WO2010001502A1 - 復号化装置、垂直磁気記録再生装置、受信装置、および、復号化方法 - Google Patents
復号化装置、垂直磁気記録再生装置、受信装置、および、復号化方法 Download PDFInfo
- Publication number
- WO2010001502A1 WO2010001502A1 PCT/JP2008/073794 JP2008073794W WO2010001502A1 WO 2010001502 A1 WO2010001502 A1 WO 2010001502A1 JP 2008073794 W JP2008073794 W JP 2008073794W WO 2010001502 A1 WO2010001502 A1 WO 2010001502A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- decoding
- burst
- output
- filter
- error
- Prior art date
Links
Images
Classifications
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
- H03M13/00—Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes
- H03M13/03—Error detection or forward error correction by redundancy in data representation, i.e. code words containing more digits than the source words
- H03M13/05—Error detection or forward error correction by redundancy in data representation, i.e. code words containing more digits than the source words using block codes, i.e. a predetermined number of check bits joined to a predetermined number of information bits
- H03M13/11—Error detection or forward error correction by redundancy in data representation, i.e. code words containing more digits than the source words using block codes, i.e. a predetermined number of check bits joined to a predetermined number of information bits using multiple parity bits
- H03M13/1102—Codes on graphs and decoding on graphs, e.g. low-density parity check [LDPC] codes
- H03M13/1105—Decoding
- H03M13/1108—Hard decision decoding, e.g. bit flipping, modified or weighted bit flipping
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11B—INFORMATION STORAGE BASED ON RELATIVE MOVEMENT BETWEEN RECORD CARRIER AND TRANSDUCER
- G11B20/00—Signal processing not specific to the method of recording or reproducing; Circuits therefor
- G11B20/10—Digital recording or reproducing
- G11B20/10009—Improvement or modification of read or write signals
- G11B20/10046—Improvement or modification of read or write signals filtering or equalising, e.g. setting the tap weights of an FIR filter
- G11B20/10203—Improvement or modification of read or write signals filtering or equalising, e.g. setting the tap weights of an FIR filter baseline correction
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11B—INFORMATION STORAGE BASED ON RELATIVE MOVEMENT BETWEEN RECORD CARRIER AND TRANSDUCER
- G11B20/00—Signal processing not specific to the method of recording or reproducing; Circuits therefor
- G11B20/10—Digital recording or reproducing
- G11B20/18—Error detection or correction; Testing, e.g. of drop-outs
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11B—INFORMATION STORAGE BASED ON RELATIVE MOVEMENT BETWEEN RECORD CARRIER AND TRANSDUCER
- G11B20/00—Signal processing not specific to the method of recording or reproducing; Circuits therefor
- G11B20/10—Digital recording or reproducing
- G11B20/18—Error detection or correction; Testing, e.g. of drop-outs
- G11B20/1816—Testing
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11B—INFORMATION STORAGE BASED ON RELATIVE MOVEMENT BETWEEN RECORD CARRIER AND TRANSDUCER
- G11B20/00—Signal processing not specific to the method of recording or reproducing; Circuits therefor
- G11B20/10—Digital recording or reproducing
- G11B20/18—Error detection or correction; Testing, e.g. of drop-outs
- G11B20/1833—Error detection or correction; Testing, e.g. of drop-outs by adding special lists or symbols to the coded information
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
- H03M13/00—Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes
- H03M13/03—Error detection or forward error correction by redundancy in data representation, i.e. code words containing more digits than the source words
- H03M13/05—Error detection or forward error correction by redundancy in data representation, i.e. code words containing more digits than the source words using block codes, i.e. a predetermined number of check bits joined to a predetermined number of information bits
- H03M13/11—Error detection or forward error correction by redundancy in data representation, i.e. code words containing more digits than the source words using block codes, i.e. a predetermined number of check bits joined to a predetermined number of information bits using multiple parity bits
- H03M13/1102—Codes on graphs and decoding on graphs, e.g. low-density parity check [LDPC] codes
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
- H03M13/00—Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes
- H03M13/03—Error detection or forward error correction by redundancy in data representation, i.e. code words containing more digits than the source words
- H03M13/05—Error detection or forward error correction by redundancy in data representation, i.e. code words containing more digits than the source words using block codes, i.e. a predetermined number of check bits joined to a predetermined number of information bits
- H03M13/13—Linear codes
- H03M13/17—Burst error correction, e.g. error trapping, Fire codes
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11B—INFORMATION STORAGE BASED ON RELATIVE MOVEMENT BETWEEN RECORD CARRIER AND TRANSDUCER
- G11B20/00—Signal processing not specific to the method of recording or reproducing; Circuits therefor
- G11B20/10—Digital recording or reproducing
- G11B20/18—Error detection or correction; Testing, e.g. of drop-outs
- G11B20/1816—Testing
- G11B2020/1823—Testing wherein a flag is set when errors are detected or qualified
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11B—INFORMATION STORAGE BASED ON RELATIVE MOVEMENT BETWEEN RECORD CARRIER AND TRANSDUCER
- G11B20/00—Signal processing not specific to the method of recording or reproducing; Circuits therefor
- G11B20/10—Digital recording or reproducing
- G11B20/18—Error detection or correction; Testing, e.g. of drop-outs
- G11B20/1833—Error detection or correction; Testing, e.g. of drop-outs by adding special lists or symbols to the coded information
- G11B2020/185—Error detection or correction; Testing, e.g. of drop-outs by adding special lists or symbols to the coded information using an low density parity check [LDPC] code
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11B—INFORMATION STORAGE BASED ON RELATIVE MOVEMENT BETWEEN RECORD CARRIER AND TRANSDUCER
- G11B2220/00—Record carriers by type
- G11B2220/20—Disc-shaped record carriers
- G11B2220/25—Disc-shaped record carriers characterised in that the disc is based on a specific recording technology
- G11B2220/2508—Magnetic discs
- G11B2220/2516—Hard disks
Definitions
- the present invention relates to a decoding device, a perpendicular magnetic recording / reproducing device, a receiving device, and a decoding method, and in particular, a decoding device for detecting a burst error in a data signal encoded by a low density parity check code, and a perpendicular magnetic
- the present invention relates to a recording / reproducing apparatus, a receiving apparatus, and a decoding method.
- LDPC low density parity check
- the low density parity check (LDPC) code has a very good decoding performance against random errors, it is known that the decoding performance is extremely deteriorated against continuous (burst) errors (burst errors). Yes.
- iterative decoding is performed in the decoding process of the LDPC code, so that error data generated in the burst error interval is propagated (distributed), resulting in decoding performance (error correction). There is a problem that the performance is extremely lowered.
- a burst error is an error that continues over a long section of data, and due to its generation mechanism, for example, in a magnetic recording / reproducing system such as a hard disk drive, a thermal asperity ( there is a burst error such as pole erase that occurs due to overwriting from an adjacent track due to residual magnetism.
- the data storage device described in Patent Document 1 is an iterative decoder that performs iterative decoding processing on encoded data to which an RLL (Run-Length Limited) code is further added.
- the log likelihood ratio (LLR: Log Likelihood Ratio) information detects a burst error interval by determining whether or not it violates the RLL coding rule, and the log likelihood corresponding to the detected burst error interval It is described that adjustment is performed by attenuating the ratio.
- the recording / reproducing apparatus described in Patent Document 2 detects the occurrence of thermal asperity based on signal amplitude fluctuations, outputs an erasure flag signal, masks channel information over a period in which the erasure flag is on, performs iterative calculation, and performs ECC ( It is described that the error correcting code is not required.
- the decoding device described in Patent Document 3 determines the authenticity of each data in each block based on hard information based on the ECC code with respect to the encoded data to which the ECC code is further added. It describes that the likelihood of each piece of data of a block determined as “no error” is replaced with a maximum value.
- Non-Patent Documents 1 and 2 assuming a burst error due to a medium defect, a burst error interval is specified by detecting deterioration in reproduction amplitude, and the LLR of the burst error interval is set to 0. By performing such operations as this, error propagation in the iterative decoding process is suppressed.
- the conventional decoding apparatus can be widely applied to various burst errors such as medium loss, thermal asperity, and pole erase without adding a special code (redundant code) for detecting a burst error. There was a problem that there was no detection means.
- the present invention has been made in view of the above, and does not add a special code (redundant code) for detecting a burst error, and is resistant to a burst error that does not involve amplitude fluctuation, such as pole erase. It is an object of the present invention to provide a decoding device, a perpendicular magnetic recording / reproducing device, a receiving device, and a decoding method that can perform decoding.
- a special code redundant code
- a decoding apparatus is a decoding apparatus that performs a decoding process on an encoded data signal, and the above-described encoding encoded by a low density parity check code
- a burst detection means for outputting burst information by performing a parity check of the data signal is provided.
- a decoding device that performs a decoding process on an encoded data signal, and performing a parity check on the encoded data signal encoded by a low density parity check code, Since information is output and decoding processing is performed on the encoded data signal, a special code (redundant code) for detecting a burst error is not added, and even for a burst error without amplitude fluctuation. Resistant decoding can be performed. In other words, even a bit-flipping burst error where the burst position cannot be specified from the signal amplitude, etc., such as pole erase, can be detected. In addition, it is not necessary to perform decryption for that purpose, and a decrease in storage capacity and an increase in processing time can be suppressed.
- the decoding apparatus is characterized in that, in the decoding apparatus described above, the burst information includes information specifying a position and length of a burst error of the encoded data signal.
- the burst information since the burst information includes information specifying the position and length of the burst error of the encoded data signal, it corresponds to the position and length of the burst error based on the burst information in the iterative decoding process. It is possible to suppress propagation (dispersion) of error data in the burst error section.
- the burst detection means in the decoding device described above performs a parity check on the encoded data signal for each row of a parity check matrix, and For each column of the parity check matrix corresponding to each bit, the number of bits “1” in the row detected as an error by the parity check is added up, and the burst error of the encoded data signal according to the distribution of the added number
- the burst information is generated and output by specifying the position and the length.
- the parity check is performed on the encoded data signal for each row of the parity check matrix, and an error is detected by the parity check for each column of the parity check matrix corresponding to each bit of the encoded data signal. Since the number of bits “1” in a row is added and burst information is generated and output by specifying the position and length of the burst error of the encoded data signal according to the distribution of the added number, a low-density parity check code The burst error interval can be identified appropriately based on the parity check matrix.
- the decoding device of the present invention is the decoding device according to the above, wherein the encoding corresponding to the position and the length of the burst error based on the burst information output by the burst detector. It further comprises decoding means for repeatedly executing decoding processing while suppressing an increase in the likelihood of bits of the data signal.
- the burst detector based on the burst information output by the burst detector, the increase in the likelihood of the bit of the encoded data signal corresponding to the position and length of the burst error is suppressed, and the iterative decoding process is performed. As a result, error propagation from burst errors can be suppressed, and decoding performance can be further improved.
- the decoding apparatus is the decoding apparatus described above, wherein the burst detection means performs a hard decision by a hard decision means that makes a hard decision on the decoding result output from the decoding means, and the hard decision means.
- the decoded result is subjected to a parity check based on a parity check matrix to generate a parity flag, and a moving average filter is applied to the parity flag generated by the parity flag generating means.
- Filter output means for performing filter output, and determining the position and length of the burst error by performing threshold determination on the filter output outputted by the filter output means, and decoding the burst information And burst information output means for outputting to.
- the decoding result output from the decoding unit is hard-determined, and the hard-decision-decoding result is subjected to parity check based on the parity check matrix to generate the parity flag, and the generated parity Applying the moving average filter to the flag and performing filter output, and by determining the threshold value for the output filter output, the burst error position and length are identified and burst information is output to the decoding means. Can be appropriately identified, and decoding resistant to burst errors can be performed.
- the decoding apparatus of the present invention is characterized in that, in the decoding apparatus described above, the filter output means applies the moving average filter to the parity flag in multiple stages.
- the moving average filter is applied to the parity flag in multiple stages, it is possible to detect a short bit flipping signal burst, and to perform decoding that is resistant to a short burst error.
- the decoding apparatus is the decoding apparatus described above, wherein the filter output means is the length of the burst error specified by the threshold value judgment of the filter output performed by the burst information output means.
- the filter output means is the length of the burst error specified by the threshold value judgment of the filter output performed by the burst information output means.
- the moving average filter that outputs a moving average of a section longer than the applied moving average filter is provided. Since the filter output is performed again after being applied, the burst error can be detected with high accuracy using the moving average filter corresponding to the length of the burst error.
- the decoding apparatus is the decoding apparatus described above, wherein the decoding means performs a posteriori probability decoding process on the encoded data signal, and includes the decoding result including the likelihood. And an APP decoder that suppresses the increase in likelihood output from the APP decoder based on the burst information output from the burst detector. An SP (sum-product) decoder that executes an iterative decoding process on the decoding result output from, and outputs the decoded decoding result including the likelihood as an input of the APP decoder. It is characterized by that.
- the a posteriori probability decoding process is performed on the encoded data signal and the decoding result including the likelihood is output, and the burst information output from the burst detector is used. , Suppressing an increase in the likelihood output from the APP decoder, executing an iterative decoding process on the decoding result output from the APP decoder, and obtaining the decoded result including the updated likelihood in the APP decoder. Since the SP decoder that is output as an input is provided, highly accurate decoding that is resistant to burst errors can be performed by an appropriate combination of an inner decoder and an outer decoder.
- the decoding apparatus is the decoding apparatus described above, wherein the SP decoder suppresses the increase in the likelihood output from the APP decoder based on the burst information by weighting. It is characterized by.
- the increase in the likelihood output from the APP decoder is suppressed by weighting based on the burst information, it is possible to suppress error propagation from the burst error by appropriate weighting, and to improve the decoding performance. Can be further improved.
- the encoded data signal is further encoded by an RS (Reed-Solomon) code in the decoding device described above, and the decoding device includes the decoding means.
- An RS decoder for decoding the RS code with respect to the output is further provided.
- the configuration further comprising an RS decoder for decoding the RS code with respect to the output of the decoding means can withstand burst errors even when a short bit flipping signal burst occurs. Decoding can be performed.
- the decoding apparatus of the present invention is characterized in that, in the decoding apparatus described above, the burst error is caused by simultaneous occurrence of a medium defect and a bit-flipping signal burst.
- a burst error can be detected even when a medium defect and a bit-flipping signal burst occur simultaneously.
- the present invention also relates to a perpendicular magnetic recording / reproducing apparatus, which is a perpendicular magnetic recording / reproducing apparatus provided with the decoding device described above.
- a perpendicular magnetic recording / reproducing apparatus that is resistant to a burst error in a perpendicular magnetic recording / reproducing apparatus with a bit-flipping signal burst that is difficult to detect from a reproduced waveform, an APP decoder output, or the like. Can do.
- the present invention also relates to a receiving device, and is characterized in that the receiving device includes the decoding device described above.
- the present invention it is possible to provide a receiving device that is resistant to burst errors in communication via a binary input additive white Gaussian channel or the like.
- the present invention also relates to a decoding method, which is a decoding method for performing a decoding process on an encoded data signal, wherein the encoded data signal encoded by a low density parity check code is encoded.
- a decoding method for performing a decoding process on an encoded data signal, and by performing a parity check of an encoded data signal encoded by a low density parity check code, burst information Is output and the decoding process is performed on the encoded data signal, so it is resistant to burst errors without amplitude fluctuations without adding a special code (redundant code) for burst error detection.
- Decoding can be performed. That is, even a burst error in which the burst position cannot be specified from the signal amplitude or the like like pole erase can be detected. For burst detection, encoding other than the low-density parity check code and decoding for it Therefore, it is not necessary to reduce the storage capacity and increase the processing time.
- the decoding method of the present invention is characterized in that, in the decoding method described above, the burst information includes information specifying a position and length of a burst error of the encoded data signal.
- the burst information since the burst information includes information specifying the position and length of the burst error of the encoded data signal, it corresponds to the position and length of the burst error based on the burst information in the iterative decoding process. It is possible to suppress propagation (dispersion) of error data in the burst error section.
- the decoding method of the present invention is the decoding method described above, wherein the burst detection step performs a parity check on the encoded data signal for each row of a parity check matrix, For each column of the parity check matrix corresponding to each bit, the number of bits “1” in the row detected as an error by the parity check is added up, and the burst error of the encoded data signal according to the distribution of the added number
- the burst information is generated and output by specifying the position and the length.
- the parity check is performed on the encoded data signal for each row of the parity check matrix, and an error is detected by the parity check for each column of the parity check matrix corresponding to each bit of the encoded data signal. Since the number of bits “1” in a row is added and burst information is generated and output by specifying the position and length of the burst error of the encoded data signal according to the distribution of the added number, a low-density parity check code The burst error interval can be identified appropriately based on the parity check matrix.
- the decoding method of the present invention is the decoding method described above, wherein the code corresponding to the position and the length of the burst error is based on the burst information output in the burst detection step. And a decoding step of repeatedly performing a decoding process while suppressing an increase in the likelihood of the bit of the coded data signal.
- iterative decoding processing is performed while suppressing an increase in the likelihood of the bit of the encoded data signal corresponding to the position and length of the burst error. Therefore, error propagation from a burst error can be suppressed, and decoding performance can be further improved.
- the decoding method of the present invention is the decoding method described above, wherein the burst detection step includes a hard decision step that makes a hard decision on the decoding result output in the decoding step, and a hard decision step.
- a parity check is performed based on a parity check matrix to generate a parity flag for the hard-decision decoded result, and the parity flag generated in the parity flag generation step includes a moving average.
- a filter output step for applying a filter to perform filter output; and determining the position and length of the burst error by performing threshold determination on the filter output output in the filter output step, and determining the burst information Burst information output step for outputting to the decoding step, And wherein the Mukoto.
- the decoding result output from the decoding unit is hard-determined, and the hard-decision-decoding result is subjected to parity check based on the parity check matrix to generate the parity flag, and the generated parity A filter is applied by applying a moving average filter to the flag, and the burst error is output to the decoding step by identifying the position and length of the burst error by performing threshold judgment on the output filter output. Can be appropriately identified, and decoding resistant to burst errors can be performed.
- the decoding method of the present invention is characterized in that, in the decoding method described above, the filter output step applies the moving average filter to the parity flag in multiple stages.
- the moving average filter is applied to the parity flag in multiple stages, it is possible to detect a short bit flipping signal burst, and to perform decoding that is resistant to a short burst error.
- the decoding method of the present invention is the decoding method described above, wherein the filter output step includes the burst error specified by performing the filter output threshold determination in the burst information output step.
- the filter output is performed again by applying the moving average filter that outputs a moving average of a section longer than the applied moving average filter.
- the moving average filter that outputs a moving average of a section longer than the applied moving average filter is provided. Since the filter output is performed again after being applied, the burst error can be detected with high accuracy using the moving average filter corresponding to the length of the burst error.
- the decoding method according to the present invention is the decoding method described above, wherein the decoding step performs a posteriori probability decoding process on the encoded data signal, and includes the decoding result including the likelihood.
- the APP (a posteriori probability) decoding step that outputs, and the burst information output in the burst detection step, the increase in the likelihood output in the APP decoding step is suppressed, and the APP An SP (sum-product) decoding step that performs an iterative decoding process on the decoding result output in the decoding step and outputs the decoded decoding result including the likelihood as an input of the APP decoding step; It is characterized by including.
- the a posteriori probability decoding process is performed on the encoded data signal, the decoding result including the likelihood is output, and output based on the burst information output in the burst detection step.
- the increase in likelihood is suppressed, the iterative decoding process is performed on the output decoding result, and the decoding result including the updated likelihood is output as the input of the posterior probability decoding process.
- the combination of the decoder process and the outer decoder process makes it possible to perform highly accurate decoding that is resistant to burst errors.
- the decoding method of the present invention is the decoding method described above, wherein the SP decoding step suppresses the increase in the likelihood output in the APP decoding step by weighting based on the burst information. It is characterized by this.
- the increase in the likelihood output in the posterior probability decoding process is suppressed by weighting, so that the propagation of errors from the burst error can be suppressed by appropriate weighting, Decoding performance can be further improved.
- the decoding method of the present invention is the decoding method described above, wherein the encoded data signal is further encoded by an RS (Reed-Solomon) code, and the output in the decoding step is An RS decoding step for decoding the RS code is further included.
- RS Random-Solomon
- the configuration further comprising an RS decoder for decoding the RS code with respect to the output of the decoding step can withstand burst errors even when a short bit flipping signal burst occurs. Decoding can be performed.
- the decoding method of the present invention is characterized in that, in the decoding method described above, the burst error is caused by simultaneous occurrence of a medium defect and a bit-flipping signal burst.
- a burst error can be detected even when a medium defect and a bit-flipping signal burst occur simultaneously.
- the present invention without adding a special code (redundant code) for detecting a burst error, it is possible to perform decoding that is resistant to a burst error that does not involve amplitude fluctuations, such as pole erase. There is an effect.
- FIG. 1 is a block diagram showing an example of the configuration of a decoding device 60 to which the present invention is applied.
- FIG. 2 is a flowchart showing an example of basic processing of the decoding device 60.
- FIG. 3 is a flowchart showing an example of burst detection processing by the burst detection means 61.
- FIG. 4 is a block diagram of an LDPC encoding / iterative decoding method including a burst detector using a parity check matrix of an LDPC code.
- FIG. 1 is a block diagram showing an example of the configuration of a decoding device 60 to which the present invention is applied.
- FIG. 2 is a flowchart showing an example of basic processing of the decoding device 60.
- FIG. 3 is a flowchart showing an example of burst detection processing by the
- FIG. 7 is a diagram showing error rate characteristics of the LDPC-PCBD system when the burst detector is not operated.
- FIG. 8 is a block diagram of a burst detector 61 ′ proposed in this embodiment.
- FIG. 9 is a diagram showing the flow of burst detection proposed in this embodiment.
- FIG. 10 is a diagram illustrating an example of a parity check.
- FIG. 11 is a diagram showing parity flags actually detected when there is no burst.
- FIG. 12 is a diagram showing parity flags actually detected when there is a burst.
- FIG. 13 is a diagram showing the filter output per sector.
- FIG. 11 is a diagram showing parity flags actually detected when there is no burst.
- FIG. 12 is a diagram showing parity flags actually detected when there is a burst.
- FIG. 14 is a diagram showing the relationship between the detected average burst length and TH flip .
- FIG. 15 is a diagram illustrating a frequency distribution of detected burst lengths.
- FIG. 16 is a diagram illustrating a relationship between a required SN ratio and L flip for decoding predetermined data without error.
- FIG. 17 is a diagram showing a recording / reproducing system block diagram of an LDPC encoding / iterative decoding method in which RS codes are connected.
- FIG. 18 is a diagram showing the relationship between the required S / N ratio and the burst length L flip for decoding predetermined data without error, obtained by computer simulation.
- FIG. 19 is a diagram showing the relationship between the required SN ratio and the medium defect length for decoding predetermined data without error.
- FIG. 21 is a diagram showing the relationship between the detected average burst length and TH flip .
- FIG. 24 is a diagram showing the relationship between the required SN ratio and L flip for decoding predetermined data without error when three types of MA filters are individually applied.
- FIG. 25 is a flowchart illustrating an example of processing of the filter 613 ′.
- FIG. 26 is a diagram illustrating error rate characteristics of the LDPC encoding / iterative decoding scheme.
- FIG. 27 is a diagram illustrating a relationship between a required SN ratio and L flip for decoding predetermined data without error.
- FIG. 28 is a block diagram of an LDPC encoding / iterative decoding system including a burst detector using a parity check matrix of an LDPC code applied to a binary input additive white Gaussian channel.
- FIG. 29 is a diagram showing BER characteristics in a binary input AWGN communication path.
- FIG. 1 is a block diagram showing an example of the configuration of a decoding device 60 to which the present invention is applied, and conceptually shows only the portion related to the present invention.
- the decoding apparatus 60 schematically performs a decoding process on a data signal encoded by a low density parity check code (hereinafter referred to as “LDPC code”). And a burst unit for outputting burst information specifying a burst error section (for example, the position and length of a burst error) of the encoded data signal by performing a parity check based on the encoded data signal.
- the detecting means 61 is connected.
- a decoding unit 65 is a decoding unit that performs a decoding process on a data signal encoded by a low density parity check code. Specifically, the decoding means 65 is based on the burst information output from the burst detection means 61, and the likelihood of the bit of the encoded data signal corresponding to the burst error section (for example, the position and length of the burst error). It suppresses the increase in the number of times, and repeats the decoding process.
- the “likelihood” is the reliability of each bit.
- the probability P (x 1
- y) / P (x 0
- the decoding unit 65 performs a posteriori probability (APP) decoding process on the encoded data signal to perform decoding including likelihood.
- APP posteriori probability
- an output of the APP decoder 62 that suppresses an increase in the likelihood output from the APP decoder 62 based on the burst information output from the burst detector 61 and the APP decoder 62 that outputs the conversion result.
- An SP decoder 63 that performs iterative decoding processing on the decoding result and outputs the decoding result including the updated likelihood as an input of the APP decoder 62 may be provided. When this configuration is adopted, the SP decoder 63 may suppress the increase in the likelihood output from the APP decoder 62 based on the burst information by weighting.
- the burst detection means 61 performs LDPC parity check on the basis of the encoded data signal output from the decoding means 65 (specifically, the APP decoder 62), thereby obtaining the encoded data signal.
- This is burst detection means for outputting burst information specifying a burst error section to decoding means 65 (specifically, SP decoder 63).
- the burst detection means 61 performs a parity check on the encoded data signal for each row of the parity check matrix, and causes an error in the parity check for each column of the parity check matrix corresponding to each bit of the encoded data signal. Burst information is generated and output by adding the number of bits “1” of the detected rows and specifying the burst error interval of the encoded data signal according to the distribution of the added number.
- the burst detection unit 61 performs a hard decision by a hard decision unit 611 and a hard decision unit 611 that makes a hard decision on the decoding result output from the APP decoder 62.
- the decoded result is subjected to parity check based on a parity check matrix to generate a parity flag, and a moving average filter is applied to the parity flag generated by the parity flag generation unit 612.
- Information output means 614 Information output means 614.
- the filter output means 613 may apply a moving average filter to the parity flag in multiple stages.
- you may provide the RS decoder 66 which decodes an RS code with respect to the output of the decoding means 65.
- FIG. The above is the configuration of the decoding device 60 in the present embodiment.
- FIG. 2 is a flowchart showing an example of basic processing of the decoding device 60.
- FIG. 3 is a flowchart showing an example of burst detection processing by the burst detection means 61.
- the decoding device 60 performs a parity check based on the encoded data signal by the LDPC code by the processing of the burst detection means 61 (step SA-1).
- the decoding device 60 outputs burst information specifying the burst error section (for example, the position and length of the burst error) of the encoded data signal by the processing of the burst detection means 61 (step SA-2).
- the decoding device 60 repeatedly performs decoding processing while suppressing an increase in the likelihood of bits of the encoded data signal corresponding to the burst error interval based on the output burst information by the processing of the decoding unit 65. Is executed (step SA-3).
- the burst detection means 61 may perform the burst detection processing shown in FIG. 3 in steps SA-1 and SA2.
- the burst detection unit 61 performs a hard decision on the decoding result (APP output sequence) output from the APP decoder 62 by the processing of the hard decision unit 611 using 0 as a threshold (step). SB-1).
- the burst detection means 61 performs a parity check for each row on the basis of the hard matrix of the APP output sequence based on the LDPC code check matrix by the processing of the parity flag generation means 612 (step SB-2).
- the burst detection means 61 generates a parity flag from the parity check result by the processing of the parity flag generation means 612 (step SB-3). Specifically, the burst detection unit 61 adds the number of bits “1” of the row detected as an error in the parity check for each column of the parity check matrix and generates a parity flag.
- the burst detection unit 61 shapes the generated parity flag through the moving average filter by the processing of the filter output unit 613, and performs filter output (step SB-4).
- the filter output means 613 applies the moving average filter in multiple stages (for example, two stages in which the first stage is the filter length L MA1 and the second stage is the filter length L MA2 (L MA2 > L MA1 )). May be.
- the filter output means 613 applies a moving average filter that outputs a moving average of a section longer than the applied moving average filter. Then, the filter output may be performed again with respect to the burst detecting means 61.
- BL max is 120 or more and less than 300
- the burst detection means 61 creates a burst information specifying the burst position and length by determining the filter output as a threshold by the processing of the burst information output means 614 and outputs it to the SP decoder 63 (step SB). -5).
- the SP decoder 63 that has received the burst information from the burst detecting means 61 repeats the SP decoding process a predetermined number of times in consideration of the burst information. For example, the SP decoder 63 repeatedly performs decoding processing while suppressing an increase in the likelihood of bits corresponding to the position and length of the burst error based on the burst information.
- the SP decoder 63 may perform the suppression of the increase in likelihood by weighting.
- the burst detecting means 61 described above may be configured to detect the burst only for the first time with respect to the output of the APP decoder 62 and output the burst information to the SP decoder 63.
- the burst information may be treated as valid in the subsequent iterative decoding process.
- the SP decoder 63 returns the decoding result to the APP decoder 62, and controls so that the decoding process is repeatedly performed a predetermined number of times between the SP decoder 63 and the APP decoder 61.
- the decoding device 60 makes a hard decision on the log likelihood ratio output from the SP decoder 63 to obtain an output data sequence. This completes the description of the processing of the decoding device 60 in the present embodiment.
- Example 1-1 a new burst detection method using a parity check matrix of an LDPC code has been proposed.
- a new burst detection for a bit-flipping burst that cannot be detected by a conventional burst detector using a reproduced waveform, an equalizer output, and an APP decoder output will be considered.
- the LDPC encoding / iterative decoding method provided with the burst detector proposed in the embodiment 1-1 and the conventional RS encoding with erasure error correction are provided. Performance comparison with the method was performed by computer simulation. As a result, it is clear that iterative decoding using the burst detector proposed in the present application shows better characteristics than the conventional RS encoding method and can allow a bit-flipping burst of about 1500 bits. It was.
- HDD hard disk drive
- LDPC Low-density parity check
- the LDPC code is a sum-product (SP) algorithm ([2] RR Kschishang, BJ Frey, and HA Loeliger, “Factor graphs and the sum-product algorithm E Tr. It is known that decoding characteristics very close to the Shannon limit can be obtained by combining with iterative decoding using Theory, vol.47, no.2, pp.498-519, Feb. 2001. [[3] DJ C. Mackay and RM Neal, “Near Shannon limit performance of low-density parity-check” odes, "Electron. Lett., vol. 32, pp. 1645-1646, Aug. 1996., [4] S. Y. Chung, GD Forney, Jr., T.J. Richardson, and R. Urbanke, “On the design of low-density parity-check codes within 0.0045 dB of the Shannon limit,” IEEE Commun. Lett., 2. vol. ).
- SP sum-product
- FIG. 4 is a block diagram of an LDPC encoding / iterative decoding method including a burst detector using a parity check matrix of an LDPC code.
- this method is referred to as an “LDPC-PCBD method”.
- the upper stage (a) and the lower stage (b) in FIG. 4 show an encoding block and a decoding block, respectively.
- the encoding block includes an RLL (run-length-limited) encoder 30, an LDPC encoder 31, a burst generator 32, a magnetic recording / reproducing system (PR1 channel 401), and an equalizer 42. Consists of
- the input data sequence ⁇ a k ′′ ⁇ is 128/130 (0, 16/8) encoded by the RLL encoder 30 ([8] Hidetoshi Saito, Toshihiko Iga, Masahiro Shirakawa, Yoshihiro Okamoto, Hisashi Osawa, “ The structure of a highly efficient run length constraint code and its evaluation, “Science Theory (C), vol. J86-C, no. 8, pp. 952-961, May 2003.), becomes ⁇ b k ′ ⁇ . However, in this embodiment, a long sector format of 4096 bytes / sector is assumed.
- the LDPC encoder 31 performs LDPC encoding on a sector basis to obtain ⁇ c k ⁇ .
- 24-byte error correction [9] T. Morita, Y. Sato, and T. T. using a GF (2 10 ) Galois field used in the current 512-byte / sector HDD is used.
- Sugawara “ECC-Less LDPC coding for magnetic recording channels,” IEEE Trans.Magn., Vol.38, no.5, pp.2304-2306, Sep.2002.
- the row weight and the column weight are set to 22 and 3, respectively.
- bit flipping signal burst is assumed as an error that cannot be detected from a reproduced waveform or the like. In the computer simulation, this is realized by inverting ⁇ c k ⁇ over the length L flip normalized by the user bit interval T b .
- a recording sequence ⁇ d k ⁇ including a bit flipping error only once for each sector is NRZ-recorded on the perpendicular magnetic recording medium.
- the waveform of the following equation (1) is assumed as an isolated reproduction waveform with respect to the stepped recording waveform ([10] Yoshihiro Okamoto, Mitsuteru Sato, Hidetoshi Saito, Hisashi Osawa, Hiroaki Muraoka, Yoshihisa Nakamura, A study of 3/4 MTR coded PRML system in perpendicular recording system using film media, “Science Technical Report, MR2000-8, June 2000.”.
- A the saturation level of h (t)
- T 50 the time required for h (t) to change from ⁇ A / 2 to A / 2.
- jitter medium noise generated when the magnetization transition point changes to white Gaussian and white-Gaussian noise (AWGN: additive white Gaussian noise) having an average value of 0 as system noise.
- AWGN additive white Gaussian noise
- the equalizer 42 includes a sixth-order Butterworth low-pass filter having a normalized cutoff frequency x h normalized by f b and a transversal filter having a tap number N t , and the characteristics of the equalizer 42 are recorded on the recording head.
- PR1 characteristics [11] ER Kretzmer, “Generalization of a binary for data communication,” IEEE Trans. Commun. Technol., Vol. Feb. 1966.).
- the (b) decoding block functioning as a decoding device has an APP decoder 62 ([12] Y. Nakamura, Y. Okamoto, H. Osawawa) that uses an AR (autogressive) channel model as a decoder input estimator. , H. Saito, H. Muraoka and Y. Nakamura, “A study of turbo decoding with embedding AR channel, perl. E.n. 2003.), SP decoder 63, burst detector 61 'functioning as burst detection means, hard decision device 4, composed of RLL decoder 67.
- the iterative decoding apparatus 601 functioning as a decoding unit that executes iterative decoding processing
- the APP decoder 62 performs decoding for inner coding, and a log likelihood ratio (LLR) ⁇ L ( ⁇ c k ) ⁇ (“ ⁇ ” is originally written on the next alphabet (hereinafter the same)).
- LLR log likelihood ratio
- the range L c taking into account the correlation of the decoder input noise in the AR channel model is 5 ([12] Y. Nakamura, Y. Okamoto, H. Osawa, H. Saito, H. Murakaka and Y. Nakamura).
- ⁇ L ( ⁇ c k ) ⁇ is input to the SP decoder 63 and the burst detector 61 ′.
- the burst detector 61 ′ detects the position and length of the burst from ⁇ L ( ⁇ c k ) ⁇ by using the parity check matrix of the LDPC code, and transmits it to the SP decoder 63 as burst information.
- the SP decoder 63 repeats SP decoding taking the burst information into consideration a predetermined number of times. Further, the SP decoder output is returned to the APP decoder 62 to perform iterative decoding.
- the number of iterations in the SP decoder 63 is i sp
- the number of iterations returned from the SP decoder 63 to the APP decoder 62 is i in . In this embodiment, the number of repetitions is 5 for both i sp and i in .
- the MAX-Log-MAP algorithm [13] P. Robertson, E. Villebrun, P.
- Example 1-1 the bit data rate (BER) characteristic is obtained by comparing the input data sequence and the output data sequence, and the performance evaluation of the LDPC encoding / iterative decoding scheme is performed. .
- FIG. 5 and 6 are diagrams showing the output of the APP decoder for one sector.
- K 1.5
- R J 80%
- SNR 21.5 dB
- x h 0.4
- L defect 1000 normalized by T b by media defect occurs
- the ⁇ (black circle) and x (cross) marks in the figure indicate the correctly decoded LLR and the erroneously decoded LLR, respectively.
- the number of erroneously decoded LLRs (Number of) Error). As a result, it can be seen that the number of erroneously decoded LLRs is not greatly different although the types of bursts are different.
- the detected burst flag is indicated by a solid line in the figure, and the burst signal degradation caused by a medium defect indicates that the LLR that is the output of the APP decoder at the location where the medium defect occurs is shown in FIG. It can be seen that the burst position can be almost detected by the threshold judgment, but the bit flipping signal bar in FIG. When the burst occurs, the APP decoder output does not drop at the burst occurrence location, and it can be seen that no burst is detected by determining the threshold value of the APP decoder output, as shown in FIG. Even when a bit-flipping signal burst occurs, the LLR value is the same level as the LLR at the position where there is no burst. It is expected that the expected effect of iterative decoding cannot be obtained.
- FIG. 7 is a diagram showing error rate characteristics of the LDPC-PCBD system when the burst detector is not operated.
- K 1.5
- R J 80%
- x h 0.4
- N t are optimum values N topt
- ⁇ white circle
- ⁇ white triangle
- ⁇ white square
- a burst detector using a parity check matrix of an LDPC code is proposed for a burst that cannot be detected by a conventional burst detector.
- 8 and 9 are a block diagram of the burst detector 61 ′ proposed in the present embodiment and a diagram showing a flow of burst detection, respectively. However, it is assumed that a signal burst occurs at a location represented by L flip in FIG.
- the burst detector 61 ′ proposed in this embodiment includes a hard decision unit 611 ′ functioning as a hard decision unit, a parity flag generator 612 ′ functioning as a parity flag generation unit, a filter 613 ′ functioning as a filter output unit, and And a burst information output unit 614 ′ functioning as burst information output means.
- the hard decision unit 611 ′ makes a hard decision on the APP decoder output sequence ⁇ L ( ⁇ c k ) ⁇ with 0 as a threshold value.
- the parity flag generator 612 ′ performs a parity check for each row on the basis of the hard decision result of ⁇ L ( ⁇ c k ) ⁇ based on the parity check matrix of the LDPC code.
- the parity flag generator 612 ′ creates a parity flag from the parity check result. Specifically, the parity flag generator 612 ′ adds up the number of bits “1” of the row detected as an error in the parity check for each column of the parity check matrix to form a parity flag. Accordingly, the parity flag in the case of an LDPC code having a column weight of 3 has values of “0” to “3” as shown in FIG.
- the filter 613 'shapes the parity flag through the filter. This filter facilitates burst detection by increasing the value of the signal burst position as shown in FIG. 9 by calculating the sum or moving average in a predetermined interval.
- FIG. 10 is a diagram illustrating an example of a parity check.
- the LDPC code used in this example is a parity check matrix having a code length of 16, a row weight of 4, and a column weight of 3. Each partial matrix of the LDPC code is surrounded by a solid line, and a blank portion indicates “0”. Also, a burst error has occurred in the section marked L flip . That is, the bit value of the bit flipping state is intentionally generated as the bit value of the detected data sequence by inverting the bit value of the recording data sequence in the L flip section.
- the parity flag generator 612 ′ performs a parity check on the hard decision result (detected data series) of the APP decoder output for each row to generate a parity flag. That is, it is checked whether or not the value of the check data sequence corresponding to “1” in the first row of the parity check matrix is “1”, and the number of parity check matrix values that are “1” is counted (first number). The first line is “2”). In this embodiment, since even parity is normal, the parity result is recorded on the first line as normal ( ⁇ in the figure).
- the parity flag generator 612 ′ executes this for all the rows (1st to 12th rows) of the parity check matrix, and determines that the parity check result is abnormal ( ⁇ in the figure) as shown in FIG. Check the line “1” (mark “1” with a circle). Then, the parity flag generator 612 ′ sets the parity flag by adding together the number of checked “1” s (“1” with a circle) in the same column of the parity check matrix. As shown in FIG. 10, in the LDPC code having a column weight of 3, when an error is detected in the same column of all the sub-matrices, the parity flag is “3”, so the parity flag is “0” to “3”. Distributed by value.
- FIG. 11 and FIG. 12 are diagrams showing parity flags actually detected.
- the parameters in the computer simulation are the same as those in FIGS.
- level 3 the maximum value in the LDPC code with column weight 3
- a section in which the value of the parity flag is continuously high can be estimated as a burst error section candidate.
- a moving average filter that is weighted when the parity flag level is 3 is applied.
- the filter 613 ′ has a two-stage configuration that takes a moving average between the front and rear 100 channel bits and again takes a moving average between the front and rear 200 channel bits.
- FIG. 13 is a diagram showing the filter output per sector. However, the parameters in the computer simulation are the same as those in FIGS. A dotted line in FIG. 13 indicates a section in which a bit flipping signal burst occurs. From the figure, it can be confirmed that the level of the filter output in this section is high. Therefore, it can be expected that burst detection is possible by threshold determination.
- FIG. 14 is a diagram showing the relationship between the detected average burst length and TH flip .
- TH flip is a filter output level threshold value for detecting a burst generated by bit flipping.
- the parameters in the computer simulation are the same as those in FIGS.
- FIG. 15 is a diagram illustrating a frequency distribution of detected burst lengths.
- both ends of the detected burst flag are expanded by 100 bits to form a burst error period, and burst information is output.
- the SP decoder 63 Attenuates the LLR value corresponding to the burst error interval based on the burst information, and performs iterative decoding until the set number of repetitions isp is reached.
- the burst information is detected only for the first time with respect to the output of the APP decoder, and is valid in the subsequent iterative decoding.
- the inventors of the present application have shown that error propagation caused by the burst portion can be suppressed by controlling the LLR value in the SP decoder 63 and performing iterative decoding that suppresses the increase in LLR ([7 ] Yasuaki Nakamura, Mitsuhiro Nishimura, Yoshihiro Okamoto, Hisashi Osawa, Hiroaki Muraoka, Yoshihisa Nakamura, "Examination of LDPC code correction capability in perpendicular magnetic recording and reproducing system with burst," IEICE Tech. Report, MR2007-23, Oct. 2007. .). Also in the present embodiment, an excessive increase in the LLR is suppressed by multiplying the weight coefficient W La in the repetition in the SP decoder 63.
- FIG. 16 is a diagram illustrating a relationship between a required SN ratio and L flip for decoding predetermined data without error.
- K 1.5
- R J 80%
- x h 0.4
- N t N top
- the total number of user data is 10 M bits.
- the circles (white circles), triangles (white triangles), and squares (white squares) in the figure indicate amplitude fluctuations in the LDPC-PCBD system equipped with the burst detector of the system proposed in this embodiment as described above.
- LDPC-BD system with conventional burst detector by detection RLL-RS-EC system with erasure error correction by adding RS code for burst error detection ([7] Yasuaki Nakamura, Mitsuhiro Nishimura , Yoshihiro Okamoto, Hisashi Osawa, Hiroaki Muraoka, Yoshihisa Nakamura, “Examination of LDPC code correction capability in perpendicular magnetic recording / reproducing system with burst,” IEICE Technical Report, MR2007-23, Oct.2007.) Show.
- the RS code in the RLL-RS-EC system is a code capable of 190-byte error correction using the GF (2 12 ) Galois field (the same document).
- a new burst detector using a parity check matrix of an LDPC code has been proposed.
- an LDPC encoding / iterative decoding system equipped with a burst detector proposed in this embodiment
- a conventional burst using amplitude fluctuations A performance comparison between an LDPC encoding / iterative decoding scheme (LDPC-BD scheme) equipped with a detector and an RS encoded erasure error correction scheme (RLL-RS-EC scheme) was performed by computer simulation.
- the LDPC encoding and iterative decoding method (LDPC-PCBD method) using the burst detector proposed in this embodiment can tolerate bit-flipping signal bursts with a bit flipping length L flip of up to about 1500.
- Example 1-2 In Example 1-1, it was shown that burst detection using a parity check matrix of an LDPC code is effective for a bit-flipping signal burst that is difficult to detect depending on a reproduction waveform, an APP decoder output, or the like. .
- the burst detection method using the check matrix creates burst information using a moving average filter, there is room for improvement in detecting a short burst of 200 channel bits or less. Therefore, in the present embodiment 1-2, a system capable of decoding even a short signal burst without error by connecting an RS code to the LDPC coding / iterative decoding method (LDPC-PCBD method) is proposed. The performance is evaluated by applying it to a perpendicular magnetic recording / reproducing system with a bit-flipping signal burst.
- LDPC-PCBD method a system capable of decoding even a short signal burst without error by connecting an RS code to the LDPC coding / iterative decoding
- FIG. 17 is a diagram showing a recording / reproducing system block diagram of an LDPC encoding / iterative decoding method in which RS codes are concatenated.
- the RLL code is a 128/130 (0, 16/8) RLL code
- the RS code is a 20-code symbol error-correcting RS code using GF (2 12 ) that can correct an error of 200 channel bits or more.
- a regular LDPC code having an information length of 33280, a row weight of 24, and a column weight of 3 is used, and the overall coding rate is 0.84.
- the bit flipping-like signal burst length normalized by the user bit interval T b L flip occurs only once for each sector (see Example 1-1).
- the isolated reproduction waveform is the waveform of the above-described formula (1), and white Gaussian noise and jittery medium noise are used as noise at the readout point.
- the normalized linear density normalized by Tb is defined by T 50 / T b .
- the SP decoder 63 performs decoding in consideration of the burst information obtained by the burst detector 61 ′ using the LDPC code check matrix (see Example 1-1).
- FIG. 18 is a diagram showing the relationship between the required S / N ratio and the burst length L flip for decoding predetermined data without error, obtained by computer simulation.
- the PR system is the PR1 system, and the AR channel model used in the APP decoder 62 ([3] A.
- ⁇ (white circles) and ⁇ (white squares) indicate the RS-LDPC-PCBD method in which the RS code proposed in Example 1-2 is concatenated and the LDPC- proposed in Example 1-1, respectively.
- Example 1-2 the LDPC encoding / iterative decoding method (RS-LDPC-PCBD method) in which the RS code is concatenated is applied to the perpendicular magnetic recording / reproducing system with a bit-flipping burst. Proposed and evaluated the performance. As a result, it has been clarified that the correction capability for a short signal burst is improved by connecting the RS code to the LDPC code.
- RS-LDPC-PCBD method the LDPC encoding / iterative decoding method in which the RS code is concatenated is applied to the perpendicular magnetic recording / reproducing system with a bit-flipping burst.
- Example 1-3 The inventors of the present application use burst information detected by using amplitude fluctuation from the output of the APP decoder for burst-like signal degradation such as a medium defect occurring in a hard disk device, thereby generating a burst-like signal.
- An LDPC encoding / iterative decoding system that is resistant to degradation ([1] R. Berger et al., IEEE Trans. Magn., Vol. 38, no. 5, pp. 2435-2437, Sept. 2002.) ([2] Nakamura et al., IEICE Technical Report, MR2007-23, Oct.2007.).
- Example 1-1 burst detection using a parity check matrix of an LDPC code is effective for a bit-flipping signal burst that is difficult to detect from a reproduction waveform, an APP decoder output, or the like. showed that.
- a medium defect and a bit-flipping signal burst occur at the same time, it is not possible to deal with a bit-flipping signal burst only by performing burst detection using amplitude fluctuations from a reproduction waveform or the like.
- Example 1-3 performance evaluation of an LDPC encoding / iterative decoding method provided with a burst detector using a parity check matrix of an LDPC code in a perpendicular magnetic recording / reproducing system accompanied by burst-like signal deterioration due to a medium defect I do.
- the configuration of the first embodiment 1-3 is the same as that shown in FIG.
- a long sector format of 4096 bytes / sector is assumed.
- the isolated reproduction waveform is assumed to be the waveform of the above-described equation (1), white Gaussian noise at the readout point, and jittery medium noise. Further, it is assumed that the assumption media defect as a burst-like signal degradation, a medium defect of the medium defect length L defect normalized occurs only once per sector bit interval T b.
- the SP decoder 63 performs decoding in consideration of the burst information obtained by the burst detector 61 ′ using the LDPC code check matrix.
- the LLR is weighted in the SP decoder 63 ([2] Nakamura et al., Shingaku Giho, MR2007-23, Oct. 2007.). And performance evaluation is performed by calculating
- FIG. 19 is a diagram showing the relationship between the required SN ratio and the medium defect length for decoding predetermined data without error.
- the number of iterations i sp and i in in the SP decoder 63 and the iterative decoding apparatus 601 are both 5 times, and the weight coefficient W La to be multiplied by the LLR in the SP decoder 63 is 0.7 (same document).
- ⁇ (white circles), ⁇ (white squares), and ⁇ (white triangles) in the figure are LDPC-PCBD systems and LDPC-BD systems that use burst amplitude detection from APP decoder output (the same document). 2 shows characteristics of the RLL-RS-EC scheme with erasure error correction by adding an RS code for burst error detection.
- LDPC-PCBD iterative decoding method with a burst detector using a parity check matrix of an LDPC code. System
- Example 1-4 As shown in Example 1-1, it was difficult for the burst detector using the parity check matrix of the LDPC code to detect a relatively short burst. Therefore, in this embodiment 1-4, improvement in detection capability for short bit flipping signal bursts is studied, and a burst detector that is resistant to short bursts is realized by improving the moving average filter.
- an LDPC encoding / iterative decoding method (an improved LDPC-PCBD method) provided with a burst detector proposed in Embodiment 1-4
- the performance comparison of the RS coded erasure error correction method was performed by computer simulation.
- the LDPC encoding / iterative decoding scheme improved LDPC-PCBD scheme
- the burst detector proposed in Embodiment 1-4 is resistant to short bit flipping signal bursts. It became.
- the LDPC code not only provides high decoding gain when combined with iterative decoding, but also identifies burst information using amplitude fluctuations for burst-like signal degradation, and is a burst consisting of burst position and length. It is known to show high performance by performing iterative decoding using information ([2] W. Tan, J. R. Cruz, “Array codes for erasure correction in magnetic recording channels,” “IEEE Trans. Magn., Vol. 39, no. 5, pp. 2579-2581, Sept. 2003. [3] T. Morita, Y. Sato, T. Sugawara, “ECC-less LDPC c” ding for magnetic recording channels, “IEEE Trans. Magn., vol. 38, no.
- Example 1-1 the burst detector using the LDPC code parity check matrix proposed in Example 1-1 has difficulty in detecting a signal burst shorter than 50 bits depending on the configuration of a moving average (MA) filter.
- MA moving average
- FIG. 4 is referred to as a configuration diagram.
- the encoding block includes an RLL encoder 30, an LDPC encoder 31, a magnetic recording / reproducing system, and an equalizer 42.
- the input data sequence ⁇ ak ′′ ⁇ is encoded by the RLL encoder 30 128/130 (0, 16/8) ([8] Hidetoshi Saito, Toshihiko Iga, Masahiro Shirakawa, Yoshihiro Okamoto, Hisashi Osawa, “High-efficiency run length
- the configuration of the constraint code and its evaluation “Science Theory (C), vol. J86-C, no. 8, pp.
- the isolated reproduction waveform with respect to the stepped recording waveform is the expression (1) described above in Embodiment 1-1 ([9] Yoshihiro Okamoto, Mitsuteru Sato, Hidetoshi Saito, Hisashi Osawa, Hiroaki Muraoka, Nakamura Yoshihisa, “A Study on 3/4 MTR Coded PRML System in Perpendicular Recording System Using Double-Layer Media,” IEICE Technical Report, MR2000-8, June 2000.). Further, as defined above formula was also above the normalized normalized line density T b (2).
- the noise at the readout point is jittery medium noise caused by fluctuation of the magnetization transition point to white Gaussian, and white Gaussian noise (AWGN: additive white Gaussian noise) with an average value of 0 as system noise
- AWGN white Gaussian noise
- the SN ratio at the point is defined by the above-described equation (3). Further, the ratio of the jitter medium noise power to the total noise power at the readout point is defined by the above equation (4).
- the equalizer 42 is composed of a sixth-order Butterworth low-pass filter and a transversal filter taps Nt with normalized cut-off frequency x h normalized with f b, until the equalizer output from the recording head PR1 Characteristics ([10] ER Kretzmer, “Generalization of a technique for binary data communication,” IEEE Trans. Commun. Technol., Vol. COM-14, pp. 67-68, pp. 67-68. Perform waveform equalization.
- the (b) decoding block functioning as a decoding device includes an APP decoder 62 ([11] Y. Nakamura, Y. Okamoto, H. Osawa, H. H. Saito, H. Muraoka, Y. Nakamura, “A study of turbo decoding with embedded AR channel for perennial recording,” IE. Trans. ), SP decoder 63, burst detector 61 ′ functioning as a burst detection means, hard decision unit 64, and RLL decoder 67.
- the APP decoder 62 performs a posteriori probability decoding on the PR1 channel 401 to obtain a log likelihood ratio (LLR) sequence ⁇ L ( ⁇ c k ) ⁇ .
- LLR log likelihood ratio
- a range Lc 5 in which the correlation of the decoder input noise is taken into account (the same document).
- ⁇ L ( ⁇ c k ) ⁇ is input to the SP decoder 63 and the burst detector 61 ′.
- the burst detector 61 ′ detects a burst generation position and length from ⁇ L ( ⁇ c k ) ⁇ by using a parity check matrix that complies with the code constraint of the LDPC code, and detects the burst information to the SP decoder 63. (See Example 1-1).
- the SP decoder 63 Attenuates the LLR value based on the burst information and performs iterative decoding until the set number of iterations is reached.
- the burst information is detected only for the first time with respect to the output of the APP decoder, and is valid in the subsequent iterative decoding.
- the iterative decoding device 601 is configured by returning the SP decoder output to the APP decoder 62.
- the number of iterations in the SP decoder 63 is i sp
- the number of iterations to return from the SP decoder 63 to the APP decoder 62 is i in .
- the number of repetitions is 5 for both i sp and i in .
- the burst detector 61 ′ includes a hard decision unit 611 ′, a parity flag generator 612 ′, a filter 613 ′, and a burst information output unit 614 ′.
- the hard decision unit 611 ′ makes a hard decision on the APP decoder output sequence ⁇ L (c k ) ⁇ with 0 as a threshold value.
- the parity flag generator 612 ′ performs a parity check for each row based on the hard decision result of ⁇ L ( ⁇ c k ) ⁇ based on the parity check matrix of the LDPC code, and creates a parity flag from the parity check result.
- the filter 613 ′ facilitates the burst detection by increasing the value of the signal burst generation position by a two-stage MA (Moving Average) filter.
- MA filter of two-stage configuration first, taking a moving average parity flag over between the front and rear L MA1 channel bits, and has a configuration that takes a moving average over a period again before and after L MA2 channel bits.
- the first-stage and second-stage MA filters are denoted as MA 1 and MA 2 , respectively. Due to the structure of the MA filter, there is a possibility that the burst may be detected shorter than before before and after the burst generation position and end position. For this reason, burst information is generated by using the detected burst expanded from the previous and subsequent LMA1 channel bits as a burst period, and is output to the SP decoder 63.
- K 1.5
- R J 80%
- SNR 21.5 dB
- x h 0.4
- L MA1 100
- L MA2 200.
- FIG. 21 is a diagram showing the relationship between the detected average burst length and TH flip .
- the simulation conditions are the same as those in FIG.
- the dotted line in the figure indicates the generated burst length L flip .
- K 1.5
- R J 80%
- x h 0.4
- N t are optimum values N topt
- W La 0.7
- TH flip 0.12.
- the configuration of an MA filter capable of detecting a burst even when a signal burst having an L flip shorter than 200 occurs will be considered.
- Three types of two-stage MA filters of filter 3 are used.
- FIG. 24 is a diagram showing a relationship between a required SN ratio and L flip for decoding predetermined data without error when three types of two-stage MA filters are individually applied.
- the total number of user data is 10 Mbits
- K 1.5
- R J 80%
- x h 0.4
- N t N topt
- ⁇ (white circles), ⁇ (white triangles), and ⁇ (white squares) indicate the characteristics when the filter 1, the filter 2, and the filter 3 are applied, respectively.
- FIG. 25 is a flowchart illustrating an example of processing of the filter 613 ′ proposed in the present embodiment 1-4.
- the burst length BL max detected the longest by the three types of filters is BL max ⁇ 120, 120 ⁇ BL max ⁇ 300, 300 ⁇ BL max , filter 1 and filter 2 respectively.
- FIG. 26 is a diagram illustrating error rate characteristics of the LDPC encoding / iterative decoding scheme (LDPC-PCBD scheme).
- K 1.5
- R J 80%
- x h 0.4
- N t N top
- the total number of data is 10M bits.
- the white marks ( ⁇ , ⁇ ) indicate the improved LDPC-PCBD system equipped with the burst detector proposed in Example 1-4
- the black marks ( ⁇ , black ⁇ ) indicate the examples.
- FIG. 27 is a diagram showing the relationship between the required SN ratio and L flip for decoding predetermined data without error.
- ⁇ white circles
- ⁇ white triangles
- ⁇ white squares
- RS-EC system with erasure error correction used as a comparison system
- the RS code used in the RS-EC system is a code capable of 190 code symbol error correction using a GF (212) Galois field (the same document), and the erasure position information is an equalizer. Since the output amplitude variation is used for determination, it is impossible to detect a bit flipping signal burst, and erasure correction is not performed. From the figure, the improved LDPC-PCBD method using the burst detector of the method proposed in the embodiment 1-4 can be used even in a signal burst whose L flip is shorter than 60 that cannot be decoded without error in the non-improved LDPC-PCBD method. It can be seen that it is resistant.
- Embodiment 1-4 the configuration of a burst detector using a parity check matrix of an LDPC code that can be detected even when a short bit flipping signal burst of 50 channel bits or less occurs is examined. went. Also, in a magnetic recording / reproducing type in which a bit-flipping signal burst occurs, an LDPC encoding / iterative decoding method (improved LDPC-PCBD) provided with a burst detector proposed in the embodiment 1-4, and implementation Comparing the performance of LDPC coding / iterative decoding method (non-improved LDPC-PCBD) with burst detector of Example 1-1 and RS coding erasure error correction method (RS-EC method) by computer simulation went.
- RS-EC method RS coding erasure error correction method
- the LDPC encoding / iterative decoding method (improved LDPC-PCBD) using the burst detector proposed in the embodiment 1-4 can tolerate a burst whose bit flipping length L flip is less than 50. It became.
- a burst with L flip shorter than 300 it is clear that a burst can be allowed with a lower S / N ratio than the non-improved LDPC-PCBD system by using a burst detector of the improved LDPC-PCBD system. It became.
- 120 ⁇ L flip ⁇ 1000 it became clear that a burst can be allowed with an SN ratio that is lower by about 1 dB or more than the RS coded erasure error correction method (RS-EC method).
- RS-EC method RS coded erasure error correction method
- the present invention was applied on the assumption that an LDPC code can be used as an error correction code for a hard disk device, and it was confirmed that a sufficiently long burst could be supported by computer simulation. . In addition, it was confirmed by computer simulation that the present invention can be used for detection of asynchronous writing which may cause a problem in a next-generation hard disk device using a bit pattern medium.
- Example 2 in which the decoding apparatus according to this embodiment is applied to a receiving apparatus will be described.
- FIG. 28 shows an LDPC encoding / iterative decoding scheme (LDPC ⁇ ) having a burst detector using a parity check matrix of an LDPC code applied to a binary input additive white Gaussian noise (AWGN) channel.
- LDPC ⁇ LDPC encoding / iterative decoding scheme
- AWGN binary input additive white Gaussian noise
- a bit flipping signal burst is assumed as an error that is difficult to detect even in a binary input AWGN channel.
- this is realized by inverting ⁇ b k ⁇ over the length L flip normalized by the user bit interval T b .
- a recording sequence ⁇ c k ⁇ including a bit flipping error only once per sector is sent to the binary input AWGN communication path.
- a code word of a binary linear code is sent to the channel after binary-bipolar conversion at the time of transmission ([1] Tadashi Wadayama, “Low-density parity check code” And its decoding method, "Triqueps, June 2002.).
- the SN ratio in the binary input AWGN communication channel is defined by the following equation (the same document).
- E b is the information symbol 1 per bit transmission energy
- N 0 represents the one-sided power spectral density of white Gaussian noise, respectively.
- the SN ratio in Expression (5) does not include a burst.
- the channel output ⁇ L ( ⁇ b k ) ⁇ is regarded as reliability and decoding is performed.
- ⁇ L ( ⁇ b k ) ⁇ is input to the SP decoder 63 and the burst detector 61 ′.
- the burst detector 61 ′ uses the parity check matrix of the LDPC code to detect the position and length of the burst from ⁇ L ( ⁇ c k ) ⁇ and transmits it to the SP decoder 63 as burst information.
- the burst detection method is performed in the same process as that of the magnetic recording / reproducing system described above.
- the SP decoder 63 repeats SP decoding taking the burst information into consideration a predetermined number of times i sp .
- FIG. 29 is a diagram showing BER characteristics in a binary input AWGN communication path.
- L flip 1000.
- ⁇ (white circles) and ⁇ (white squares) indicate when the burst detector is used and when it is not used.
- the burst detector when the burst detector is not used, the error rate tends to be saturated, but by using the berth detector, decoding can be performed without error at about 1.6 dB. Further, it was found that by using the burst detector, it is possible to suppress the SN ratio deterioration of about 0.5 dB as compared with the characteristics when no signal burst occurs.
- the present invention is not limited to this, and the present invention is not limited to a recording apparatus such as a magnetic disk apparatus or a storage such as an optical disk.
- the present invention can be widely applied to products, digital signal processing in the communication field such as wireless LAN, wireless Internet, and long-distance optical communication.
- the present invention provides not only error correction for the hard disk device described above, but also asynchronous write detection for the next generation hard disk device using a bit pattern medium, burst error detection in communication, burst error for semiconductor memory. It is extremely useful for detection.
- the decoding device 60 has been described with respect to the example configured by the combination of the APP decoder 62 and the SP decoder 63.
- the present invention is not limited to this, and the LDPC code can be decoded.
- Known decoding means may be used, or a combination of a soft output soft input inner decoder and an outer decoder using likelihood may be used.
- the weighting is described as a method for suppressing the increase in the likelihood.
- the present invention is not limited to this.
- the likelihood corresponding to the position and length of the detected burst error is set.
- the likelihood increase may be suppressed by masking a certain number of repetitions.
- the decryption device 60 performs processing in a stand-alone form has been described as an example. However, the decryption device 60 performs processing in response to a request from a client terminal configured in a separate casing, and the decryption is performed. You may comprise so that a conversion result may be returned to the said client terminal.
- all or part of the processes described as being automatically performed can be performed manually, or the processes described as being performed manually can be performed. All or a part can be automatically performed by a known method.
- each illustrated component is functionally conceptual and does not necessarily need to be physically configured as illustrated.
- the processing functions provided in each device of the decoding device 60 are all or any part of the CPU (Central Processing Unit).
- a program interpreted and executed by the CPU or may be realized as wired logic hardware.
- the program is recorded on a recording medium to be described later, and is mechanically read by the decoding device 60 as necessary.
- a storage unit such as a ROM or an HD stores a computer program for performing various processes by giving instructions to the CPU in cooperation with an OS (Operating System). This computer program is executed by being loaded into the RAM, and constitutes a control unit in cooperation with the CPU.
- the computer program may be stored in an application program server connected to the decryption device 60 via an arbitrary network, and may be downloaded in whole or in part as necessary. is there.
- the program according to the present invention can also be stored in a computer-readable recording medium.
- the “recording medium” refers to any “portable physical medium” such as a flexible disk, a magneto-optical disk, a ROM, an EPROM, an EEPROM, a CD-ROM, an MO, and a DVD, or a LAN, WAN, or Internet. It includes a “communication medium” that holds the program in a short period of time, such as a communication line or a carrier wave when the program is transmitted through a network represented by
- program is a data processing method described in an arbitrary language or description method, and may be in any form such as source code or binary code.
- program is not necessarily limited to a single configuration, but is distributed in the form of a plurality of modules and libraries, or in cooperation with a separate program represented by an OS (Operating System). Including those that achieve the function.
- OS Operating System
- a well-known configuration and procedure can be used for a specific configuration for reading a recording medium, a reading procedure, an installation procedure after reading, and the like in each device described in the embodiment.
- Various databases stored in the storage means are storage means such as a memory device such as RAM and ROM, a fixed disk device such as a hard disk, a flexible disk, and an optical disk.
- Various programs, tables, databases, etc. used for various processes Is stored.
- the decryption device 60 is connected to an information processing device such as a known personal computer or workstation, and the software (including programs, data, etc.) for realizing the method of the present invention is installed in the information processing device. It may be realized.
- the specific form of distribution / integration of the devices is not limited to that shown in the figure, and all or a part of them may be functional or physical in arbitrary units according to various additions or according to functional loads. Can be distributed and integrated.
- a special code for detecting a burst error is not added, and it is resistant to a burst error that does not involve amplitude fluctuations such as pole erase.
- a decoding device, a perpendicular magnetic recording / reproducing device, a receiving device, and a decoding method capable of performing certain decoding can be provided, so that the next generation magnetic recording / reproducing system, semiconductor field, storage memory This is extremely useful in a wide variety of fields such as fields and information and communication fields.
Landscapes
- Engineering & Computer Science (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- Probability & Statistics with Applications (AREA)
- Theoretical Computer Science (AREA)
- Error Detection And Correction (AREA)
Abstract
Description
601 繰り返し復号化装置
61 バースト検出手段
61´ バースト検出器
611 硬判定手段
611´ 硬判定器
612 パリティフラグ発生手段
612´ パリティフラグ発生器
613 フィルタ出力手段
613´ フィルタ
614 バースト情報出力手段
614´ バースト情報出力器
65 復号手段
62 APP復号器
63 SP復号器
64 硬判定器
66 RS復号器
67 RLL復号器
30 RLL符号器
31 LDPC符号器
32 バースト発生器
33 RS符号器
401 PR1チャネル
402 AWGNチャネル
41 磁気記録媒体(ディスク)
42 等化器
まず、本発明にかかる復号化装置の構成の一例について、図1を参照して説明する。図1は、本発明が適用される復号化装置60の構成の一例を示すブロック図であり、該構成のうち本発明に関係する部分のみを概念的に示している。
次に、このように構成された復号化装置60による基本処理の一例について、以下に図2および図3を参照して詳細に説明する。図2は、復号化装置60の基本処理の一例を示すフローチャートである。図3は、バースト検出手段61によるバースト検出処理の一例を示すフローチャートである。
上記実施の形態では、LDPC符号の検査行列を利用した新たなバースト検出法を提案した。本実施例1では、再生波形や等化器出力、APP復号器出力を利用した従来のバースト検出器では検出できないビットフリッピング状バーストに対する新たなバースト検出を検討する。また、このようなバーストを伴う垂直磁気記録再生系において、本実施例1-1で提案するバースト検出器を備えたLDPC符号化・繰り返し復号化方式と、従来の消失誤り訂正を伴うRS符号化方式との性能比較を、計算機シミュレーションにより行った。その結果、本願で提案するバースト検出器を用いて繰り返し復号を行うことで、従来のRS符号化方式に比べて良好な特性を示し、約1500ビットのビットフリッピング状バーストを許容できることが明らかとなった。
K=T50/Tb (2)
実施例1-1では、再生波形やAPP復号器出力等によっては検出が困難なビットフリッピング状の信号バーストに対して、LDPC符号のパリティ検査行列を利用したバースト検出が有効であることを示した。しかし、検査行列によるバースト検出法は、移動平均フィルタを用いてバースト情報を作成するため、200チャネルビット以下の短いバーストの検出について改善の余地があった。そこで、本実施例1-2では、LDPC符号化・繰り返し復号化方式(LDPC-PCBD方式)にRS符号を連結することで、短い信号バーストに対しても、誤りなく復号できるシステムを提案し、ビットフリッピング状の信号バーストを伴う垂直磁気記録再生系に適用して性能評価する。
本願発明者らは、ハードディスク装置で発生する媒体欠陥のようなバースト状の信号劣化に対して、APP復号器出力から振幅変動を利用して検出されたバースト情報を用いることで、バースト状の信号劣化([1]R.Berger et al., IEEE Trans.Magn., vol.38, no.5, pp.2435-2437, Sept.2002.)に耐性のあるLDPC符号化・繰り返し復号化方式を構築した([2]仲村他,信学技報, MR2007-23, Oct.2007.)。また、実施例1-1では、再生波形やAPP復号器出力等からの検出が困難なビットフリッピング状の信号バーストに対しては、LDPC符号のパリティ検査行列を用いたバースト検出が有効であることを示した。媒体欠陥とビットフリッピング状の信号バーストが同時に発生した場合、再生波形等から振幅変動を利用してバースト検出を行うだけではビットフリッピング状の信号バーストに対応することができない。本実施例1-3では、媒体欠陥によるバースト状の信号劣化を伴う垂直磁気記録再生系において、LDPC符号の検査行列を利用したバースト検出器を備えたLDPC符号化・繰り返し復号化方式の性能評価を行う。
実施例1-1で示したように、LDPC符号の検査行列を利用したバースト検出器は、比較的短いバーストの検出が困難であった。そこで、本実施例1-4では、短いビットフリッピング状の信号バーストに対する検出能力の改善について検討し、移動平均フィルタを改良することで、短いバーストに耐性のあるバースト検出器を実現する。また、ビットフリッピング状の信号バーストを伴う垂直磁気記録再生系において、本実施例1-4で提案するバースト検出器を備えたLDPC符号化・繰り返し復号化方式(改良型LDPC-PCBD方式)と、RS符号化消失誤り訂正方式(RLL-RS-EC方式)の性能比較を計算機シミュレーションにより行った。その結果、本実施例1-4で提案するバースト検出器を備えたLDPC符号化・繰り返し復号化方式(改良型LDPC-PCBD方式)は、短いビットフリッピング状信号バーストにも耐性があることが明らかとなった。また、40~1000ビット程度のビットフリッピング状信号バーストに対して、RS符号化消失誤り訂正方式(RLL-RS-EC方式)よりも良好な特性が得られることが明らかとなった。
さて、これまで本発明の実施の形態について説明したが、本発明は、上述した実施の形態以外にも、特許請求の範囲に記載した技術的思想の範囲内において種々の異なる実施の形態にて実施されてよいものである。
Claims (24)
- 符号化データ信号に対して復号化処理を実行する復号化装置であって、
低密度パリティ検査符号により符号化された上記符号化データ信号のパリティ検査を行うことにより、バースト情報を出力するバースト検出手段、
を備えたことを特徴とする復号化装置。 - 請求項1に記載の復号化装置において、
上記バースト情報は、上記符号化データ信号のバーストエラーの位置および長さを特定した情報を含むこと、
を特徴とする復号化装置。 - 請求項2に記載の復号化装置において、
上記バースト検出手段は、
上記符号化データ信号に対してパリティ検査行列の行毎にパリティ検査を行い、上記符号化データ信号の各ビットに対応する上記パリティ検査行列の列毎に、上記パリティ検査で誤りとして検出した行のビット「1」の数を合算し、上記合算した数の分布に従って上記符号化データ信号の上記バーストエラーの上記位置および上記長さを特定することにより上記バースト情報を生成して出力すること、
を特徴とする復号化装置。 - 請求項2に記載の復号化装置において、
上記バースト検出手段により出力された上記バースト情報に基づいて、上記バーストエラーの上記位置および上記長さに対応する、上記符号化データ信号のビットの尤度の上昇を抑制して繰り返し復号化処理を実行する復号手段、
を更に備えたことを特徴とする復号化装置。 - 請求項4に記載の復号化装置において、
上記バースト検出手段は、
上記復号手段から出力された復号化結果を硬判定する硬判定手段と、
上記硬判定手段により硬判定された上記復号化結果について、パリティ検査行列に基づいてパリティ検査を行って、パリティフラグを発生させるパリティフラグ発生手段と、
上記パリティフラグ発生手段により発生された上記パリティフラグに、移動平均フィルタを適用してフィルタ出力を行うフィルタ出力手段と、
上記フィルタ出力手段により出力された上記フィルタ出力について閾値判定を行うことにより上記バーストエラーの上記位置および上記長さを特定して上記バースト情報を上記復号手段に出力するバースト情報出力手段と、
を備えたことを特徴とする復号化装置。 - 請求項5に記載の復号化装置において、
上記フィルタ出力手段は、
上記パリティフラグに上記移動平均フィルタを多段階で適用すること、
を特徴とする復号化装置。 - 請求項5または6に記載の復号化装置において、
上記フィルタ出力手段は、
上記バースト情報出力手段により上記フィルタ出力の閾値判定が行われることにより特定された上記バーストエラーの上記長さが所定値以上の場合、適用した上記移動平均フィルタより長い区間の移動平均をフィルタ出力する上記移動平均フィルタを適用して再度フィルタ出力を行うこと、
を特徴とする復号化装置。 - 請求項4乃至7のいずれか一つに記載の復号化装置において、
上記復号手段は、
上記符号化データ信号に対して事後確率復号化処理を実行して、上記尤度を含む上記復号化結果を出力するAPP(a posteriori probability)復号器と、
上記バースト検出器から出力された上記バースト情報に基づいて、上記APP復号器から出力された上記尤度の上昇を抑制して、上記APP復号器から出力された上記復号化結果に対する繰り返し復号化処理を実行し、更新した上記尤度を含む復号化結果を上記APP復号器の入力として出力するSP(sum-product)復号器と、
を備えたことを特徴とする復号化装置。 - 請求項8に記載の復号化装置において、
上記SP復号器は、
上記バースト情報に基づいて、上記APP復号器から出力された上記尤度の上昇を重み付けにより抑制すること、
を特徴とする復号化装置。 - 請求項4乃至9のいずれか一つに記載の復号化装置において、
上記符号化データ信号は、更にRS(Reed-Solomon)符号により符号化されており、
上記復号化装置は、
上記復号手段の出力に対して、上記RS符号の復号化を行うRS復号器、
を更に備えたことを特徴とする復号化装置。 - 請求項2乃至10のいずれか一つに記載の復号化装置において、
上記バーストエラーは、媒体欠陥とビットフリッピング状の信号バーストが同時に発生したことによるものであること、
を特徴とする復号化装置。 - 請求項1乃至11のいずれか一つに記載の復号化装置を備えた垂直磁気記録再生装置。
- 請求項1乃至11のいずれか一つに記載の復号化装置を備えた受信装置。
- 符号化データ信号に対して復号化処理を実行する復号化方法であって、
低密度パリティ検査符号により符号化された上記符号化データ信号のパリティ検査を行うことにより、バースト情報を出力するバースト検出ステップ、
を含むことを特徴とする復号化方法。 - 請求項14に記載の復号化方法において、
上記バースト情報は、上記符号化データ信号のバーストエラーの位置および長さを特定した情報を含むこと、
を特徴とする復号化方法。 - 請求項15に記載の復号化方法において、
上記バースト検出ステップは、
上記符号化データ信号に対してパリティ検査行列の行毎にパリティ検査を行い、上記符号化データ信号の各ビットに対応する上記パリティ検査行列の列毎に、上記パリティ検査で誤りとして検出した行のビット「1」の数を合算し、上記合算した数の分布に従って上記符号化データ信号の上記バーストエラーの上記位置および上記長さを特定することにより上記バースト情報を生成して出力すること、
を特徴とする復号化方法。 - 請求項15に記載の復号化方法において、
上記バースト検出ステップにて出力された上記バースト情報に基づいて、上記バーストエラーの上記位置および上記長さに対応する、上記符号化データ信号のビットの尤度の上昇を抑制して繰り返し復号化処理を実行する復号ステップ、
を更に含むことを特徴とする復号化方法。 - 請求項17に記載の復号化方法において、
上記バースト検出ステップは、
上記復号ステップにて出力された復号化結果を硬判定する硬判定ステップと、
上記硬判定ステップにて硬判定された上記復号化結果について、パリティ検査行列に基づいてパリティ検査を行って、パリティフラグを発生させるパリティフラグ発生ステップと、
上記パリティフラグ発生ステップにて発生された上記パリティフラグに、移動平均フィルタを適用してフィルタ出力を行うフィルタ出力ステップと、
上記フィルタ出力ステップにて出力された上記フィルタ出力について閾値判定を行うことにより上記バーストエラーの上記位置および上記長さを特定して上記バースト情報を上記復号ステップに出力するバースト情報出力ステップと、
を含むことを特徴とする復号化方法。 - 請求項18に記載の復号化方法において、
上記フィルタ出力ステップは、
上記パリティフラグに上記移動平均フィルタを多段階で適用すること、
を特徴とする復号化方法。 - 請求項18または19に記載の復号化方法において、
上記フィルタ出力ステップは、
上記バースト情報出力ステップにて上記フィルタ出力の閾値判定が行われることにより特定された上記バーストエラーの上記長さが所定値以上の場合、適用した上記移動平均フィルタより長い区間の移動平均をフィルタ出力する上記移動平均フィルタを適用して再度フィルタ出力を行うこと、
を特徴とする復号化方法。 - 請求項17乃至20のいずれか一つに記載の復号化方法において、
上記復号ステップは、
上記符号化データ信号に対して事後確率復号化処理を実行して、上記尤度を含む上記復号化結果を出力するAPP(a posteriori probability)復号ステップと、
上記バースト検出ステップにて出力された上記バースト情報に基づいて、上記APP復号ステップにて出力された上記尤度の上昇を抑制して、上記APP復号ステップにて出力された上記復号化結果に対する繰り返し復号化処理を実行し、更新した上記尤度を含む復号化結果を上記APP復号ステップの入力として出力するSP(sum-product)復号ステップと、
を含むことを特徴とする復号化方法。 - 請求項21に記載の復号化方法において、
上記SP復号ステップは、
上記バースト情報に基づいて、上記APP復号ステップにて出力された上記尤度の上昇を重み付けにより抑制すること、
を特徴とする復号化方法。 - 請求項17乃至22のいずれか一つに記載の復号化方法において、
上記符号化データ信号は、更にRS(Reed-Solomon)符号により符号化されており、
上記復号ステップにおける出力に対して、上記RS符号の復号化を行うRS復号ステップ、
を更に含むことを特徴とする復号化方法。 - 請求項15乃至23のいずれか一つに記載の復号化方法において、
上記バーストエラーは、媒体欠陥とビットフリッピング状の信号バーストが同時に発生したことによるものであること、
を特徴とする復号化方法。
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US13/000,717 US8739002B2 (en) | 2008-06-30 | 2008-12-26 | Decoder, perpendicular magnetic recording and reproducing device, receiving device, and decoding method |
JP2010518878A JP5628670B2 (ja) | 2008-06-30 | 2008-12-26 | 復号化装置、垂直磁気記録再生装置、受信装置、および、復号化方法 |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2008171121 | 2008-06-30 | ||
JP2008-171121 | 2008-06-30 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2010001502A1 true WO2010001502A1 (ja) | 2010-01-07 |
Family
ID=41465612
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2008/073794 WO2010001502A1 (ja) | 2008-06-30 | 2008-12-26 | 復号化装置、垂直磁気記録再生装置、受信装置、および、復号化方法 |
Country Status (3)
Country | Link |
---|---|
US (1) | US8739002B2 (ja) |
JP (1) | JP5628670B2 (ja) |
WO (1) | WO2010001502A1 (ja) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102404260A (zh) * | 2010-09-09 | 2012-04-04 | 株式会社东芝 | 判决反馈式均衡器 |
CN112997158A (zh) * | 2018-10-16 | 2021-06-18 | 美光科技公司 | 用于错误校正的方法和装置 |
Families Citing this family (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8225252B2 (en) * | 2010-06-25 | 2012-07-17 | Intel Corporation | Systems, methods, apparatus and computer readable mediums for use in association with systems having interference |
US8589841B2 (en) * | 2012-04-05 | 2013-11-19 | International Business Machines Corporation | Automatic parity checking identification |
GB2506658B (en) * | 2012-10-05 | 2015-01-21 | Broadcom Corp | Method and apparatus for signal detection and decoding |
TWI543178B (zh) * | 2014-06-10 | 2016-07-21 | 群聯電子股份有限公司 | 解碼方法、記憶體儲存裝置及記憶體控制電路單元 |
JP6567238B1 (ja) * | 2019-02-22 | 2019-08-28 | 三菱電機株式会社 | 誤り訂正復号装置および誤り訂正復号方法 |
JPWO2020245883A1 (ja) * | 2019-06-03 | 2020-12-10 | ||
JP2021034825A (ja) * | 2019-08-21 | 2021-03-01 | 株式会社東芝 | 磁気ディスク装置 |
US10778248B1 (en) * | 2020-01-30 | 2020-09-15 | TenaFe, Inc. | Low-density parity-check decoding with de-saturation |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2003068024A (ja) * | 2001-06-11 | 2003-03-07 | Fujitsu Ltd | 記録再生装置、信号復号回路、エラー訂正方法、及び反復型復号器 |
JP2005166089A (ja) * | 2003-11-28 | 2005-06-23 | Toshiba Corp | ディスク記憶装置、データ再生装置及びデータ再生方法 |
JP2006139815A (ja) * | 2004-11-10 | 2006-06-01 | Victor Co Of Japan Ltd | 記録装置、再生装置及び記録媒体 |
JP2007087530A (ja) * | 2005-09-22 | 2007-04-05 | Rohm Co Ltd | 信号復号方法、信号復号装置および信号記憶システム |
JP2008065969A (ja) * | 2006-08-09 | 2008-03-21 | Fujitsu Ltd | 符号化装置、復号化装置、符号化方法、復号化方法および記憶装置 |
Family Cites Families (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP3808769B2 (ja) * | 2001-12-27 | 2006-08-16 | 三菱電機株式会社 | Ldpc符号用検査行列生成方法 |
JP3745709B2 (ja) * | 2002-06-28 | 2006-02-15 | インターナショナル・ビジネス・マシーンズ・コーポレーション | 符号化装置、復号化装置、符号化方法、復号化方法、プログラム、プログラム記録媒体、及びデータ記録媒体 |
US7296216B2 (en) * | 2003-01-23 | 2007-11-13 | Broadcom Corporation | Stopping and/or reducing oscillations in low density parity check (LDPC) decoding |
US20050003769A1 (en) * | 2003-07-02 | 2005-01-06 | Foerster Jeffrey R. | Ultra-wideband transceiver architecture and associated methods |
US7447984B2 (en) * | 2005-04-01 | 2008-11-04 | Broadcom Corporation | System correcting random and/or burst errors using RS (Reed-Solomon) code, turbo/LDPC (Low Density Parity Check) code and convolutional interleave |
JP5219699B2 (ja) | 2007-08-30 | 2013-06-26 | パナソニック株式会社 | 符号化装置及び復号装置 |
JP4946844B2 (ja) | 2007-12-13 | 2012-06-06 | ソニー株式会社 | 記録再生装置および記録再生方法 |
US8201051B2 (en) * | 2008-10-15 | 2012-06-12 | Lsi Corporation | Method for detecting short burst errors in LDPC system |
-
2008
- 2008-12-26 JP JP2010518878A patent/JP5628670B2/ja not_active Expired - Fee Related
- 2008-12-26 US US13/000,717 patent/US8739002B2/en not_active Expired - Fee Related
- 2008-12-26 WO PCT/JP2008/073794 patent/WO2010001502A1/ja active Application Filing
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2003068024A (ja) * | 2001-06-11 | 2003-03-07 | Fujitsu Ltd | 記録再生装置、信号復号回路、エラー訂正方法、及び反復型復号器 |
JP2005166089A (ja) * | 2003-11-28 | 2005-06-23 | Toshiba Corp | ディスク記憶装置、データ再生装置及びデータ再生方法 |
JP2006139815A (ja) * | 2004-11-10 | 2006-06-01 | Victor Co Of Japan Ltd | 記録装置、再生装置及び記録媒体 |
JP2007087530A (ja) * | 2005-09-22 | 2007-04-05 | Rohm Co Ltd | 信号復号方法、信号復号装置および信号記憶システム |
JP2008065969A (ja) * | 2006-08-09 | 2008-03-21 | Fujitsu Ltd | 符号化装置、復号化装置、符号化方法、復号化方法および記憶装置 |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102404260A (zh) * | 2010-09-09 | 2012-04-04 | 株式会社东芝 | 判决反馈式均衡器 |
CN112997158A (zh) * | 2018-10-16 | 2021-06-18 | 美光科技公司 | 用于错误校正的方法和装置 |
US11907061B2 (en) | 2018-10-16 | 2024-02-20 | Lodestar Licensing Group Llc | Methods and devices for error correction |
Also Published As
Publication number | Publication date |
---|---|
US20110107178A1 (en) | 2011-05-05 |
JPWO2010001502A1 (ja) | 2011-12-15 |
JP5628670B2 (ja) | 2014-11-19 |
US8739002B2 (en) | 2014-05-27 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP5628670B2 (ja) | 復号化装置、垂直磁気記録再生装置、受信装置、および、復号化方法 | |
US8151162B2 (en) | Encoding device, decoding device, encoding/decoding device, and recording/reproducing device | |
JP4879323B2 (ja) | 誤り訂正復号装置および再生装置 | |
US9048879B1 (en) | Error correction system using an iterative product code | |
JP2009506474A (ja) | ソフト復号化方法及び装置、エラー訂正方法及び装置、ソフト出力方法及び装置 | |
US8341506B2 (en) | Techniques for correcting errors using iterative decoding | |
JP5297536B2 (ja) | 蓄積装置テストに関するldpcコードのエラー訂正能力調節 | |
JP2003068024A (ja) | 記録再生装置、信号復号回路、エラー訂正方法、及び反復型復号器 | |
JP2005093038A (ja) | 記録再生装置および記録再生回路 | |
US20090327832A1 (en) | Decoder and recording/reproducing device | |
US20070094582A1 (en) | Encoding method to QC code | |
CN102063918B (zh) | 编码方法、编码设备、解码方法和解码设备 | |
Oh et al. | RS-LDPC concatenated coding for the modern tape storage channel | |
US8786968B2 (en) | Data storage device including a recording channel, a detector, and a noise prediction circuit, and method of processing a signal in a data storage device | |
US8949702B2 (en) | Systems and methods for detector side trapping set mitigation | |
Moon et al. | Detection of prescribed error events: Application to perpendicular recording | |
JP4294407B2 (ja) | 信号処理方法及び信号処理回路 | |
Kong et al. | Performance evaluation of the reed solomon and low density parity check codes for blu-ray disk channels | |
JP2004145972A (ja) | リードチャネル復号器、リードチャネル復号方法およびリードチャネル復号プログラム | |
US9548762B2 (en) | Normalization factor adaptation for LDPC decoding for hard disk drive systems | |
Marinoni et al. | Efficient design of non-binary LDPC codes for magnetic recording channels, robust to error bursts | |
Nishikawa et al. | A study on iterative decoding with LLR modulator by parity check information in SMR system | |
Nakamura et al. | A new burst detection scheme using parity check matrix of LDPC code for bit flipping burst-like signal degradation | |
Nakamura et al. | Burst detection by parity check matrix of LDPC code for perpendicular magnetic recording using bit-patterned medium | |
Nakamura et al. | A study of LDPC coding and iterative decoding system in magnetic recording system using bit-patterned medium with write error |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 08874877 Country of ref document: EP Kind code of ref document: A1 |
|
ENP | Entry into the national phase |
Ref document number: 2010518878 Country of ref document: JP Kind code of ref document: A |
|
WWE | Wipo information: entry into national phase |
Ref document number: 13000717 Country of ref document: US |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 08874877 Country of ref document: EP Kind code of ref document: A1 |