JP2005108332A - Data reproducing device using repetition decoding and method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a data recording and reproducing device using repetition decoding in which an effect of error correction by the repetition decoding is sufficiently demonstrated even when a burst error is included in a reproduced signal. <P>SOLUTION: This reproducing device for performing reproduction of data from a reproduced signal using repetition decoding has: a read error detecting means for detecting a read error region in the reproduced signal; a reproduced signal probability obtaining means for obtaining a probability value for an expected value of the reproduced signal; a control means for controlling the reproduced signal probability obtaining means so that the probability value for the expected value of the reproduced signal of the read error region becomes equal among all branches in the read error region on the basis of the detection result of the detecting means; and a repetition decoding calculating means for performing repetition decoding calculations using the probability value obtained from the reproduced signal probability obtaining means. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a data reproducing apparatus and a data reproducing method using iterative decoding, and more particularly to a data reproducing apparatus and a data reproducing method for compensating for a burst error.

[Necessity of iterative decoding]
A reduction in the SNR (Signal to Noise Ratio) of the reproduction signal becomes a problem in realizing high-density recording media. As a technique for solving this problem, an iterative decoding method, which is a signal processing method having a high detection capability for a low SNR signal, is required as one of means for realizing a high density recording medium.

[Basic configuration of iterative decoding]
In the iterative decoding method, the detection capability for low SNR signals is obtained by repeatedly decoding two or more linked encoders while feeding back one decoding result to the other between the corresponding decoders. It has been attracting attention because of its high coding gain.

  FIG. 1 is a diagram illustrating a configuration of an iterative decoding method. In the configuration of FIG. 1, a first encoder 100 that adds parity bits generated by recursive systematic convolutional code processing to user data ui, and code data xi including user data ui and parity bits are predetermined. The user data ui is encoded by an interleaver 200 that performs rearrangement based on the above rule and a second encoder 300 that further encodes the code data ci converted by the interleaver 200.

  In addition, the configuration of FIG. 1 includes second code data yi (hereinafter also referred to as a received signal sequence) to which noise is added via a communication path 900 and a second decoder 600 described later via an interleaver 700. The first decoder 400 that performs decoding processing corresponding to the second encoder using the output prior information La (ci), and the prior information La that is output from the first decoder via the deinterleaver 500 'The second decoder 600 that performs the decoding process corresponding to the first encoder 100 using the data obtained by separating the user data ui and the parity bit information from (xi), and the user data of the second decoder 600 User data ui ′ is estimated from reproduction signal data yi using hard decision unit 800 that makes a hard decision on the decoding result L (ui *) regarding ui.

  In the above configuration, the prior information La (ci) of the first decoder 400 is obtained by converting the external likelihood information Le ′ (xi) output from the second decoder 600 by the interleaver 700. . The prior information La ′ (xi) of the second decoder 600 is obtained by converting the external likelihood information Le (ci) output from the first decoder 400 by the deinterleaver 500.

  Further, the first decoder 400 has a probability P (ci = 1 | y) that the next input bit ci is 1 and a probability that ci is 0 under the condition that the received signal sequence yi is detected. Logarithmic likelihood ratio with P (ci = 0 | y), that is, L (ci *), which is the log-likelihood ratio (LLR) of the posteriori probability (APP), is calculated The external likelihood information Le (ci) is output by subtracting the prior information La (ci) from the likelihood ratio L (ci *). The second decoder 600 converts the prior information La ′ (xi) input from the first decoder 400 via the deinterleave 500 into the likelihood information L (ui) corresponding to the data bit ui and the parity bit pi. Is converted into likelihood information L (pi) corresponding to, and the log likelihood ratio L (ui *) and parity for the data bit ui using the likelihood information L (ui) and L (pi) Calculate the log likelihood ratio L (pi *) for the bit, combine the log likelihood ratios L (ui *) and L (pi *), and obtain the prior information La ′ (xi) from the result of puncturing that performs decimation. External likelihood information Le ′ (xi) is output by subtraction. The output external likelihood information Le ′ (xi) is converted by the interleaver 700 and input to the first decoder 400 as prior information La (ci), and the above decoding process is repeated repeatedly.

[Configuration of recording / reproducing apparatus applying iterative decoding]
FIG. 2 shows a configuration diagram of a data recording / reproducing apparatus 1000 to which the iterative decoding method is applied. The recording system of the data recording / reproducing apparatus 1000 of FIG. 2 includes an outer encoder 1100, a puncture (MUX & Puncturer) 1110, an interleaver (Interleaver) 1200, and a laser drive circuit (LD Driver) 1300. The puncture 1110 is encoded by combining the user data ui and the parity bit string pi generated by the outer encoder 1100 according to a predetermined rule, and thinning out bits from the obtained bit string according to the predetermined rule (puncture function). A data bit string xi is generated. The interleaver 1200 changes the arrangement of the encoded data bit sequence xi from the puncture to generate the encoded data bit sequence ci. The laser drive circuit 1300 controls laser emission based on the encoded data bit string ci and records on the optical disc.

  On the other hand, the reproduction system of this data recording / reproducing apparatus includes an amplifier (Amp) 1401, an AGC 1402, a low-pass filter (LPF) 1403, an equalizer (EQ) 1404, an analog / digital converter (A / D) 1500, and an iterative decoder. 1600 and controller 1700. The reproduction signal obtained from the optical head is waveform-shaped by an amplifier 1401, AGC 1402, low-pass filter 1403, and equalizer 1404, and sampling processing is performed by an analog / digital converter 1500. Next, the iterative decoder 1600 decodes the user data ui ′ estimated from the reproduction signal data yi after the sampling process, whereby the user data is reproduced from the recording data of the recording medium. The iterative decoder 1600 is controlled by a controller 1700 (for example, ODC in the case of a magneto-optical disk device). The iterative decoder 1600 decodes user data by performing iterative decoding the number of times determined by the controller 1700.

  When recorded data is written at a high density exceeding the limit of the playback resolution, the playback signal has waveform interference between the bits of the recorded data. By equalizing this waveform, the PR (Partial Response) characteristics are improved. Can be given. That is, the user data ui encoded by the outer encoder 1100 can be regarded as being encoded in the PR channel. Therefore, a code configuration in the iterative decoding method is realized by the recording system outer encoder 1100 and the PR channel encoding function. This code configuration is also called a turbo code.

  The iterative decoder 1600 in the reproduction system includes a decoder corresponding to the recording-system outer encoder 1100 and the reproduction-system PR channel. In the configuration shown in FIG. 1, the first decoder 400 corresponds to a decoder corresponding to the PR channel (hereinafter, the first decoder 400 is also referred to as a PR channel MAP decoder), and the second decoder 600 Corresponds to a decoder corresponding to the outer encoder 1100 (hereinafter, the second decoder 600 is also referred to as an outer code-compatible MAP decoder).

  In the first and second decoders, the MAP (Maximum A Posteriori Probability) decoding method is used. In the MAP decoding method, a value that maximizes the posterior probability of the user data ui obtained by using the probability (preliminary information) that the user data ui occurs and the value (likelihood information) defined in the channel characteristics is expressed in bits. In this method, the user data ui is estimated from the reproduced signal data yi determined and detected. As a technique for realizing the MAP decoding method, a BCJR algorithm or the like is used.

[Configuration of PR channel MAP decoder]
FIG. 3 shows a configuration of a conventional PR channel MAP decoder using the BCJR algorithm. The PR channel MAP decoder 400 includes a γ calculating unit 401, an α calculating unit 402, an α memory 403, a β calculating unit 404, a posterior probability log likelihood ratio (LLR) calculating unit 405, a reproduction signal probability acquiring unit 406, and a subtracting unit 407. It is comprised including. The posterior probability log-likelihood ratio calculation means 405 uses the occurrence probability of the reproduction signal data yi output from the reproduction signal probability acquisition means 406 and the transition between adjacent states using the prior information La (ci) input to the decoder. Logarithmic likelihood of the posterior probability of the code data ci estimated from the reproduced signal data yi using the γ calculating means 401 for obtaining the probability, the α memory 403 storing the output of the α calculating means 402, and the β calculating means 404 Output the ratio L (ci). The PR channel MAP decoder 400 subtracts the prior information La (ci) from the log likelihood ratio L (ci) output from the posterior probability log likelihood ratio calculation means 405 using the subtraction means 407, thereby The external information Le (ci), which is prior information for the decoder, is output.

  Recording media such as optical disks, magnetic disks, and magnetic tapes generally have partial defects. In particular, in the case of a replaceable medium, defects increase due to the influence of dust adhering and scratches that occur during handling.

  Although iterative decoding shows a large detection capability against a decrease in SNR in high-density recording, when a reproduction signal of a defective portion (burst error) of a recording medium is input, a sample value different from the original value is obtained. Therefore, there is a problem in that the probability value of the path that should be originally increased is lowered and the probability value of the wrong path is increased.

  The above problem occurs in a single decoder constituting the iterative decoding. In the iterative decoding, the probability propagation between the decoders is repeatedly performed, so that an error generated in one decoder in the iterative process is decoded in the other decoding. Error information due to burst errors tends to increase.

  Therefore, in the conventional iterative decoder, even when error correction using an error correction code is performed in an iterative process, the limit of error correction code correction capability is exceeded due to an increase in error information when a burst error occurs. Therefore, there is a problem that an expected decoding result cannot be obtained.

  SUMMARY OF THE INVENTION An object of the present invention is to provide a data recording / reproducing apparatus using iterative decoding, in which the effect of error correction by iterative decoding can be sufficiently obtained even when a burst signal is included in the reproduced signal. And

  The present invention relates to a reproduction apparatus that reproduces data from a reproduction signal using iterative decoding, a read error detection unit that detects a read error region in the reproduction signal, and a probability value for an expected value of the reproduction signal. Based on the detection result of the reproduction signal probability acquisition means to be acquired and the detection means, the reproduction value is such that the probability value with respect to the expected value of the reproduction signal in the read error area is the same among all branches in the read error area. It comprises control means for controlling signal probability acquisition means, and iterative decoding calculation means for performing iterative decoding calculation using a probability value obtained from the reproduction signal probability acquisition means.

  In the process of probability calculation in the iterative decoding process, the present invention completely eliminates a probability value that becomes an element of error propagation by a reproduction signal outside the effective range, thereby including a burst error in the reproduction signal. There is an effect that a sufficient error correction effect by iterative decoding can be obtained.

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

[Description of configuration]
FIG. 7 shows the configuration of a data recording / reproducing apparatus to which the PR channel MAP decoder according to the present invention is applied. The configuration of the data recording / reproducing apparatus in FIG. 7 is the same as that of the conventional data recording / reproducing apparatus shown in FIG. The configuration of the present invention is different from the conventional configuration in that a read error detector 1800 and a PR channel controller 1900 are provided.

  The read error detector 1800 described above detects a read error region from the reproduction signal converted into digital data by the A / D converter 1500. Based on the detected result, the PR channel controller 1900 sets the output of the reproduction signal probability acquisition means 406 at the sample position corresponding to the reading error area to substantially the same value for all branches between adjacent states in the reading error area. To control. The reproduction signal probability acquisition means 406 may be realized as a configuration for calculating a probability value based on a sample value of a reproduction signal that is an input value, or from a list in which combinations of input values and probability values are set. You may implement | achieve as a structure which acquires applicable probability value.

  The read error detector 1800 has a memory means composed of a logic circuit or the like, and has a function of storing information on the read error position until the iterative decoding operation of the corresponding data block is completed.

  FIG. 8 shows an embodiment of a PR channel MAP decoder according to the present invention. The PR channel MAP decoder of FIG. 8 has the same basic configuration as the conventional PR channel MAP decoder shown in FIG. The configuration of the present invention is different from the conventional configuration in that the reproduction signal probability acquisition unit 406 is controlled by the PR channel controller 1900.

[Outline of BCJR algorithm]
Next, decoding processing of the PR channel MAP decoder 400 using the BCJR algorithm will be described with reference to FIG. The trellis diagram shown in FIG. 5 shows how the reproduction signal data yi detected from time t = 0 to time t = 7 transitions between the state S1 and the state S0. A section in the trellis diagram represents the state of the reproduction signal data yi that is stochastically determined based on the detection level of the reproduction signal detected at each time point, and the state S1 is a data string estimated from the reproduction signal data yi. The ci indicates a state of 1, and the state S0 indicates a state where the data string ci estimated from the reproduction signal data yi is 0.

  In addition, branches (hereinafter referred to as “branches”) connecting adjacent nodes in the trellis diagram of FIG. 5 represent transitions between adjacent states, and each branch has a γ value that is the probability of transition between states. Is assigned as a characteristic value. That is, a branch represents the probability of transition from a state at a certain point in time to a certain adjacent state. A set of branches selected up to a specific clause through a plurality of clauses is called a path metric.

In the BCJR algorithm, a γ value that is a transition probability (branch) between adjacent states, an α value that is a transition probability (path metric) from the front, and a β that is a transition probability (path metric) from the rear. Using the value, the log likelihood ratio of the posterior probability of each section is obtained. In the example of FIG. 5, an example is shown in which the posterior probability is obtained from time t = 2 to time t = 3. That is, the γ value that is the transition probability (branch) between states in the section from time t = 2 to time t = 3, and the forward transition probability (path) from time t = 0 to time t = 2. Metric) and a β value that is the probability of transition in the reverse direction (path metrics) from time t = 7 to time t = 3, from time t = 2 to time t = 3 The posterior probability and the log likelihood ratio of the posterior probability are obtained.
[Relationship between recording data bit string and PR channel output]
FIG. 4 shows the correspondence between all combinations of recording data bits in the PR (1,1) system and the expected value of the detection level of the reproduction signal P (t). The detection level of the reproduction signal P (t) in the PR (1,1) system is a value reflecting the state of two bits.

  For example, 3) in FIG. 4 detects the reproduction signal P (t-1) = 1 from the bit (0,1) at time t-1, and reproduces the reproduction signal P (t from bit (1,0) at time t. ) = 1 is detected. At this time, the state S0 changes to the state S1 from the time t-2 to the time t-1, and the state S1 changes to the state S0 from the time t-1 to the time t. The same applies to the other combinations in FIG. Note that the reproduction signal data yi is a value obtained by sampling after shaping the waveform of the reproduction signal P (t) described above.

In the trellis diagram shown in FIG. 5, the state transition indicated by the thick arrow is the transition from time t = 0 to time t = 2 in FIG.
The transition from time t = 1 to time t = 3 is shown in FIG.
It corresponds to. That is, in the trellis diagram shown in FIG. 5, the reproduction signal P (t = 1) = 0 is detected at time t = 1, and the reproduction signal P (t = 2) = 1 is detected at time t = 2. Become.
[Relationship between basic trellis structure and playback signal detection level]
FIG. 6 shows only the basic part of the trellis diagram shown in FIG. From time t-1 to time t in FIG. 6, the transition from the state S0 to the state S0 corresponds to detection of the reproduction signal P (t) = 0. In other words, when the reproduction signal P (t) = 0 is detected, the probability γ (0,0) of transition from the state S0 to the state S0 from the time t-1 to the time t is the transition between other states. It is more likely than the probability of occurrence (high probability of transition).

  From time t-1 to time t in FIG. 6, the transition from the state S0 to the state S1 and the transition from the state S1 to the state S0 correspond to the detection of the reproduction signal P (t) = 1. In other words, when the reproduction signal P (t) = 1 is detected, the probabilities γ (0,1) and γ () that transition from the state S0 to the state S1 or from the state S1 to the state S0 from time t-1 to time t. 1,0) is more likely (high probability of transition) than the probability of transition between other states.

From time t-1 to time t in FIG. 6, the transition from the state S1 to the state S1 corresponds to the detection of the reproduction signal P (t) = 2. In other words, when the reproduction signal P (t) = 2 is detected, the probability γ (1,1) of transition from the state S1 to the state S1 from the time t-1 to the time t is the transition between other states. It is more likely than the probability of occurrence (high probability of transition).
[Gamma calculation formula]
The above γ value is represented by likelihood information P (yi) and prior probability Pa (ci). These are handled in logarithm for convenience of calculation. Accordingly, the γ value is expressed by the following equation (1).

γ (S (t-1), S (t)) = Pa (ci) + P (yi) (1)
Prior probability Pa (ci) represents the occurrence probability of information bit ci that is known before decoding, and is the outer code for the other decoder (for example, PR channel MAP decoder) between the first and second decoders. It is obtained from prior information La (ci) obtained from the output of the decoding result of the corresponding MAP decoder. Prior information La (ci) is expressed by the logarithmic ratio of the above-mentioned prior probability Pa (ci), and since the method for obtaining prior probability Pa (ci) from this prior information La (ci) is known, the explanation will be given. Omitted.

  The likelihood information P (yi) represents the occurrence probability of the reproduction signal data yi for the information bit ci input to the communication path, and each branch between adjacent states according to the detection level of the reproduction signal P (t). Characterizes the probability value.

  In the conventional iterative decoder, when a read error occurs, the reproduction signal data yi input to the PR channel MAP decoder has a sample value that is different from a valid range as a reproduction signal. There is a problem that the probability value of the branch to be increased is lowered and the probability value of the wrong branch is increased.

This problem is caused by the likelihood information P (yi) in the expression (1) of the γ operation being a value significantly different from the originally expected value due to the influence of the reading error that has occurred in the reproduction signal. That is, since the likelihood information P (yi) is a value that is significantly different from the originally expected value, the probability value of the branch becomes an incorrect value at the read error occurrence location. For example, Fig. 4 1)
As shown in FIG. 4, when the recording bit (0, 0) is reproduced at time t, the reproduction signal P (t) = 0 should be obtained, but the reproduction signal P (t) = 4 due to the occurrence of a read error. Suppose that In this case, in the trellis diagram shown in FIG. 6, if the probability value of γ (0,0) is originally high, the probability value of γ (1,1) becomes high due to the occurrence of a read error. The probability value of γ (0,0), which should originally be high, will be low.

  Conventionally, in an iterative decoding operation, an erroneous probability value caused by a reading error is corrected to a plausible value based on the relationship with the probability values of other sample data. However, when a burst error occurs due to defects on the recording medium, multiple erroneous probability values are included, and the expected decoding result is obtained because the limit of the correction capability by iterative decoding is exceeded. There is a problem that cannot be done.

  Further, the a posteriori probability calculation is performed by the α operation and the β operation using the value of the γ operation, and the obtained posterior probability is used as the prior information of the other decoder. The degree information is propagated between the decoders constituting the iterative decoding.

  In the iterative decoder according to the present invention, the read error detector 1800 detects the occurrence position of the read error in the reproduced signal, and the PR channel controller 1900 is used to determine the probability of the expected value of the reproduced signal in the detected error region. Control is performed such that the likelihood information P (yi) for the reproduction signal data yi in the read error area is forcibly converted to substantially the same value so that the value is the same among all branches in the read error area. The conversion value may be any real number. However, the same value is used for all branches. Since the output of the reproduction signal probability acquisition means is processed as a log likelihood ratio in the posterior probability calculation means, it is clear that the value set substantially the same among all branches is zero. In this embodiment, for simplification of description, the description will be continued for the case where all the values to be converted are P (yi) = 0.

  Since the first decoding process of the iteration treats the prior information La (ci), which is the logarithmic ratio of the prior probability Pa (ci), as La (ci) = 0, all branches in the read error region described above have prior probability Pa ( ci) and likelihood information P (yi) are 0, and γ = 0. That is, the probability of path generation is the same among all combinations of paths that can be taken in the read error area described above, and the path passing through the read error area is indeterminate. That is, the decoding result of the first decoder in the first iteration has the same probability of being 0 and 1 in the error region.

  After the second iteration, the decoding result obtained by sending the probability information obtained by the above decoding process to the other decoder is used as the prior information La (ci). That is, by repeating the above decoding process repeatedly, the likelihood information P (yi) in the γ operation is replaced with 0, but the prior probability Pa obtained from the prior information La (ci) input from the other decoder Data can be decoded using the probability of (ci).

  When the γ operation in the read error area is expressed by the following equation (2).

γ (S (t-1), S (t)) = Pa (ci) (2)
The above-mentioned PR channel controller 1900 uses the γ calculation in the γ calculator 401 as the expression (1) when the reproduction signal data yi is within the effective range, and substantially when the reproduction signal data yi is in the reading error area. The output of the reproduction signal probability acquisition unit 406 is controlled so as to satisfy Expression (2).

  The reproduction signal probability acquisition means 406, which is controlled by the PR channel controller due to the occurrence of the read error, inputs 0 as likelihood information P (yi) to the γ calculator 401. As a result, the γ arithmetic expression of the γ arithmetic unit 401 is substantially the expression (2). In the above example, the value to be replaced in the read error area is shown as an example of likelihood information P (yi) = 0, but the value is not limited to this value. By setting substantially the same value among all branches, even if the likelihood information P (yi) is a value other than 0, it can be substantially regarded as Expression (2).

  Further, in a region other than the read error region, the reproduction signal probability acquisition unit 406 obtains likelihood information P (yi) using the reproduction signal data yi, and inputs the likelihood information P (yi) to the γ calculator 401. As a result, the γ arithmetic expression of the γ arithmetic unit 401 becomes the expression (1).

  Next, the output of the PR channel MAP decoder 400 becomes the input (preliminary information La ′ (xi)) of the decoder 600 corresponding to the outer encoder through the deinterleaver 500. Using this prior information La ′ (xi), a decoding operation is performed in decoder 600 corresponding to the outer encoder.

  From the second iteration onward, the output of the outer encoder-compatible decoder 600 is fed back to the PR channel MAP decoder 400 and used as prior information La (ci). In the PR channel MAP decoder 400 after the second iteration, as in the first iteration, only the likelihood information P (yi) for the read signal data yi in the read error area is adjacent in the read error area by the PR channel controller 1900. Set to substantially the same value for all branches between the states to be executed. However, in the decoding operation after the second iteration, calculation with the path weighted is performed by the above-described prior information La (ci).

  As described above, in the present invention, the sample values of the reproduction signal are not replaced, or the likelihood information of the output of the PR channel MAP decoder is not replaced, and all the adjacent states in the read error region are not replaced. MAP decoding is performed by replacing only the likelihood information P (yi) with respect to the reproduced signal data yi in the branch of FIG. 1 with substantially the same value, so that even when a read error occurs, the output of one decoder is used in advance of the other decoder. It can be used effectively as information.

[Description of read error detector]
Next, an embodiment of the read error detector 1800 is shown in FIG. The read error detector 1800 includes comparators 1801 and 1802, shift registers 1803 and 1804, and an OR circuit 1805. The comparators 1801 and 1802 have inputs a and b, and when a is greater than or equal to b, the output is at a high level, and when a <b, the output is at a low level. To do. The comparator 1801 compares the reproduction signal data yi with the positive threshold value B1 of the read error detection level, and determines whether or not it is a read error portion. The comparator 1802 compares the reproduction signal data yi with the negative threshold B2 of the read error detection level, and determines whether or not it is a read error portion. The outputs of the two comparators are input to N-stage shift registers 1803 and 1804 representing a read error position (BP), and all the outputs of the shift registers are input to an OR circuit 1805. The logical sum circuit 1805 outputs a high level read error gate signal (BG) if there is at least one high level value among the values input from each shift register.

  The threshold values B1 and B2 may be stored in a register (not shown) or may be input from the controller 1700.

  FIG. 10 shows the operation of the read error detector 1800. FIG. 10A shows reproduction signal data yi 2001 having a read error portion in the reproduction signal.

  As described with reference to FIG. 9, the read error detector 1800 has the reproduction signal data yi larger than the positive threshold value B1 of the read error detection level or the negative threshold value B2 of the read error detection level. If it is smaller than this, it is determined that the read error position BP.

  The present embodiment shown in FIG. 10 shows an example in which the number of stages N of the shift registers 1803 and 1804 in FIG. 9 is N = 4, for example. Note that the number N of stages of the shift registers 1803 and 1804 is an arbitrarily determined value based on the spot diameter of the beam, the size of the recording mark on the medium, and the like, and is usually smaller than the block length, which is an iterative decoding processing unit. Value.

  As shown in FIG. 10B, the output of the shift register 1803 becomes the high level 2002 to 2005 when the reproduction signal data yi is larger than the positive threshold value B1 of the read error detection level. On the other hand, as shown in FIG. 10C, the output of the shift register 1804 becomes a high level 2006 and 2007 when the reproduction signal data yi is smaller than the negative threshold B2 of the read error detection level. . When all the N-stage outputs of the shift registers 1803 and 1804 are input to the OR circuit 1805, as shown in FIG. 10D, the output of the OR circuit 1805 has a high level during the read error period. To get the signal 2010. In this way, the read error gate signal BG having a time period in the range affected by the read error can be created.

  In this way, by supplying the read error gate signal BG created by the read error detector 1800 to the PR channel controller 1900 of FIG. 8, the output of the reproduction signal probability acquisition means 406 can be controlled.

  In the above-described embodiment, the gate is generated only behind the detection pulse. However, as another embodiment, a method of spreading forward is also conceivable. For example, if the sampling value yi is delayed by N / 2 stages of the shift register, the read error gate signal BG is opened from N / 2 before the read error position BP. It is also possible to cope with a small influence of a reading error area due to generated scratches and dust. In other words, the sample value yi that is affected by the reading error in the portion before the reading error position BP can also be set as the compensation range.

  In addition, the above-described detection of the read error area may be configured such that detection is performed using a read error detection means every time the reproduction signal is iteratively processed in the iterative decoder. The detection result may be stored in the memory until completion, and the read error position may be read from the memory in the second and subsequent iterative decoding processes. In addition, as a method of storing in the memory, an array of block lengths, which is an iterative decoding processing unit, is prepared on the memory, and the detection result (reading normal or reading error is indicated at a position corresponding to the sampling value in the processing unit block. There is a method of storing binary information). In this method, the PR channel controller refers to the detection result stored in the position corresponding to the sampling value to be decoded in the iterative decoding process, and controls the operation of the reproduction signal probability acquisition means. As another method of storing in the memory, a method of storing only the position information where the reading error is detected can be considered. In this method, the PR channel controller compares the position of the sampling value to be decoded with the detected position on the memory in the iterative decoding process, and controls the operation of the reproduction signal probability acquisition means.

  As described above, the likelihood information P (yi) for the read signal data yi in the read error area is forcibly converted to substantially the same value for all branches between adjacent states, thereby selecting in the read error area All possible path probabilities are equal. Subsequent calculation of the γ value, α value, and β value proceeds based on the output value of the reproduction signal probability acquisition means controlled by the PR channel controller. This makes it possible to prevent the probability value of a path different from the original path from increasing when a read error occurs.

  In particular, in the present invention, the sample value of the reproduction signal is not replaced, or the output of the PR channel MAP decoder is not replaced, and the probability value with respect to the expected value of the reproduction signal in the read error region is determined as the read error region. MAP decoding is performed using a probability value obtained by replacing only the likelihood information P (yi) for the reproduced signal data yi with the same value so as to be equivalent among all branches of the read signal data. The output of the decoder can be effectively used as prior information of the other decoder.

  Accordingly, the iterative decoding scheme according to the present invention has a decoding capability even at a low SNR without propagating erroneous likelihood information in an iterative process even for a burst error exceeding the correction capability of the iterative decoder. High system can be obtained.

The figure which shows the basic composition of an iterative decoding system The figure which shows the structure of the recording / reproducing system of the optical disk using the conventional iterative decoding system The figure which shows the structure of the conventional PR channel MAP decoder. The figure which showed the relation between the data bit string on the recording medium and the PR channel reproduction signal (PR (1,1) system) A diagram showing an example of a trellis diagram The figure which showed the relation between the branch and PR channel reproduction signal (PR (1,1) system) The figure which shows the structure of the optical disk recording / reproducing system using the iterative decoding system by this invention The figure which shows the structure of PR channel MAP decoder by this invention The figure which shows the Example of the reading error detector of this invention The figure which shows description of operation | movement of the reading error detector of this invention

Claims (4)

A playback device for playing back data from a playback signal using iterative decoding,
A reading error detecting means for detecting a reading error area in the reproduction signal;
Reproduction signal probability acquisition means for acquiring a probability value for an expected value of the reproduction signal;
Control means for controlling the reproduction signal probability acquisition means so that the probability value for the expected value of the reproduction signal in the read error area is equal among all branches in the read error area based on the detection result of the detection means. When,
A reproduction apparatus comprising: an iterative decoding operation unit that performs an iterative decoding operation using a probability value obtained from the reproduction signal probability acquisition unit. The playback device according to claim 1,
A playback apparatus comprising: storage means for storing information on a read error position detected by the read error detection means until iterative decoding of a playback signal corresponding to the read error position is completed. The playback device according to claim 1,
A reproduction apparatus characterized in that the processing of the reproduction signal probability calculation means is delayed by a predetermined clock with respect to the detection result of the error detection means. A method of correcting a read error in the reproduction signal reproduced by a reproduction device that reproduces data from the reproduction signal using iterative decoding,
A read error detecting step for detecting a read error area in the reproduction signal;
A reproduction signal probability acquisition step of acquiring a probability value with respect to an expected value of the reproduction signal;
Based on the detection result of the detection step, the operation of the reproduction signal probability acquisition step is controlled so that the probability value with respect to the expected value of the reproduction signal in the reading error region is the same among all branches in the reading error region. Control steps;
An iterative decoding operation step for performing an iterative decoding operation using a probability value obtained from the reproduction signal probability acquisition step;
A reading error correction method, characterized by causing a playback device to have

Images