kth (1 ≦ k) from the competitive decoded value memory 809. ≦ L2) is obtained as a competitive decoded value bc _k (S1204). In the example of (Equation 8), since 2 bits are decoded simultaneously as shown in FIG. 22, the maximum likelihood decoded value bm _k and the competitive decoded value bc _k are 2-bit signals. Therefore, when comparing the maximum likelihood decoded value bm _k and the competitive decoded value bc _k (S1205), the comparison is performed using the 0th bit and the 1st bit. Thereafter, the likelihood update is performed for each of the 0th bit and the 1st bit.

According to the above circuit configuration and processing procedure, it is possible to calculate an effective reliability even when the two-dimensional PR characteristic is used, and it is possible to improve the correction capability by using soft decision decoding. .

101: Host, 102: Error correction coding unit, 103: Modulation unit, 104: LDD, 105: Pickup, 106: Disc, 107: Spindle motor, 108: AFE, 109: ADC, 110: Waveform equalization unit, 111 : Soft output decoding unit, 112: Soft value demodulation unit, 113: Error correction unit, 801: BM operation unit, 802: ACS operation unit, 803: Likelihood candidate memory, 804: Path memory unit, 805: Prior maximum likelihood path Determination unit, 806: Maximum likelihood path determination unit, 807: Competitive path determination unit, 808: Maximum likelihood decoded value memory, 809: Competitive decoded value memory, 810: Inter-signal distance calculation unit, 811: Normalized likelihood candidate calculation unit 812: Likelihood update unit, 813: Likelihood memory, 814: LLR calculation unit, 815: Multiplication unit, 1301: Likelihood update unit, 1302: Likelihood memory for comparison, 1303: Force for the likelihood memory, 1601: likelihood updating unit, 1901: maximum likelihood path determining unit, 1902: competitive path determination unit.

Claims

In the information reproducing apparatus,
A maximum likelihood path determination unit that outputs a maximum likelihood path that is the most probable decoding result;
A contention path determination unit that outputs a contention path that is a decoding result different from the maximum likelihood path;
A likelihood calculator that outputs a likelihood that is a difference between the maximum likelihood path and the competitive path;
An inter-signal distance calculation unit that outputs an inter-signal distance that is a distance between the maximum likelihood waveform generated from the decoding result of the maximum likelihood path and the competitive waveform generated from the decoding result of the competitive path;
A reliability output unit that outputs a reliability based on a normalized likelihood obtained by normalizing the likelihood with the distance between the signals;
An information reproducing apparatus comprising:
The information reproducing apparatus according to claim 1,
A normalized likelihood calculating unit that outputs a normalized likelihood obtained by normalizing the likelihood with the distance between the signals;
The likelihood candidate at a time when the decoding result of the maximum likelihood path and the decoding result of the competitive path are different from the normalized likelihood, and when the normalized likelihood is small, the normalized likelihood is determined as the likelihood candidate. A likelihood updater,
A reliability output unit that outputs reliability based on the likelihood candidates;
An information reproducing apparatus comprising:
The information reproducing apparatus according to claim 1,
A likelihood update unit that compares the likelihood with a likelihood candidate at a time when the decoding result of the maximum likelihood path and the decoding result of the competitive path are different, and uses the likelihood as a likelihood candidate when the likelihood is small; ,
A normalized likelihood candidate calculation unit that outputs a normalized likelihood candidate obtained by normalizing the likelihood candidate by the distance between the signals;
A reliability output unit that outputs reliability based on the normalized likelihood candidates;
An information reproducing apparatus comprising:
In the information reproducing apparatus,
A maximum likelihood path determination unit that outputs a maximum likelihood path that is the most probable decoding result;
A contention path determination unit that outputs a contention path that is a decoding result different from the maximum likelihood path;
A likelihood calculator that outputs a likelihood that is a difference between the maximum likelihood path and the competitive path;
An inter-signal distance calculation unit that outputs an inter-signal distance that is a distance between the maximum likelihood waveform generated from the decoding result of the maximum likelihood path and the competitive waveform generated from the decoding result of the competitive path;
A likelihood update unit that compares the likelihood with a likelihood candidate at a time when the decoding result of the maximum likelihood path and the decoding result of the competitive path are different, and uses the likelihood as a likelihood candidate when the likelihood is small; ,
A reliability output unit that outputs a reliability based on the distance between the signals when the comparison operation is not performed, and outputs a reliability based on the likelihood candidate when the comparison operation is performed;
An information reproducing apparatus comprising:
The information reproducing apparatus according to claim 4, wherein
When the comparison operation is not performed, the reliability is output based on the minimum value of the inter-signal distance, and when the comparison operation is performed, the reliability output is output based on the likelihood candidate. An information reproducing apparatus having a section.
The information reproducing apparatus according to claim 4, wherein
When the comparison operation is not performed, the reliability is output based on the average value of the distance between the signals, and when the comparison operation is performed, the reliability output is output based on the likelihood candidate. An information reproducing apparatus having a section.
The information reproducing apparatus according to any one of claims 1 to 6,
2. The information reproducing apparatus according to claim 1, wherein the contention path is a path that is forcibly branched at a path merge point at a time when the maximum likelihood path exists.
In the information reproduction method,
Output the most likely path that is the most probable decoding result,
Outputting a contention path resulting in a decoding result different from the maximum likelihood path,
Output a likelihood that is a difference between the maximum likelihood path and the competitive path;
A distance between signals that is a distance between a maximum likelihood waveform generated from the decoding result of the maximum likelihood path and a competitive waveform generated from the decoding result of the competitive path;
A method of reproducing information, comprising: outputting a reliability based on a normalized likelihood obtained by normalizing the likelihood with the distance between signals.
The information reproduction method according to claim 8,
Outputting a normalized likelihood obtained by normalizing the likelihood by the distance between the signals;
The likelihood candidate at a time when the decoding result of the maximum likelihood path and the decoding result of the competitive path differ from the normalized likelihood, and when the normalized likelihood is small, the normalized likelihood is set as the likelihood candidate. ,
A method of reproducing information, comprising: outputting a reliability based on the likelihood candidate.
The information reproduction method according to claim 8,
Compare the likelihood with the likelihood candidate at a time when the decoding result of the maximum likelihood path and the decoding result of the competitive path are different, and if the likelihood is small, the likelihood is set as a likelihood candidate,
A normalized likelihood candidate obtained by normalizing the likelihood candidate by the distance between the signals,
A method of reproducing information, comprising: outputting a reliability based on the normalization likelihood candidate.
In the information reproduction method,
Output the most likely path that is the most probable decoding result,
Outputting a contention path resulting in a decoding result different from the maximum likelihood path,
Output a likelihood that is a difference between the maximum likelihood path and the competitive path;
A distance between signals that is a distance between a maximum likelihood waveform generated from the decoding result of the maximum likelihood path and a competitive waveform generated from the decoding result of the competitive path;
Compare the likelihood with the likelihood candidate at a time when the decoding result of the maximum likelihood path and the decoding result of the competitive path are different, and if the likelihood is small, the likelihood is set as a likelihood candidate,
When the comparison operation is not performed, the reliability is output based on the distance between the signals, and when the comparison operation is performed, the reliability is output based on the likelihood candidate. Playback method.
The information reproduction method according to claim 11,
When the comparison operation is not performed, the reliability is output based on the minimum value of the inter-signal distance, and when the comparison operation is performed, the reliability is output based on the likelihood candidate. Information reproduction method.
The information reproduction method according to claim 11,
When the comparison operation is not performed, the reliability is output based on an average value of the inter-signal distance, and when the comparison operation is performed, the reliability is output based on the likelihood candidate. Information reproduction method.
14. The information reproduction method according to claim 8, wherein:
2. The information reproduction method according to claim 1, wherein the competing path is a path that is forcibly branched at a path junction at a certain time of the maximum likelihood path.