JP2006166484A - Decoding apparatus and method, and recording medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To perform encoding using a code conversion table whereon parities of code sequences, until a code state becomes the same, are different. <P>SOLUTION: A codeword assignment of the code conversion table meets q<SB>0</SB>≠q<SB>1</SB>, for an arbitrary information sequence even if a codeword restriction length in decoding is divided into three blocks and an insertion position of a DC control bit is inserted into the position of a first bit or a second bit of an information word. For example, when information sequences d<SB>0</SB>and d<SB>1</SB>inserting temporary DC control bits 1 and 0 in the top of an information sequence [1, 1, 0, 0, 0, 1, 0] are encoded using a predetermined code conversion table with a state 3 as a starting point, code states s<SB>0</SB>and s<SB>1</SB>become the same, s<SB>0</SB>=s<SB>1</SB>=6, in the third block. A complementary q<SB>0</SB>of 2 of the total of a code sequence c<SB>0</SB>, by the time point when the code states become the same, is 0, and a complementary q<SB>1</SB>of 2 of the total of a code sequence c<SB>1</SB>is 1, so that q<SB>0</SB>≠q<SB>1</SB>is established. The present invention is applicable to a recording and reproducing apparatus or an encoding apparatus. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an encoding device, an encoding method, a recording medium, and a program, and in particular, inserts DC control bits for performing DC control of a code sequence into an information sequence, and converts an m-bit information word into an n-bit code. The present invention relates to an encoding device and an encoding method, a recording medium, and a program that are suitable for use in continuous conversion into words.

  In various recording / reproducing apparatuses and communication apparatuses, in many cases, the error rate of digital transmission information is reduced by encoding input information.

  FIG. 1 shows a block diagram of a digital signal processing circuit in a conventional general recording / reproducing apparatus 1.

  The encoding unit 11 encodes a continuously input information sequence in a ratio of m: n (m is a data bit length before encoding and n is a data bit length after encoding), and is a binary recording code. Generate a series. Here, m is an information word length, n is a code word length, and m / n is a code rate or a coding rate.

  The recording / reproducing unit 12 includes, for example, a magnetic head, an optical pickup, and a control circuit for controlling the driving thereof, and records the recording code input from the encoding unit 11 on a recording medium (not shown). The recording / reproducing unit 12 reproduces a signal recorded on the recording medium and outputs an analog reproduced wave to the analog equalizing unit 13. The analog equalizer 13 equalizes the reproduction wave input from the recording / reproducing unit 12 to a predetermined target equalization characteristic.

  The A / D conversion unit 14 converts the supplied equalized signal into a digital reproduction signal. The A / D converter 14 includes a PLL (Phase Locked Loop) circuit. As the PLL, a hybrid digital PLL that performs only phase error detection by a digital unit may be used, or a full digital PLL that performs both phase error detection and signal synchronization by a digital unit may be used.

  When equalization by the analog equalization unit 13 is not sufficient, a digital equalization unit may be further provided between the A / D conversion unit 14 and the code detection unit 15. In this case, a low-pass filter may be provided in place of the analog equalization unit 13.

  The code detector 15 receives a digital reproduction signal (equalized signal) and converts it into a code (detects the code). In recent years, a soft decision detector such as a Viterbi detector is generally used for the code detector 15. The decoding unit 16 receives detection codes that are continuously input, decodes them into detection information at a ratio of n: m, and outputs the detection information as a detection information sequence.

  In FIG. 1, various codes are put into practical use as codes used in the encoding process executed by the encoding unit 11. For example, assuming NRZI (non return to zero on one) modulation in which the recording rectangular wave is inverted by bit 1, the minimum value (minimum run) of 0s between 1 and 1 in the code sequence is d. The (d, k) run length limited (RLL) code in which the maximum value (maximum run) is limited to k is widely used in various recording and reproducing apparatuses.

  In particular, in an optical disc reproducing apparatus using an optical pickup for signal reproduction, a minimum run limit code with a minimum run d of 1 or more is widely used to reduce deterioration of reproduction signal quality due to nonlinearity of the optical pickup. . For example, in a recording / reproducing device for a compact disk (CD: Compact Disk) and a recording / reproducing device for a mini disk (MD: Mini Disk) that are widely used, encoding called EFM (eight to fourteen modulation) is used. A (2,10) RLL code with a rate of 8/17 is employed. In MD DATA2 used in a mini-disc recording / reproducing apparatus for data recording, a (1,7) RLL code with a coding rate of 2/3 is used. It has been adopted.

  Further, the (d, k) RLL code is adopted not only in the recording / reproducing apparatus, but also in, for example, a wireless communication apparatus using infrared light called IrDA-VFIr (Infrared Data Association, Very Fast Infrared) A (1,13) RLL code with a coding rate of 2/3 is employed.

  Among the (d, k) RLL codes, for example, the EFM code is a fixed length code, and the (1, 7) RLL code is a variable length code. The variable length code is generally encoded based on either a look ahead type code conversion table or a finite state type code conversion table.

  2 and 3 are commonly known (1,7) RLL code look-ahead code conversion tables according to Cohn-Jacoby-Bates, FIG. 2 is a basic conversion table, and FIG. It is a conversion table in. The information sequence is normally encoded according to the basic conversion table of FIG. 2, but when the information sequence described in the conversion table of violation in FIG. 3 is supplied, the conversion table of violation in FIG. 3 has priority. And encoding processing is executed (see, for example, Patent Document 1).

U.S. Patent, M. Cohn; G. Jacoby; C. Bates, USP 4,337,458

  Assuming that one block is a minimum unit of encoding (code conversion unit, for example, corresponding to 2 bits of input information or 3 bits of code in the basic conversion table of FIG. 2), the code conversion tables of FIGS. The encoding unit 11 that performs the encoding process based on the information needs to look ahead for the information sequence for one block in order to execute the encoding process. Further, the look-ahead type code conversion tables shown in FIGS. 2 and 3 are limited by reassigning a code word delayed from the information word to the information word by the same amount as the look-ahead amount necessary in the encoding unit 11. It is possible to convert to a state type code conversion table.

  Here, the code word delayed by one block is reassigned to the information word, for example, in FIG. 3, since the two code words 001.000 are assigned to the two information words 10.00, the immediately preceding information word In the state where is 10, the code word 001 is assigned to the information word 00.

  FIG. 4 is a five-state code conversion table of a finite state type (1,7) RLL code converted from the look-ahead type code conversion tables of FIGS. In FIG. 4, 3 bits shown before the slash (/) indicate a code word, and a number shown after the slash indicates a transition state.

  That is, the finite state type code conversion table of FIG. 4 is obtained by assigning a 3-bit codeword delayed by one block to each 2-bit information word in the look-ahead type code conversion tables of FIGS. For example, Non-Patent Document 1).

A. Gallopoulos, C. Heegard, and P. Siegel, “The Power Spectrum of Run-Length-Limited Codes,” IEEE Trans. On Com., Vol.37, no.9, pp.906-917, Sep.1989 .

  When performing encoding based on the look-ahead type code conversion tables of FIGS. 2 and 3, the encoding unit 11 uses the code word assigned to the input information word to perform the look-ahead of the information word. However, only code words delayed by one block can be output. In addition, when the encoding unit 11 performs encoding based on the finite state type code conversion table of FIG. 4, it is not necessary to look ahead of the information word, but the finite state type code conversion table of FIG. 2 and FIG. 3, a code word delayed by one block is assigned to an information word.

  That is, the same information sequence is encoded when encoded based on the look-ahead code conversion table of FIGS. 2 and 3 and when encoded based on the finite-state code conversion table of FIG. A code sequence can be obtained. Therefore, the look-ahead type code conversion table and the finite state type code conversion table appear to express different codes, but the only difference is how the encoding rules are expressed. However, assuming that the internal circuit configuration in the encoding unit 11 is configured in accordance with the encoding rule of the code conversion table, the encoding unit 11 performs the look-ahead type and the finite state type in the encoding process. Depending on which code conversion table is used, the circuit configuration differs.

  Further, when the code is not a variable length code but a fixed length code, there is no need to look ahead of the information sequence, so the code conversion table is only a finite state type. Whether the code represented by the code conversion table of the finite state type is a variable length code or a fixed length code depends on whether the same code word is assigned to a plurality of information words in the code conversion table, It can be distinguished by whether it is assigned to an information word.

  In addition, the design of the look-ahead code conversion table often has to rely on heuristic techniques, but the code conversion table of the finite state type uses a technique called ACH (Adler-Coppersmith-Hassner) algorithm, Can be designed theoretically.

  Here, an ACH algorithm for designing a finite state type code conversion table will be described. In the ACH algorithm, first, using the number of bits of the target codeword as a parameter, based on a finite state transition diagram of the code, a vector called an approximate eigenvector having a dimension that matches the number of states is obtained. Ask. Next, after creating a finite state table with the same number of elements in the finite state transition diagram as the number of elements in the approximate eigenvector, each operation called state splitting and state merging is performed. Is repeated several times to finally obtain a simplified finite state table.

  In the ACH algorithm, there are many approximate eigenvectors, and the finally obtained finite state table may differ depending on the value of the approximate eigenvector obtained first. However, since there are many selection branches in each operation process of state division and state fusion, even if the same approximate eigenvector is used, the finally obtained finite state table is not necessarily the same. It ’s not hard.

  A code conversion table of a finite state type can be obtained by appropriately assigning each code word in the finite state table thus obtained to a necessary information word. In addition, all codes designed according to this method have a finite codeword constraint length and can be decoded state-independently. That is, a code designed according to this method is known to be capable of sliding-block decoding.

  Details of the ACH algorithm are disclosed in Non-Patent Document 2, for example.

R. Adler, D. Coppersmith, and M. Hassner, “Algorithms for Sliding Block Codes,” IEEE Trans. On Information Theory, vol. IT-29, no. 1, pp. 5-22, Jan. 1983.

  FIG. 5 is a finite state transition diagram of 8 states giving (1, 7) RLL on the assumption of NRZI modulation. However, the state numbers in the finite state transition diagram shown in FIG. 5 do not correspond to the state numbers in the finite state type code conversion table.

  In general, [2,3,3,3,2,2,2,1] or [3,5,5,4,4] as an 8-dimensional approximate eigenvector obtained from the finite state transition diagram of FIG. Two approximate eigenvectors of 4, 4, 3, 2] are known. The value of each element of the approximate eigenvector indicates the number (so-called state splitting number) by which each state in the finite state transition diagram is first divided. For example, the five-state code conversion table shown in FIG. 4 can be obtained from the look-ahead type code conversion tables shown in FIGS. 2 and 3, and approximate eigenvectors [3, 5, 5, 4, 4, 4, 3, 2] can be theoretically obtained from the finite state transition diagram of FIG. 5 by the ACH algorithm. On the other hand, when the approximate eigenvector is [2, 3, 3, 3, 2, 2, 2, 1] and the finite state type code conversion table is obtained by the ACH algorithm, the four state finite state type code conversion table is obtained. Obtainable.

  FIG. 6 is a four-state code conversion table of the finite state type (1,7) RLL code by Weathers-Wolf. The code conversion table of FIG. 6 is known as a finite state type code conversion table having the minimum number of states in the (1,7) RLL code (for example, Non-Patent Document 3).

AD Weathers and JK Wolf, “A New Rate 2/3 Sliding Block Code for the (1,7) Runlength Constraint with the Minimal Number of Encoder States,” IEEE Trans. On Info. Theory, vol.37, no.3, pp.908-913, May 1991.

  In an optical disk recording / reproducing apparatus or an optical disk reproducing apparatus, it is generally necessary to suppress a low frequency component of a code spectrum to some extent in order to extract a servo signal used in a low frequency band. If the low-frequency component of the code spectrum is not suppressed at all, the low-frequency component of the reproduction signal in the data area is superimposed as noise on the extracted servo signal, which causes a problem that the quality of the servo signal is significantly deteriorated. appear.

  In order to suppress the low frequency component of the code spectrum, the DC component control of the code sequence is performed so that the amplitude value (DSV: digital-sum variation) of the DC accumulated charge (RDS) of the code sequence is as small as possible. (DC control) is effective. Here, the RDS is a recording code sequence or transmission code sequence after modulation, for example, if the code sequence is based on NRZI modulation, the code polarity represented by ± 1 in the code sequence after NRZI modulation is This is the cumulative value.

  Conventionally, the following three methods are mainly used as specific methods for performing DC control of code sequences.

  In the first method, first, encoding is executed regardless of the DSV of the code sequence, RDS is observed, and then the DC bit is set to the code sequence at regular intervals so as to make the DSV of the code sequence as small as possible. It is a method of inserting. However, on the assumption that the DC control bit becomes an RLL control bit depending on the case, the first encoding may be performed after the insertion of the DC control bit so that the RLL restriction of the code sequence is observed.

  Examples of the code using the first method include the EFM code that has been put into practical use in the above-described CD and MD. However, in an EFM code, if a certain information sequence continues, it becomes impossible to perform DC control in order to keep the (2, 10) RLL restriction, and the DSV may not be finite. However, since input information generally used has high randomness, this point is not a problem in practice.

  The first method has an advantage that it can be applied to any RLL code, but the number of DC control bits to be inserted in order to comply with the RLL restriction increases, and the influence of the DC control bits is included. Sign redundancy is likely to increase.

  The second method is a method in which the code conversion table is designed in advance so that the code conversion table is a finite state type and the DSV of the code sequence is as small as possible. As a code using the second method, for example, there is a (2,10) RLL code called EFMPlus that converts an 8-bit information word into a 16-bit codeword, and a recording code of a DVD (digital versatile-disk) Has been put to practical use. The EFM code described above is a fixed-length code, but the EFMPlus code is a variable-length code in which the same code word is assigned to a plurality of information words.

  However, in the EFMPlus code, of the 256 information words, there are only 95 information words that can be DC controlled, and if any of the other 161 information words is continuous, the DSV is not finite. Sometimes. However, in this case as well, as in the case of the EFM code described above, the input information to be used is generally highly random, and this is not a problem in practice.

  The second method has an advantage that it is not necessary to insert DC control bits. However, in some codes, it is difficult to design a code conversion table at a high coding rate, and in many cases, the code conversion table is used. Therefore, the configuration of the encoder and decoder tends to be complicated.

  In the third method, the code conversion table is a look-ahead type, and a code word is pre-assigned to an information word so that the two's complement of the sum of the information words and the two's complement of the sum of the code words always match, At the time of encoding, DC control bits are inserted at regular intervals in the information sequence so that the DSV of the code sequence is as small as possible. When codeword assignment is performed based on the third method, since the inversion of the code polarity that follows is guaranteed when the DC control bit to be inserted is 0 or 1, the DC control of the code sequence is performed. It can be done efficiently.

  Since the code word assignment method according to the third method is performed so that the parity of the information sequence and the parity of the code sequence always coincide, the parity preserving (PP) code word assignment (word assignment) method and Called.

  In other words, the parity-preserving codeword assignment method devised by Kahlman-Immink in 1994 is that in the look-ahead code conversion table, p is the two's complement of the total information word, and the two's complement of the total codeword is When q is assumed, a code word is assigned to an information word so that the following expression (1) is established (see, for example, Patent Document 2).

        p = q (1)

US patent, J. Kahlman; A. Immink, USP 5,477,222

  In the third method, no matter what DC control bits are inserted into the information sequence, the RLL restriction of the code sequence is not disturbed. Therefore, when the third method is used, the DC control bit is compared with the first method in which the DC control bit is inserted into the code sequence when an equivalent suppression effect is to be obtained for the low frequency component of the code spectrum. It is often possible to reduce the redundancy.

  However, in the third method, the polarity inversion of the code sequence is guaranteed by the selection of the DC control bit, but the polarity inversion of the RDS within the control interval is not necessarily guaranteed. For this reason, if certain input information in which the polarity of the RDS is not reversed regardless of whether the DC control bit is selected as 0 or 1, the DSV may not be finite. However, as in the case of the EFM code and the EFMPlus code described above, generally used input information has high randomness, so this point is not a big problem in practical use.

  Further, when a DC control method for inserting DC control bits into an information sequence is used, the code sequence can be obtained with a certain degree of probability even if parity-preserving codeword assignment is not performed in the look-ahead code conversion table used. Since the polarity of the RDS can be reversed, DC control of the code sequence is not impossible at all. In other words, the advantage of using the parity-preserving codeword assignment method is that the DC control of the code sequence is further improved by greatly increasing the probability that the polarity of the RDS of the code sequence is inverted when the DC control bit is inserted into the information sequence. It is a point that can be performed efficiently.

  In the code conversion table, a specific method for assigning parity-preserving codewords is known only for the look-ahead code conversion table. That is, no code word assignment method that is a parity-preserving type has been found for the finite state type code conversion table obtained from the ACH algorithm. Immmink, one of the inventors of the parity-preserving codeword assignment method, clearly points out this in his book (Non-Patent Document 4).

K. Immink, “Codes for Mass Data Storage,” Shannon Foundation publishers, Netherlands, p.290, 1999.

  In addition, the code represented by the first look-ahead type code conversion table that has been assigned the parity-preserving codeword assignment disclosed in Patent Document 2 described above is a (1,8) RLL code, In (1,7) RLL code, it is known that a look-ahead code conversion table in which parity-preserving codeword assignment is performed can be designed.

  7 and 8 are look-ahead type code conversion tables in which (1, 7) RLL codes are assigned parity-preserving codeword assignments.

  FIG. 7 is a basic conversion table used for encoding an information sequence at normal time, and FIG. 8 is a conversion table at violation used for encoding at the time of violation. When encoding is performed based only on the basic conversion table of FIG. 7, it becomes a (1,8) RLL code, and a (1,7) RLL code can be obtained by combining the conversion table at the time of violation in FIG. it can. In other words, in the (1,7) RLL code, when the information sequence described in the conversion table at the time of violation is supplied during the encoding process of the information sequence, the violation of FIG. The time conversion table is preferentially used. In both the basic conversion table of FIG. 7 and the conversion table at the time of violation in FIG. 8, the 2's complement of the total information word and the 2's complement of the codeword all match (for example, Patent Documents). 3).

Japanese Patent Laid-Open No. 11-346154

Further, as described in Patent Document 3, in an optical disc recording / reproducing apparatus or an optical disc reproducing apparatus, a long minimum run is a bit error rate in a situation where defocusing, tangential tilt, or the like occurs. Adversely affects. For this reason, there has been proposed a method of limiting the maximum continuous number of minimum runs to 6 by further adding a code conversion table at the time of violation to the look-ahead type code conversion tables of FIGS. However, here, the maximum number of consecutive runs of the minimum run is R, for example, when the minimum run is a code of 1 and NRZI modulation is assumed, r consecutive of 01 is expressed as (01) ^r In addition, in the code sequence of 001 (01) ^r 00, r ≦ R is always satisfied.

  FIG. 9 is a foresight-type violation in which parity-preserving codeword assignment is performed, which is added to FIGS. 7 and 8 to limit the maximum continuous number of minimum runs of (1,7) RLL code to 6. It is a code conversion table.

  Also in the conversion table at the time of violation of FIG. 9, the 2's complement of the total information word and the 2's complement of the codeword are the same. By performing encoding according to the code conversion rule in which the code conversion table in violation of FIG. 9 is added to the code conversion table of FIG. 7 and FIG. 8, the maximum number of consecutive runs of the minimum run while maintaining the parity-preserving codeword assignment. Can be limited to six.

  The (1, 7) RLL code represented by the look-ahead type code conversion table to which the parity-preserving type codeword assignment shown in FIG. 7, FIG. 8, and FIG. It is used for Blu-ray recording codes.

  As described above, the parity-preserving codeword assignment method is one of excellent methods for efficiently performing DC control of a code sequence, but a specific codeword assignment method is a look-ahead code. It is based on a conversion table and its design can only rely on heuristics. For this reason, at present, very few specific code conversion tables for which parity-preserving codeword assignment has been performed are disclosed.

  On the other hand, a finite-state code conversion table can be theoretically designed using the ACH algorithm. Therefore, if a codeword allocation method having the same effect as the parity-preserving codeword allocation method in the look-ahead type code conversion table can be found in the finite state type code conversion table, the code of the code excellent in DC control efficiency can be found. The conversion table can be designed theoretically.

  Conventionally, a technique for converting a look-ahead type code conversion table into a finite state type code conversion table in a variable length code has been disclosed. However, a finite state type code conversion table is converted to a look-ahead type code conversion table. The technology for conversion is not disclosed, and the specific method is not known. If the code conversion table of the finite state type can be converted into the code conversion table of the look-ahead type, the difference between the code word allocation method in the above-described finite state type code conversion table and the conventional parity preservation type code word allocation method Can be easily clarified.

  In addition, when the encoding unit 11 performs encoding according to the encoding rule of the look-ahead type code conversion table in which the parity-preserving codeword allocation disclosed so far is performed, the conventional encoding method is used. Compared to the case, the configuration becomes complicated.

  For example, in the look-ahead type code conversion tables of FIGS. 7 and 8, since the look-ahead amount of the necessary information sequence is 3 blocks, the code word delayed by 3 blocks is reassigned to the information word, so that the finite state type When converted to the code conversion table, the number of states becomes 104. The encoding unit 11 does not necessarily follow the encoding rules of the finite state type code conversion table. However, the fact that the number of states is large when converted into the finite state type code conversion table means that the look-ahead type code is used. Even in the conversion table, since the encoding rule is more complicated, the configuration of the encoding unit 11 is generally complicated.

  This is particularly noticeable when the maximum number of consecutive minimum runs is limited. For example, when the look-ahead code conversion table described with reference to FIGS. 7 and 8 is further combined with the look-ahead code conversion table of FIG. 9, the look-ahead amount of the necessary information sequence is 5 blocks. When the code word delayed by 5 blocks is reassigned to the information word and converted into the code conversion table of the finite state type, the number of states becomes 1691, and the number of states becomes very large.

  Further, when encoding is performed based on a look-ahead type code conversion table in which parity-preserving codeword allocation has been performed so far, the codeword constraint length at the time of decoding is, for example, FIG. 2 and FIG. There arises a problem that it becomes longer than the code word constraint length at the time of decoding when encoding is performed according to the code conversion table of FIG. Here, the codeword constraint length at the time of decoding is the number of codeword blocks that need to be retained for sliding block decoding.

  When the codeword constraint length at the time of decoding becomes long, not only the configuration of the decoding unit 16 becomes complicated, but also the error propagation length at the time of decoding becomes long, and the bit error rate after decoding deteriorates. Here, the error propagation length at the time of decoding is the maximum number of bits that a 1-bit error in the code sequence may propagate to the information sequence. This problem occurs similarly when the maximum number of consecutive runs of the minimum run is limited.

  FIG. 10 shows the number of states at the time of encoding, the code word constraint length at the time of decoding, and the error propagation length at the time of decoding when using the (1,7) RLL code described with reference to FIGS. The calculation result of is shown. However, in FIG. 10, the number of states of the code conversion table combining FIG. 7 and FIG. 8 and the code conversion table combining FIG. 7, FIG. 8 and FIG. The number of states generated when the code conversion table of the finite state type is converted by the method described above.

  As shown in FIG. 10, encoding is performed according to the encoding rule of FIG. 7 and FIG. 8 in which parity-preserving codeword assignment is performed, or the code conversion table combining FIG. 7, FIG. 8, and FIG. 7) The RLL code has 104 or 1691 states at the time of encoding, and the number of states 5 described with reference to the code conversion table of FIGS. 2 and 3 or the code conversion table of FIG. Compared to the number of states 4 described with reference to the code conversion table of FIG. Also, the (1, 7) RLL code encoded according to the encoding rule of the code conversion table combining FIG. 7 and FIG. 8 or FIG. 7, FIG. 8 and FIG. Also, the codeword constraint length at the time of decoding and the error propagation length at the time of decoding are twice as large as when the code conversion table of FIGS. 2 and 3 or the code conversion table of FIG. 4 or the code conversion table of FIG. 6 is used. It can be seen that both of them are greatly deteriorated.

  In other words, the codes represented by the look-ahead type code conversion table that has been assigned parity-preserving codewords known so far are the number of states at the time of encoding, the codeword constraint length at the time of decoding, and the time of decoding Any of these error propagation lengths is significantly deteriorated compared to codes that have not been assigned a parity-preserving codeword, so it has been desired to improve these values.

  Further, if the maximum number of consecutive minimum runs can be shortened, the bit error rate in situations such as defocusing and tangential tilt in an optical disc (recording) reproducing apparatus can be improved more effectively. However, in the (1,7) RLL code, the minimum value of the maximum continuous number of minimum runs achieved so far has been 6. A technique for making the maximum continuous number of minimum runs in the (1,7) RLL code shorter than 6 is not currently disclosed.

  The present invention has been made in view of such a situation. DC control bits for performing DC control of a code sequence are inserted into an information sequence, and an m-bit information word is continuously inserted into an n-bit code word. In the encoding method to be converted, the number of states at the time of encoding can be made smaller than before, and the code word constraint length at the time of decoding and the error propagation length at the time of decoding can be made shorter than in the conventional case. Thus, the maximum number of consecutive minimum runs in the (1,7) RLL code can be made shorter than the conventional six.

  The encoding device of the present invention includes an encoding unit that continuously converts an m-bit information word into an n-bit code word, and the encoding unit has a minimum run limit of 1 and a maximum run limit of 7. In addition, the encoding process is performed based on an encoding rule in which the maximum number of consecutive runs of the codeword is 3 to 5.

  The encoding rule used by the encoding means may be such that the number of states when expressed using a code conversion table is 8, and the maximum continuous number of minimum runs is 5.

  The encoding means can perform code conversion so that the codeword constraint length at the time of decoding is 4 blocks.

  The encoding method of the present invention is based on an encoding rule in which the minimum run limit is 1, the maximum run limit is 7, and the maximum continuous number of the minimum run of the codeword is 3 to 5, m-bit information. It includes an encoding step for continuously converting words into n-bit codewords.

  The program recorded on the recording medium of the present invention is based on an encoding rule in which the minimum run limit is 1, the maximum run limit is 7, and the maximum continuous number of codeword minimum runs is 3 to 5. , Including an encoding step of successively converting m-bit information words into n-bit code words.

  The program of the present invention converts an m-bit information word based on an encoding rule in which the minimum run restriction is 1, the maximum run restriction is 7, and the maximum number of consecutive runs of the code word is 3 to 5. A computer is caused to execute a process including an encoding step of continuously converting to an n-bit code word.

  In the encoding apparatus, the encoding method, and the program according to the present invention, the minimum run limit is 1, the maximum run limit is 7, and the maximum number of consecutive runs of the codeword is 3 to 5 The m-bit information word is continuously converted into an n-bit code word.

  According to the present invention, in addition to being able to perform code conversion, the encoding rule is such that the minimum run limit is 1, the maximum run limit is 7, and the maximum continuous number of code word minimum runs is 3 to 5. Based on this, the encoding process can be executed.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  FIG. 11 is a block diagram showing a configuration of a recording / reproducing apparatus 31 to which the present invention is applied. In addition, the same code | symbol is attached | subjected to the part corresponding to the conventional case, The description is abbreviate | omitted suitably. That is, the recording / reproducing apparatus 31 in FIG. 11 has the same configuration as that in FIG. 1 except that an encoding unit 51 is provided instead of the encoding unit 11.

  11 includes an encoding device 41 including an encoding unit 51 and a recording / reproducing unit 12, a recording / reproducing unit 12, an analog equalization circuit 4, an A / D conversion unit 14, and a code. You may make it comprise as two apparatuses with the decoding apparatus 42 comprised from the detection part 15 and the decoding part 16. FIG. Note that the decoding device 42 may acquire and decode a transmission code string encoded by the encoding device 41 and transmitted to a network (not shown).

  In the recording / reproducing apparatus 31 or the decoding apparatus 42 shown in FIG. 11, as in the conventional case described with reference to FIG. 1, if the equalization by the analog equalization unit 13 is not sufficient, the A / D A digital equalization unit may be further provided between the conversion unit 14 and the code detection unit 15. In this case, a low-pass filter may be provided in place of the analog equalization unit 13.

  Next, a process performed by the encoding unit 51 in FIG. 11 to perform DC control of a code sequence by inserting a DC control bit into an information sequence and performing encoding will be described.

  FIG. 12 is a block diagram showing a configuration of encoding section 51 in FIG. 11 that performs DC control of a code sequence by inserting DC control bits into an information sequence.

The DC control bit insertion unit 101 continuously receives an information sequence in units of bytes (8 bits). DC control bit inserting unit 101 in accordance with a preset DC control interval T _dc, information sequence d ₀ DC control bit 0 of the temporary is inserted, and the information sequence d ₁ which the control bit 1 the temporary is inserted The information sequence d ₀ is generated and supplied to the m / n converter 102 and the information sequence d ₁ is supplied to the m / n converter 103.

When DC control bits are inserted into the information sequence supplied to the DC control bit insertion unit 101 at intervals of T _dc bits, the output sequence has a DC control redundancy of 1 / (T _dc +1) with respect to the input sequence. (DC-control redundancy).

  The DC control bit insertion unit 101 further outputs a DC control signal indicating the DC control time to the m / n conversion unit 102, the m / n conversion unit 103, and the code selection signal generation unit 106.

m / n conversion unit 102, the information sequence d ₀ supplied, m: at n conversion ratio to convert the temporary code sequence c _0, the provisional code sequence c _0, by n bits, the delay unit 104 and Output to the code selection signal generator 106. Similarly, the m / n conversion unit 103 converts the supplied information sequence d ₁ into a temporary code sequence c ₁ with an m: n conversion ratio, and delays the temporary code sequence c ₁ in units of n bits. To unit 105 and code selection signal generator 106.

The delay unit 104 delays the supplied temporary code sequence c ₀ by the DC control interval and outputs the delayed code sequence c ₀ to the code selection unit 107. Similarly, the delay unit 105 delays the supplied temporary code sequence c ₁ by the DC control interval and outputs the delayed code sequence c ₁ to the code selection unit 107.

After the code selection signal generation unit 106 calculates the RDS of the provisional code sequence c ₀ and the provisional code sequence c ₁ supplied, the provisional code to be selected to make the DSV of the code sequence as small as possible is the provisional code sequence c A code selection signal indicating which temporary code sequence is ₀ or the temporary code sequence c ₁ is generated, and the m / n conversion unit 102 and the m / n conversion unit 103 are according to the timing of the input DC control signal. And to the code selection unit 107.

The code selection unit 107 receives the delayed provisional code sequence c ₀ and provisional code sequence c ₁ and reduces the DSV of the code sequence as much as possible based on the code selection signal input from the code selection signal generation unit 106. Only one of the possible temporary code sequences is selected and output as a code sequence.

Here, each of the m / n conversion unit 102 and the m / n conversion unit 103 has an internal register for storing the code state at each time, and based on the input DC control signal and code selection signal. When the provisional code sequence c ₀ is selected, the code state of the m / n conversion unit 102 is stored in the internal register of the m / n conversion unit 103. When the provisional code sequence c ₁ is selected, m / n conversion is performed. The code state of the unit 103 is copied to the internal register of the m / n conversion unit 102, respectively.

Next, the encoding process executed by the encoding unit 51 will be described with reference to the flowchart of FIG. Here, the encoding unit 51 receives an input information sequence of bytes N _i, at intervals of the bit number T _dc, by inserting DC control bits, so it is possible to perform DC control code sequence Shall be made.

In step S1, the DC control bit insertion unit 101 initializes the variable i and the variable j to 0, and the m / n conversion unit 102 substitutes the initial state s _i for the state s ₀ of the register included therein. In addition, the m / n conversion unit 103 assigns the initial state s _i to the state s ₁ of the register included therein. Here, the variable i is a variable indicating the number of processed bytes, and the variable j is a variable for counting the number of bits in order to insert DC control bits at a correct insertion interval.

  The DC control bit insertion unit 101 acquires a 1-byte information series in step S2, and sets variable i to i = i + 1 and variable j to j = j + 8 in step S3.

In step S4, the DC control bit insertion unit 101 determines whether or not the variable j satisfies j < _Tdc .

If it is determined in step S4 that the variable j does not satisfy j <T _dc , in step S5, the DC control bit insertion unit 101 inserts the DC control bit 0 into the supplied information sequence, and the information sequence d ₀ is generated and output to the m / n converter 102, and the DC control bit 1 is inserted into the supplied information sequence to generate the information sequence d _1, which is output to the m / n converter 103. Further, the variable j is set as j = j−T _dc .

If it is determined in step S4 that the variable j satisfies j <T _dc , or after the process of step S5 is completed, the m / n converter 102 and the m / n converter 103 are supplied in step S6. The encoding of the coding rate m / n is performed on the two information series d ₀ or information series d ₁ . The m / n conversion unit 102 outputs the generated provisional code sequence c ₀ to the code selection signal generation unit 106 and the delay unit 104, and the m / n conversion unit 103 performs code selection on the generated provisional code sequence c _1. The data is output to the signal generator 106 and the delay unit 105.

In step S7, the code selection signal generation unit 106 calculates DSV for each of the two provisional code sequences c ₀ and provisional code sequence c ₁ , sets the DSV of the provisional code sequence c ₀ to DSV ₀ , The DSV of the code sequence c ₁ is set as DSV ₁ .

In step S8, the code selection signal generation unit 106 compares the calculation results in step S7 and determines whether DSV ₀ > DSV ₁ is satisfied.

If it is determined in step S8 that DSV ₀ > DSV ₁ is not satisfied, the code selection signal generation unit 106 generates a code selection signal indicating that the provisional code c ₀ is selected as the output code sequence, as the m / n conversion unit 102. , The code selection unit 107 sets the output code sequence c ₂ to c ₀ , and the m / n conversion unit 103 stores the internal register in the internal register. The state s ₁ is replaced with the state s ₀ of the register inside the m / n conversion unit 102.

If it is determined in step S8 that DSV ₀ > DSV ₁ , the code selection signal generation unit 106 generates a code selection signal indicating that the provisional code c ₁ is selected as an output code sequence. , The code selection unit 107 sets the output code sequence c ₂ to c ₁ and the m / n conversion unit 102 stores the internal register in the internal register. The state s ₀ is replaced with the state s ₁ of the register inside the m / n conversion unit 103.

In step S11, the code selecting unit 107 outputs the code sequence c _2.

In step S12, DC control bit inserting section 101 compares the variable i, and a number of bytes N _i supplied, the variable i is judged whether to satisfy the i <N _i. If it is determined in step S12 that the variable i satisfies i <N _i , the process returns to step S2 and the subsequent processes are repeated. If it is determined in step S12 that the variable i does not satisfy i <N _i , the process is terminated.

By such processing, the temporary DC control bit 0 is inserted at a bit interval of T _dc bits, and the encoded temporary code sequence c ₀ and the temporary DC control bit 1 are inserted and encoded. The code sequence c ₁ is generated, and one of the temporary code sequences is selected and output as the code sequence c _{2 based on} the DSV calculation result, so that the DC control of the code sequence can be performed.

  As for the code conversion table used in the encoding process executed in step S6 in FIG. 13, as described above, the method for allocating the parity-preserving codeword is known only for the a priori code conversion table. It is not designed, and its design depends only on heuristics. However, if a codeword allocation method having the same effect as the parity-preserving codeword allocation method can be obtained in a finite state type code conversion table, a theoretical code conversion table using the ACH algorithm can be designed. Therefore, a code superior to the conventional code can be obtained.

  Next, as executed in the encoding unit 51 described with reference to FIG. 12, DC control bits are inserted into the information series at predetermined bit intervals, and information is obtained according to the encoding rule of the finite state type code conversion table. A characteristic required for a code conversion table that can efficiently perform DC control of a code sequence when the sequence is encoded will be described.

As described with reference to FIGS. 12 and 13, for example, an information sequence d ₀ and a temporary DC control bit 1 are inserted at a specific position such as the head of the information sequence, with the temporary DC control bit 0 inserted. When the information sequence is the information sequence d ₁ , these two information sequences d ₀ and information sequence d ₁ are encoded according to the encoding rules of the finite state type code conversion table starting from the same state. Are the temporary code sequence c ₀ and the temporary code sequence c ₁ , respectively. In this case, the code states at the time of encoding stored in the registers inside the m / n conversion unit 102 and the m / n conversion unit 103 are: State s ₀ and state s ₁ .

For example, when encoding is performed using the code conversion table described with reference to FIG. 6, when 7 bits of [1, 1, 0, 0, 0, 1, 0] are input as an information sequence, Information sequence d ₀ and information sequence d ₁ with provisional DC control bits inserted at the beginning thereof, provisional code sequence c ₀ and provisional code sequence c ₁ encoded using the code conversion table of FIG. Also, the code state s ₀ and the code state s ₁ at the time of encoding are, as shown in FIG. 14, an information sequence d ₀ : 01,10,00,10, a code sequence c ₀ : 100,101,000,010. , Code state s ₀ : 2,3,1,3, information sequence d ₁ : 11,10,00,10, code sequence c ₁ : 101,010,000,010, code state s ₁ : 4,3,1 , 3.

In the encoding shown in FIG. 14, the state of each provisional code sequence is different from the state s ₀ and the state s ₁ even if the initial state is the same because the first input information word is different by 1 bit. State, s ₀ = 2 and s ₁ = 4. However, after that, the state of each code sequence is the same at s ₀ = s ₁ = 3 in the second block, and once the state becomes the same, each temporary code sequence in the subsequent third block and thereafter The state is the same.

Thus, any information sequence, an information sequence d ₀ was inserted DC control bit 0 of the temporary, and the information sequence d ₁ of inserting the DC control bit 1 of the temporary, provisional code sequence c ₀ obtained by respectively encoding Even if the initial code state c ₁ and the state s ₀ and state s ₁ are the same, the first input information word (that is, the temporary DC control bit) is different by 1 bit. If this is the case, it will be in a different state. However, in the subsequent conversion, when the state s ₀ and the state s ₁ become the same state, the corresponding temporary code sequence and the state are the same in each of the two systems regardless of the information sequence that follows.

Therefore, if the polarities after the NRZI modulation of the temporary code sequence c ₀ and the temporary code sequence c ₁ are inverted at the time when the state s ₀ and the state s ₁ are the same, the subsequent temporary code sequence c ₀ and Since the sign of the RDS of the temporary code sequence c ₁ is always reversed, either the temporary code sequence c ₀ or the temporary code sequence c ₁ is selected as the final code sequence so as to make the DSV of the code sequence as small as possible. Then, the DC control of the code sequence can be performed efficiently.

Here, at the time when the state s ₀ and state s ₁ of the provisional code sequence c ₀ and the temporary code sequence c ₁ becomes equal, reversing the polarity of the NRZI-modulated provisional code sequence c ₀ and the temporary code sequence c ₁ Therefore, the value of the parity q ₀ that is the two's complement of the total of the provisional code sequence c ₀ at the time when the states of the provisional code sequence become the same and the two's complement of the sum of the provisional code sequence c ₁ are used. It suffices if the value of the parity q ₁ is different.

That is, when encoding an information sequence based on a finite state type code conversion table, in order to efficiently perform DC control of the code sequence by inserting DC control bits into the information sequence at regular intervals, A temporary code sequence c ₀ that has been temporarily encoded from an arbitrary start point state, and a temporary code sequence that has been temporarily encoded from the same start point state. In the code sequence c ₁ , it is sufficient that the parity q ₀ and the parity q ₁ at the time when the state s ₀ and the state s ₁ become the same always satisfy the following equation (2).

q ₀ ≠ q ₁ (2)

However, after the state s ₀ and the state s ₁ becomes equal, since the provisional code sequence c ₀ and the temporary code sequence c ₁ is the same formula when the state s ₀ and the state s ₁ becomes identical ( If 2) holds, the parity q ₀ and the parity q ₁ of the provisional code sequence including the provisional code sequence after the state s ₀ and the state s ₁ are the same satisfy Expression (2).

In the case of FIG. 14 described above, the total number of [ ₁ , ₀ , ₀ , ₁ , ₀ , 1], which is the code sequence c ₀ up to the second block in which the state s ₀ and the state s ₁ are the same, is 2. the complement q ₀ 1, Similarly, a code sequence c ₁ until the second block [1,0,1,0,1,0] 2's complement q ₁ of a total of from 1, In this case, Formula (2) is not materialized. That is, when the insertion position of the DC control bit is the first bit of the information word, it can be seen that Expression (2) does not hold for an arbitrary information sequence in the code conversion table of FIG.

  Further, in the code conversion table of FIG. 6, even when the insertion position of the DC control bit is the second bit of the information word, the expression (2) does not hold for a certain information series in the same manner.

  Here, a case has been described in which Expression (2) does not hold for a single information series. However, if Expression (2) holds for a certain information series, Expression (2) also applies to other arbitrary information series. ) Is necessary to check whether or not.

That is a specific position bits (e.g. first bit) and information sequence d ₀ of the block length L to 0, the information sequence d ₁ Metropolitan to 1 the bit of a specific location, these two information sequences, start state In the temporary code sequence c ₀ and the temporary code sequence c ₁ obtained by fixing and encoding each temporary code sequence until the state of each temporary code sequence becomes the same first except for the start point The two two's complements q ₀ and q ₁ are obtained, and it is checked whether or not they are different for all information sequences of L blocks that can be generated and all start point states.

  However, when the state becomes the same with the number of blocks shorter than L, the information series until the state becomes the same is the same, and for an information series longer than that, it is checked whether or not the equation (2) is satisfied. There is no need. Further, since it is impossible to make the observed block length L infinite in order to make the length of the information sequence arbitrary, it is assumed that L has a certain finite length. Furthermore, if the states are not the same within the range of the observed block length L, this is ignored.

  However, in a certain code conversion table, if Equation (2) holds for an arbitrary information sequence having a certain finite block length L, Equation (2) also applies to an arbitrary information sequence having an infinite block length. Whether or not is satisfied will be described later with reference to FIG.

For example, in the four-state code conversion table of FIG. 6, when the code word assignment to each information word is changed, the result of verifying whether or not there is a code word assignment satisfying the expression (2) for an arbitrary information sequence When the length of the information sequence L including the DC control bits is 4 blocks or more, regardless of whether the insertion position of the DC control bits is the first bit or the second bit of the information word, Also in the case of (4!) ⁴ = 331, 776 in the ^four -state code conversion table of FIG. 6, codewords satisfying Expression (2) for any information sequence among all codeword assignments There were no assignments.

Similarly, in the five-state code conversion table of FIG. 4, when the code word assignment to each information word is changed, whether or not there is a code word assignment satisfying Expression (2) for an arbitrary information sequence As a result of verification, when the length of the information sequence L including the DC control bits is 5 blocks or more, the insertion position of the DC control bits is the first bit or the second bit of the information word. In any case, among the (4!) ⁵ = 7,962,624 codeword assignments in the ^five -state code conversion table of FIG. 4, a codeword assignment satisfying equation (2) for an arbitrary information sequence is There was no one.

  As described above, in the code conversion table of the conventionally known 4-state or 5-state finite state type (1, 7) RLL code, even if the codeword assignment is changed, an expression for an arbitrary information sequence is obtained. There was no codeword assignment satisfying (2).

  On the other hand, in the look-ahead code conversion tables of FIGS. 7 and 8, 104 state finite state code conversion tables in which each 3-bit codeword is delayed by 3 blocks and reassigned to 2-bit information words, and In the look-ahead code conversion tables of FIGS. 7, 8, and 9, a 1691-state finite state code conversion table in which each 3-bit codeword is delayed by 5 blocks and reassigned to a 2-bit information word. In any case, the formula (2) was satisfied for any information series. However, first, after designing a finite-state code conversion table with as few states as possible and finding a codeword assignment that satisfies Equation (2) for an arbitrary information sequence, parity-preserving codeword assignment is performed. It is very difficult to reduce the number of states than the code in the case of breakage.

  For this reason, by designing a new finite state table by the ACH algorithm using an approximate eigenvector different from the conventionally used approximate eigenvector, arbitrary information can be obtained in a finite state type code conversion table with a relatively small number of states. It was made possible to obtain a codeword assignment for which the formula (2) holds for a sequence.

  That is, an approximate eigenvector [4, 6, 6, 6, 4, 4, 4, 2] was newly derived as an 8-dimensional approximate eigenvector obtained from the finite state transition diagram of FIG. This approximate eigenvector is obtained by doubling each element of the conventionally used approximate eigenvector [2, 3, 3, 3, 2, 2, 2, 1]. Based on this new approximate eigenvector, a new seven-state finite state table was obtained using the ACH algorithm.

FIG. 15 shows a seven-state finite state that gives the (1,7) RLL restriction, obtained using the ACH algorithm based on the approximate eigenvector [4, 6, 6, 6, 4, 4, 4, 2]. It is a table. In the 7-state finite state table of FIG. 15, codeword assignment to information words has not been performed yet. In order to obtain the code conversion table from the finite state table of FIG. 15, it is necessary to assign the 3-bit codeword of each state to four types of 2-bit information words of 00, 01, 10, and 11, respectively. There are a total of (4!) ⁷ = 4,586,471,424 such codeword assignments in the finite state table of FIG.

  Next, in the finite state table of FIG. 15, processing for determining whether or not there is a codeword assignment that satisfies Equation (2) for an arbitrary information sequence when codeword assignment to an information word is performed will be described. To do.

  In a finite state table, a process for searching for a codeword assignment satisfying Expression (2) for an arbitrary information sequence regardless of the observed block length L is executed using, for example, a personal computer shown in FIG. Can do.

  FIG. 16 illustrates a configuration example of the personal computer 121. The CPU (Central Processing Unit) 131 of the personal computer 121 has various programs according to a program stored in a ROM (Read Only Memory) 132 or a program loaded from a hard disk (HDD) 138 to a RAM (Random Access Memory) 133. Execute the process. The RAM 133 also appropriately stores data necessary for the CPU 131 to execute various processes.

  The CPU 131, the ROM 132, and the RAM 133 are connected to each other via an internal bus 134. An input / output interface 135 is also connected to the internal bus 134.

  The input / output interface 135 includes an input unit 137 including a keyboard and a mouse, a CRT (Cathode Ray Tube), an LCD (Liquid Crystal Display), a display for displaying images and text, a speaker for outputting sound, and the like. An output unit 136, an HDD 138 for recording / reproducing information, a network interface 140 including a modem, a terminal adapter, and the like are connected. The network interface 140 performs communication processing via a network such as the Internet, for example.

  A drive 139 is connected to the input / output interface 135 as necessary, and a magnetic disk 141, an optical disk 142, a magneto-optical disk 143, a semiconductor memory 144, or the like is appropriately mounted, and a computer program read from them is Installed in the HDD 138 as necessary.

  A code that satisfies Equation (2) for an arbitrary information sequence in a certain finite state table executed by the personal computer 121 described with reference to FIG. Processing for searching for word assignment will be described.

  In step S <b> 31, the CPU 131 acquires a finite state table used for the search using, for example, an ACH algorithm based on the user operation input from the input unit 137 and information stored in the HDD 138 in advance. For example, the CPU 131 obtains an eight-dimensional approximate eigenvector [4, 6, 6, 6, 4, 4, 4, 2] from the finite state transition diagram of FIG. 5 and uses the ACH algorithm to (1, 7). The 7-state finite state table of FIG. 15, which is a 7-state finite state table giving the RLL restriction, is obtained.

In step S <b> 32, the CPU 131 reads the variable L and the variable N ₀ temporarily stored in the RAM 133, initializes them by substituting 0 respectively, and stores them again in the RAM 133. Here, the variable L is the observed block length, and the variable N ₀ is the number of codeword assignments that satisfy Equation (2) for an arbitrary information sequence of the variable L indicating the block length.

In step S33, the CPU 131 increments the value of the variable L temporarily stored in the RAM 133 by 1 to L = L + 1, sets the variable M to M = ₀ , substitutes N ₀ for the variable N ₁ , and N ₁ = N _0, and after, by substituting 0 to N _0, save to again RAM133. Here, the variable N ₁ is the previously calculated value of N ₀ , and the variable M is the number of codeword assignment searches.

  In step S34, the CPU 131 increments the value of the variable M temporarily stored in the RAM 133 by 1 and stores it in the RAM 133 again as M = M + 1, and changes and updates the codeword assignment. For example, when the finite state table of FIG. 15 is acquired in step S31, allocation of 3-bit code words to four types of 2-bit information words, that is, code word allocation has not been searched so far in step S34. Changed to new codeword assignment and updated. However, codeword assignment at the time of the first search may be arbitrary, and a new codeword assignment that has not been searched so far may be used after the second time.

In step S35, as described with reference to FIG. 12 to FIG. 14, the CPU 131 sets the equation (2) when the state s ₀ and the state s ₁ match in the temporary code sequence c ₀ and the temporary code sequence c ₁ . Whether or not there is at least one information sequence having a block length L satisfying q ₀ = q ₁ is determined. If it is determined in step S35 that there is no information sequence of block length L where q ₀ = q ₁ , the process proceeds to step S36. If it is determined in step S35 that there is an information sequence having a block length L such that q ₀ = q ₁ , the process proceeds to step S37.

  However, in step S35, if there is no information sequence in which the states of the two temporary code sequences match, the process may proceed to step S36 or to step S37. In this case, the process proceeds to step S37. However, in this case, the process is executed on the assumption that at least one of all the searched codeword assignments has at least one information sequence in which the states of the two temporary code sequences match.

If it is determined in step S35 that there is no block length L information sequence satisfying q ₀ = q ₁ , in step S36, the CPU 131 increments the value of the variable N ₀ by 1 and N ₀ = N _The code conversion table using the codeword assignment at that time, or information necessary and sufficient to express the code conversion table using the codeword assignment at that time is stored in the RAM 133.

In step S37, CPU 131 includes a variable M which is temporarily stored in the RAM 133, reads out the total number M _p codewords assigned comparison, it is determined whether or not M <M _p. The total number M _p of codeword assignments is, for example, (4!) ⁷ = 4,586,471,424 in the case of the finite state table described with reference to FIG. In step S37, if it is determined that M <M _p, the process returns to step S34, and the subsequent processes are repeated.

If it is determined in step S37 that M <M _p is not satisfied, in step S38, the CPU 131 reads the variables N ₁ and N ₀ temporarily stored in the RAM 133, and N ₀ ≠ N ₁ and N ₀ ≠ 0. Judge whether there is. If it is determined in step S38 that N ₀ ≠ N ₁ and N ₀ ≠ 0, the process returns to step S33, and the subsequent processes are repeated. If it is determined in step S38 that N ₀ ≠ N ₁ and N ₀ ≠ 0, the process ends.

When the execution of the search process described with reference to FIG. 17 is completed, the code conversion table stored by the search using the last L is represented by the formula (2) for an arbitrary information series regardless of the block length L to be observed. All codeword assignments satisfying N0, and N ₀ represents the number. That is, if N ₀ = 0, there is no single codeword assignment that satisfies Equation (2) for an arbitrary information sequence in the finite state table.

Next, according to the flowchart of FIG. 17, the block length L of the information sequence including the DC control bits is changed for all codeword assignments (4!) ⁷ = 4,586,471,424 in the 7-state finite state table of FIG. In this case, the number N (L) of codeword assignments for which Equation (2) holds for an arbitrary information sequence was obtained.

  However, since the finite state table of FIG. 15 is created from the ACH algorithm, no matter what codeword assignment is used in the finite state table of FIG. 15, the code encoded according to the code conversion table after the codeword assignment. Can be subjected to sliding block decoding, but there are four block code lengths and three block code word constraint lengths at the time of decoding. There are 382,205,952 codeword assignments corresponding to 1/12 of the total codeword constraint length when decoding is 3 blocks.

  FIG. 18 shows the number N (L) of codeword assignments satisfying equation (2) for an arbitrary information sequence of block length L when the observed block length L is changed in the finite state table of FIG. The code word constraint length at the time of decoding is classified into 4 blocks and 3 blocks. Further, even if the insertion position of the DC control bit is fixed to the first bit or the second bit of the information word, the calculation of the number N (L) of codeword assignments satisfying Expression (2) The result is the same.

  That is, for constraint length 4, when block length L is 1, 955,514,880 codeword assignments satisfy Equation (2). Hereinafter, as the number of blocks increases, codeword assignment satisfying Equation (2) is satisfied. When the number N (L) decreases and the block length L is 6, the number is 327,680. Even when the block length L is 7, the number N (L) of codeword assignments is the same as when the block length L is 6, and is 327,680.

  On the other hand, in the constraint length 3, when the block length L is 1, 63,700,992 codeword assignments satisfy Expression (2), and when the block length L is 2, 68,950,144 codeword assignments satisfy Expression (2 ) Is satisfied. Hereinafter, as the number of blocks increases, the number N (L) of codeword assignments that satisfy Equation (2) decreases. When the block length L is 6, the number is 65,536. Even when the block length L is 7, the number N (L) of codeword assignments is the same as when the block length L is 6, and is 65,536.

  That is, as shown in FIG. 18, the number N (L) of codeword assignments satisfying the expression (2) is such that the block length L is 6 except for the case where the block length L increases from 1 to 2 in the constraint length 3. In the following cases, it decreases as the block length increases, but does not decrease when the block length L increases from 6 to 7.

  In addition, the allocation of 393,216 code words (constraint length 4 and constraint length 3) satisfying equation (2) with an arbitrary information sequence when the block length L is 6 increases the block length L to 7 or 8. Even in such a case, Expression (2) is satisfied for an arbitrary information series. Furthermore, codeword assignments obtained by randomly extracting some of the 393,216 codeword assignments can be expressed by the formula (2) for an arbitrary information sequence even when the block length L is increased to 12, for example. )

  This is because if the block length L is increased to some extent, all two temporary code sequences whose code states do not match each other become a repetitive pattern. Therefore, even if the block length L is set to infinity (actually, the block length L cannot be set to infinity), the 393,216 codeword assignments can be expressed by the formula ( 2) is satisfied.

  That is, when the number N (L) of codeword assignments satisfying equation (2) is calculated by changing the observed block length L, the codeword assignments found when the following equation (3) holds: The number N (L) is the number of codeword assignments that satisfy Expression (2) for an arbitrary information sequence regardless of the observed block length L.

    N (L) = N (L-1) (3)

In FIG. 18, when the codeword constraint length at the time of decoding is 3 blocks, the number N (L) of codeword allocations when the block length L is 2 than when the observed block length L is 1 The value of is small. This, when the extremely short length of the input information sequence, the code word assignment number of cases where the state s ₀ of the provisional code sequence c ₀ and state s ₁ of the provisional code sequence c ₁ match becomes zero This is because it is assumed that such a codeword assignment does not hold Equation (2).

  As described with reference to FIG. 18, in the finite state table of FIG. 15, when the insertion position of the DC control bit is fixed to either the first bit or the second bit of the information word, the block length to be observed Regardless, there are a total of 393,216 codeword assignments that satisfy Equation (2) for an arbitrary information sequence. However, as described above, even if the insertion position of the DC control bit is fixed to the first bit or the second bit of the information word, Expression (2) is obtained for an arbitrary information sequence. Although the number of codeword assignments to satisfy is the same, the codeword assignments are not necessarily the same.

  The number of codeword assignments satisfying equation (2) for an arbitrary information sequence is the same as that described above regardless of whether the insertion position of the DC control bit is inserted in either the first bit or the second bit of the information word. Of the 393,216 streets, there are 196,608 streets, corresponding to exactly half.

  Therefore, when the insertion position of the DC control bit is inserted into at least one of the first bit and the second bit of the information word, the number of codeword assignments satisfying Expression (2) for an arbitrary information sequence is There are 3 × 196,608 = 589,824 ways. These 589,824 are the total number of codeword assignments that satisfy Equation (2) regardless of the observed block length in the finite state table of FIG. This number of codeword assignments is 1/7776 of the total number of codeword assignments in the finite state table of FIG.

  Here, in the finite state table of FIG. 15, 589,824 codeword assignments that satisfy Equation (2) are assigned to DC control bits when Equation (2) is satisfied for an arbitrary information sequence, regardless of the observed block length. It is classified into three types according to the insertion position. The three types are “type 1 of code word assignment that satisfies the expression (2) for an arbitrary information sequence only when a DC control bit is inserted in the first bit of the information word”, Only when a DC control bit is inserted into the second bit, type 2 which is “codeword assignment satisfying equation (2) for an arbitrary information sequence”, “first bit or second bit of information word” Even if a DC control bit is inserted at the position of, the codeword allocation that satisfies the expression (2) for an arbitrary information sequence is type 3.

  FIG. 19 shows the number of codeword assignments of type 1 to type 3 when the codeword constraint length at the time of decoding is 3 blocks and when it is 4 blocks. As shown in FIG. 19, the number of codeword assignments in the case of a constraint length of 4 blocks is five times that in the case of a constraint length of 3 blocks, and for Type 1 to Type 3, the number is the same.

  Next, with respect to the code conversion table when specific codeword assignment is performed on the finite state table of FIG. 15, three codeword assignments that are relatively similar to each other are shown in FIG. 20 to FIG. The code conversion table will be described.

  The code conversion table of FIG. 20 is a first example of a finite state type 7-state code conversion table of a (1,7) RLL code.

  The codeword assignment in the code conversion table of FIG. 20 is that the codeword constraint length at the time of decoding is 3 blocks, and the insertion position of the DC control bit is set to either the first bit or the second bit of the information word. Even if it is inserted, the codeword assignment satisfying the equation (2) for an arbitrary information sequence, and this codeword assignment is one of 32,768 codeword assignments of type 3 shown in FIG.

  The code conversion table of FIG. 21 is a second example of a finite state type 7-state code conversion table of the (1, 7) RLL code.

  The codeword assignment of the code conversion table of FIG. 21 is similar to the code conversion table of FIG. 20, the codeword constraint length at the time of decoding is 3 blocks, and the insertion position of the DC control bit is set to the first bit of the information word. And any position of the second bit, the code word assignment satisfying the expression (2) for any information sequence is performed in 32,768 types 3 of type 3 shown in FIG. One of the codeword assignments. The code conversion table of FIG. 21 has features different from the code conversion table of FIG. 20, and this point will be described later.

  The code conversion table of FIG. 22 is a third example of the finite state type 7-state code conversion table of the (1, 7) RLL code.

  The codeword assignment of the code conversion table of FIG. 22 is arbitrary information only when the codeword constraint length at the time of decoding is 3 blocks and the insertion position of the DC control bit is inserted in the first bit of the information word. This is a codeword assignment that satisfies equation (2) for the sequence, and this codeword assignment is one of the 32,768 type 1 codeword assignments shown in FIG. The code conversion table of FIG. 22 has different features from the code conversion tables shown in FIGS. 20 and 21, and this point will be described later.

  Here, as a specific example in which the code conversion table of FIG. 21 satisfies Expression (2) using FIG. 23, the 7 bits of the information sequence [1, 1, 0, 0, 0, 1, 0] are The case where it is input will be described.

When 7 bits of [ ₁ , ₁ , _{0, 0, 0} , ₁ , ₀ ] are supplied as an information sequence, an information sequence d ₀ and an information sequence d ₁ in which a temporary DC control bit is inserted at the head thereof, state 3 23, the temporary code sequence c ₀ and the temporary code sequence c ₁ encoded using the code conversion table of FIG. 21, and the code state s ₀ and the code state s ₁ at the time of encoding are shown in FIG. Information sequence d ₀ : 01,10,00,10, code sequence c ₀ : 000,100,001,001, code state s ₀ : 1,6,6,3, information sequence d ₁ : 11, 10, 00, 10, code sequence c ₁ : 000, 010, 101, 001, code state s ₁ : 4, 7, 6, 3.

In this example, the code state s ₀ and code states s ₁ of the provisional code sequence c ₀ and the temporary code sequence c ₁ in third block is, s has become the same in ₀ = s ₁ = 6, the code state the same The code sequence c ₀ = [0, 0, 0, _1, 0, 0, 0, 0, ₁ ] up to the point when the two's complement q ₀ is 0, and the code sequence c ₁ = [0, 0, [0, 0, 1, 0, 1, 0, 1], the total two's complement q ₁ is 1. Therefore, in this case, Expression (2) is established.

  Further, in the code conversion using the code conversion table of FIG. 21, when a code sequence different from 7 bits of [1, 1, 0, 0, 0, 1, 0] is converted, Even when the insertion position is the second bit of the information word, the expression (2) is similarly established when the above-described two provisional code sequences match.

  The finite-state code conversion tables of FIGS. 20 to 22 have different characteristics. However, from the finite-state code conversion tables of FIGS.

  Next, processing for converting the finite state code conversion tables of FIGS. 20 to 22 into the look-ahead code conversion table, the characteristics of each look-ahead code conversion table, and the differences between them will be described.

  As described above, converting the look-ahead type code conversion table into the finite state type code conversion table has been conventionally performed, but there is no example in which the reverse conversion is performed. To convert a finite-state code conversion table into a look-ahead code conversion table, code words advanced by the amount of look-ahead required at the time of encoding can be reassigned to information words. Since it is not always easy to know the necessary amount of look-ahead from the code conversion table, the process of converting a finite state type code conversion table to a look-ahead type code conversion table is not possible. Compared with the process of converting the code conversion table of the type into the code conversion table of the finite state type, it is difficult.

  For example, when a finite-state code conversion table is designed by the ACH algorithm, it is generally known that the maximum value of the foreseeing amount is a value obtained by subtracting 1 from the maximum value of each element of the approximate eigenvector used. It has been. However, since this value is the maximum value of the foreseeing amount, it does not necessarily match the minimum necessary foreseeing amount.

Here, in the variable-length code which has been used to encode the past, when the look-ahead amount necessary for decoding processing was L _d block, look-ahead amount required for the encoding process, when L _d is even, L _{When d} / 2 and L _d are odd, (L _d −1) / 2. The amount of look-ahead necessary in the decoding process can be known when examining the codeword constraint length during the decoding process based on the code conversion table of the finite state type.

  For example, in codes encoded according to the code conversion tables of FIGS. 20 to 22, the codeword constraint length at the time of decoding is 3 blocks including the current time, and the amount of look-ahead required at the time of decoding is 2 blocks. Therefore, it can be estimated that the amount of look-ahead necessary for encoding is one block.

  By utilizing this fact, in the code conversion tables of FIGS. 20 to 22, the code words advanced by one block are reassigned to the information words, so that these finite state type code conversion tables are converted into the look-ahead type code conversion tables. Can be converted to

  For example, in FIG. 21, the code word 010 generated in the state 4 is always 11 because the code word advanced by one block is reassigned to the information word. Is assigned to the information word 11.

  FIG. 24 is a look-ahead code conversion table of the (1,7) RLL code converted from the code conversion table of the finite state type of FIG. 20 by reassigning the code word advanced by one block to the information word. is there.

  FIG. 25 is a look-ahead code conversion table of the (1,7) RLL code converted from the code conversion table of the finite state type of FIG. 21 by reassigning the code word advanced by one block to the information word. is there.

  FIG. 26 is a look-ahead code conversion table of the (1,7) RLL code converted from the code conversion table of the finite state type of FIG. 22 by reassigning the code word advanced by one block to the information word. is there.

  In the look-ahead code conversion tables of FIGS. 24 to 26, the first column indicates the supplied information bits, and the second column indicates the given codes. Further, (“0”) indicates a case where the immediately preceding sign bit is 0, and (“1”) indicates a case where the immediately preceding sign bit is 1.

  Next, each feature of the look-ahead code conversion tables of FIGS. 24 to 26 will be described.

  The look-ahead code conversion table of FIG. 24 is a parity-preserving codeword assignment expressed by equation (1) in which the parity of the information sequence and the parity of the code sequence always match. However, the (1, 7) RLL code represented by FIG. 24 has a code number of 104 to 7 at the time of encoding and a codeword at the time of decoding, compared with the codes represented by the conventional FIG. 7 and FIG. The constraint length is greatly improved from 5 blocks to 3 blocks, and the error propagation length during decoding is from 10 bits to 6 bits. Codeword assignments having characteristics similar to those of the code conversion table shown in FIG. 24 are 1/2 of the number of type 3 codeword assignments shown in FIG.

  The look-ahead code conversion in FIG. 25 is such that when the number of blocks is odd, the parity of the information sequence and the parity of the code sequence are different from each other and when the number of blocks is even, Codeword assignment is performed so that the two match, and clearly, a codeword assignment different from the parity-preserving codeword assignment expressed by the equation (1) is performed.

  However, in the look-ahead code conversion table, the code of codeword assignment as shown in FIG. 25 can be subjected to DC control equivalent to the code of parity preservation type codeword assignment described with reference to FIG. This is because, when the codeword assignment as shown in FIG. 25 is used, polarity inversion in the code sequence at the time of inserting the DC control bit is guaranteed regardless of whether the number of blocks is odd or even. The codeword assignment having the same characteristics as the code conversion table shown in FIG. 25 is 1/2 of the type 3 codeword assignment shown in FIG.

  The look-ahead code conversion table in FIG. 26 is clearly different from the parity-preserving codeword assignment, and further different from the codeword assignment as shown in FIG. The code conversion table of FIG. 26 includes the information word 01 or 11 so that the parity of the information sequence is different from the parity of the code sequence for the information sequence including the odd word information block 01 or 11, and includes the even word block of the information word 01 or 11 Code words are assigned to the information sequence so that the parity of the information sequence matches the parity of the code sequence.

  It is difficult to judge at a glance whether or not DC control can be performed efficiently when the code conversion table of FIG. 26 is used. However, even in a code having codeword assignment as shown in FIG. By fixing the insertion position of the control bits to the information word at the first bit, DC control equivalent to the code of codeword assignment described with reference to FIGS. 24 and 25 is possible.

  As described above, in the look-ahead code conversion table, in order to satisfy the condition of Expression (1), the parity of the information sequence and the parity of the code sequence need to match. On the other hand, in order to satisfy the condition of the expression (2) for the code conversion table of the finite state type, the parity of the information sequence and the parity of the code sequence do not necessarily match, and two temporary codes It suffices if the sequence parity is different. That is, the condition of Expression (2) is a condition including the condition of Expression (1).

  That is, the codeword assignment of the finite state type code conversion table searched in accordance with the flowchart of FIG. 17 is not limited to satisfying the condition of Expression (1), but also when converted into the look-ahead type code conversion table. Many codeword assignments different from the conventional parity-preserving codeword assignment described with reference to FIG. 25 and FIG. 26 are included.

  In the following, in a finite state type code conversion table, for an arbitrary information sequence, when two temporary code sequences are in the same state, a codeword assignment that always satisfies equation (2) is expressed as “parity-differential code”. It is referred to as “word allocation”, and such a code word allocation method is referred to as “parity-differential code word allocation method”.

  Next, the encoding unit 51 described with reference to FIGS. 11 to 13 performs the information sequence according to the finite-state code conversion tables described with reference to FIGS. 4 and 6 and FIGS. FIG. 27 and FIG. 28 will be used to describe the effect of suppressing the code spectrum in the low frequency band when the above coding is performed and the DC control redundancy is changed.

Specifically, as an indicator of the suppression effect of the low frequency spectrum of the code, when the channel clock was f _c, the power spectral density of f _c / 1024, it shall be used as a representative value. In addition, an operation of performing discrete Fourier transform on the code sequence of 4096 bits after NRZI modulation to obtain a spectrum component of f _c / 1024 was repeated 10,000 times for different random information sequences, and their root mean squares were measured.

FIG. 27 shows a case where encoding is executed according to the code conversion table described with reference to FIGS. 4 and 6 and FIGS. 20 to 22 with the DC control bit insertion position as the first bit of the information word. , F _c / 1024 and the DC spectral redundancy dependency of the power spectral density. FIG. 28 shows the insertion position of the DC control bit as the second bit of the information word, and FIGS. FIG. 23 shows DC control redundancy dependency of power spectrum density of f _c / 1024 when encoding is performed according to the code conversion table of FIG.

  As shown in FIG. 27, when the encoding process is executed based on the code conversion tables of FIGS. 20 to 22 with the insertion position of the DC control bit as the first bit of the information word, the DC control redundancy is 0.007. In the wide range from 0.12 to 0.12, the suppression effect of the low frequency spectrum can be improved as compared with the case of encoding based on the conventional finite state type code conversion table.

  Further, as shown in FIG. 28, when encoding processing is performed based on the code conversion tables of FIGS. 20 and 21 with the insertion position of the DC control bit as the second bit of the information word, the DC control redundancy is 0.007. In a wide range from 0.08 to 0.08, the suppression effect of the low frequency spectrum can be improved as compared with the case of encoding based on the conventional finite state type code conversion table.

  Further, as shown in FIGS. 27 and 28, when the code conversion table of FIG. 22 is used, when the insertion position of the DC control bit is the first bit of the information word, the suppression effect of the high low frequency spectrum is obtained. However, when the second bit of the information word is used, the effect of suppressing the low frequency spectrum obtained is lower than when the conventional 4-state or 5-state code conversion table is used. This is because, in the code conversion table of FIG. 22, codeword assignment is performed so that an arbitrary information sequence satisfies Expression (2) only when the insertion position of the DC control bit is the first bit of the information word. This is because.

  However, in any code conversion table, the code word assigned to information word 01 and the code word assigned to information word 10 are all interchanged with each other, or the code word assigned to information word 00 and the information By replacing all the code words assigned to the word 11 with each other, the DC control bit insertion position is the first bit of the information word and the second bit of the information word. Can be codeword assignments that are completely replaced.

  That is, in the encoding process to which the present invention is applied, it is not so important whether the insertion position of the DC control bit is the first bit or the second bit of the information word. Whether to select a codeword assignment that can increase the suppression effect in the low frequency band according to the insertion position, or whether to use the DC control bit insertion position as the first bit of the information word according to the codeword assignment used What is necessary is just to determine whether it is set as the 2nd bit.

From FIG. 27, for example, when encoding is performed according to the 7-state code conversion table of FIG. 21, compared to the case of encoding according to the 4-state code conversion table of FIG. 6, −8.5 dB f _c / 1024. In order to obtain the power spectral density at 1, the redundancy of the DC control bits can be greatly improved from 1/22 = 0.04545 to 1/40 = 0.025. However, to what extent the power spectral density at f _c / 1024 needs to be suppressed in an actual system varies depending on the system to be used, but is generally −6 dB or less. In addition, the suppression effect of the low frequency band of the code spectrum varies depending on the DC control method.

  As described above, when the DC control bits are inserted into the information series, it is possible to efficiently perform the DC control of the code series by performing encoding according to the finite state type code conversion tables of FIGS.

That is, the coding used in the processing of the encoding unit 51 that continuously converts m-bit information words into n-bit code words after inserting DC control bits for DC control of the code series into the information series In the first example of the rule, when the coding rule is expressed by a code conversion table of a finite state type, the code conversion table assumes that the DC control bit is 0 and is encoded from an arbitrary starting point state. time of the code states s ₀ of c _0, assuming DC control bit 1 and the code states s ₁ of the provisional code sequence c ₁ obtained by encoding the same starting state and the provisional code sequence c ₀ becomes identical Thus, the code word is assigned to the information word so that the two's complement of the total of the temporary code sequence c ₀ and the two's complement of the total of the temporary code sequence c ₁ are always different.

  Also, encoding used in the processing of the encoding unit 51 that continuously converts an m-bit information word into an n-bit code word after inserting DC control bits for DC control of the code sequence into the information sequence In the second example of the rule, in addition to the conditions of the first example, when the encoding rule is expressed by a look-ahead type code conversion table, one block of m-bit information word or n-bit code word is used. When the number of blocks is odd, the two's complement of the total of the information sequence is different from the two's complement of the total of the code sequence, and when the number of blocks is even, the two's complement of the total of the information sequence is The code word is assigned to the information word so that the two's complement of the total of the code sequence matches.

Also, encoding used in the processing of the encoding unit 51 that continuously converts an m-bit information word into an n-bit code word after inserting DC control bits for DC control of the code sequence into the information sequence In the third example of the rule, when the encoding rule is expressed by a finite state type code conversion table, a DC control bit is inserted into either the first bit or the second bit of the 2-bit information word. Only when the code conversion table assumes that the DC control bit is 0 and the code state s _{0 of} the temporary code sequence c ₀ encoded from an arbitrary starting point state and the DC control bit is 1, the temporary code when the the code states s ₁ of the provisional code sequence c ₁ obtained by encoding the same starting state and sequence c ₀ becomes identical, and the two's complement of the sum of the provisional code sequence c _0, the provisional code sequence c ₁ The information word is marked so that the two's complement is always different. Word is assigned.

  As described above, the parity-difference codeword allocation method in the finite-state code conversion table includes the conventional parity-conserving codeword allocation method in the look-ahead code conversion table, and includes the conventional parity-conserving code. In addition to the word assignment method, a very large number of codeword assignments with high DC control efficiency are provided. However, even if the parity-differential codeword assignment in the finite-state code conversion table results in the parity-preserving codeword assignment in the look-ahead code conversion table, the parity-differential codeword assignment However, the characteristics can be greatly improved as compared with the conventional code with parity-preserving codeword assignment described with reference to FIGS. 7 and 8.

  Next, a case where the maximum continuous number of minimum runs in the (1, 7) RLL code is made smaller than the conventional 6 and the bit error rate is improved when defocusing or tangential tilt occurs during reproduction processing. explain.

  FIG. 29 shows the relationship between the maximum number of consecutive minimum runs and the Shannon capacity of the code in the (1,7) RLL code. Here, the Shannon capacity is a theoretical maximum coding rate that can be achieved by a limited code. By using the ACH algorithm described above, it is possible to design a code conversion table of a finite state type for codes having a coding rate less than the Shannon capacity.

  From FIG. 29, the maximum continuous number of minimum runs in the (1,7) RLL code is up to the minimum run maximum continuous number 3 with a Shannon capacity of 0.6730 if the coding rate remains 2/3 = 0.6667. It can be seen that this can be reduced.

  That is, when a 2-bit information word is continuously converted into a 3-bit code word and encoding is performed using an encoding rule in which the minimum run limit of the code is 1 and the maximum run limit is 7, The maximum number of consecutive consecutive runs is limited to 3 or more and 5 or less.

  Next, as a specific example, a method for designing a specific finite state type code conversion table of the (1, 7) RLL code when the maximum continuous number of minimum runs is 5 will be described.

  FIG. 30 is a finite state transition diagram of 18 states that simultaneously gives the (1,7) RLL restriction rule and the restriction that the maximum continuous number of minimum runs is 5 when NRZI modulation is assumed.

  Here, in FIG. 30, when the maximum continuous number of minimum runs is to be set to 4, a state transition diagram in which states 17 to 18 are deleted is used, and to set to 3, state transitions in which states 15 to 18 are deleted. A figure may be used.

  First, based on the finite state transition diagram of FIG. 30, [4, 6, 6, 6, 5, 5, 4, 2, 4, 6, 3, 6, 3, 5, 3, 5, 2 as approximate eigenvectors. , 3] is derived. Based on this approximate eigenvector, an 8-state finite state table similar to the 7-state finite state table of FIG. 15 can be designed based on the ACH algorithm. FIG. 31 shows a finite state table for a (1,7) RLL code with a maximum run number of 5 as the minimum run.

Then, based on the flowchart described with reference to FIG. 17, using the finite state table of FIG. 31, all the codeword assignments are (4!) ⁸ = 110,075,314,176, depending on the block length to be observed. First, a codeword assignment satisfying Equation (2) for an arbitrary information sequence, that is, a parity-difference codeword assignment was searched. As a result, when the insertion position of the DC control bit is fixed to only one of the first bit and the second bit of the information word, Expression (3) is established at L = 7. At that time, there are 786,432 codeword allocation numbers for parity-differential codeword allocation. When the insertion position of the DC control bit is inserted into at least one of the first bit and the second bit of the information word, there are 1,179,648 codeword allocation numbers for parity-differential codeword allocation. This number of codeword assignments is 1/93312 of the total number of codeword assignments in the finite state table of FIG.

  However, since the finite state table in FIG. 31 is created using the ACH algorithm, no matter what codeword assignment is used in the finite state table in FIG. 31, the code conversion table after codeword assignment is used. The encoded code can be subjected to sliding block decoding, and there are code block constraint lengths of 5 blocks and 4 blocks at the time of decoding. Further, among the 2-bit information words having a codeword constraint length of 4 blocks at the time of decoding, there are codes having a codeword constraint length of 3 blocks of only one bit. In such a case, the codeword constraint length is still 4 blocks, but the error propagation length during decoding is shortened by 1 bit.

  With reference to FIG. 32, the breakdown of the number of 1,179,648 codeword allocations that are parity-differential codeword allocation in the finite state table of FIG. 31 will be described. The type classification in FIG. 32 is the same classification as type 1 to type 3 described with reference to FIG. Further, in FIG. 32, the codeword constraint length at the time of decoding is described by classifying the codeword allocation of 5 blocks and the codeword allocation of 4 blocks. This is the number of codeword assignments in which the codeword constraint length of one bit of a 2-bit information word is 3 blocks out of 4 block codeword assignments.

  FIG. 33 is a specific example of a finite state type 8-state code conversion table of a (1, 7) RLL code in which a parity-difference type codeword is assigned and the maximum number of consecutive consecutive minimum runs is limited to 5. The codeword allocation of the code conversion table of FIG. 33 is that the codeword constraint length at the time of decoding is 4 blocks, and the codeword constraint length of one bit of 2-bit information words is 3 blocks, and the information word This is one of 256 codeword assignments of type 3 shown in FIG. 32, which is a parity-differential codeword assignment regardless of which of the first bit and the second bit of DC is a DC control bit.

  The code according to the code conversion table of FIG. 33 is reduced from 6 to 5 in the maximum number of consecutive runs of the minimum run as compared with the codes according to the code conversion tables of FIG. 7, FIG. 8, and FIG. In addition, the number of states during encoding is greatly improved from 1691 to 8, code word constraint length during decoding from 6 blocks to 4 blocks, and error propagation length during decoding from 15 bits to 7 bits. ing.

  Thus, in the encoding method in which a 2-bit information word is continuously converted into a 3-bit code word, and the minimum run limit of the code is 1 and the maximum run limit is 7, the maximum of the minimum run of the code A code in which the number of continuations is limited to 5 has characteristics superior to those in which the maximum number of continuations of the minimum run is 6. The encoding unit 51 described with reference to FIGS. 11 to 13 performs the encoding process using a code in which the maximum continuous number of minimum runs is limited to 5 in the process of step S6 of FIG. It is possible to reduce the number of states at the time of encoding, to shorten the codeword constraint length at the time of decoding, and to reduce the error propagation length.

  Further, when the encoding rule is expressed by a finite state type code conversion table, the number of states can be set to eight.

  Also, the code conversion table of the finite state type in FIG. 33 can be converted into a look-ahead type code conversion table by reassigning the code word advanced by one block to the information word. The look-ahead type code conversion table obtained as a result is the same as the basic conversion table in FIG. 25, and is obtained by adding the conversion table at the time of violation shown in FIG. 34 to the basic conversion table in FIG.

  That is, the code conversion table at the time of violation in FIG. 34 is a code conversion table at the time of violation added to limit the maximum continuous number of minimum runs to 5 with respect to the look-ahead type code conversion table of FIG. 34, when the information word “10.11” is supplied, if the immediately preceding code word is (010), the code word (000) is not used according to the code conversion table of FIG. .001) is a code conversion table for violation.

  In the code conversion table at the time of violation shown in FIG. 34, as in the case of FIG. 25, codeword allocation is performed so that the parity of the information sequence and the parity of the code sequence match when the number of blocks is an even number. . The encoding is performed based on the code conversion rule in which the look-ahead type code conversion table of FIG. 34 is added to the look-ahead basic code conversion table of FIG. The maximum number of consecutive runs can be limited to 5. However, this encoding method is equivalent to performing encoding based on the finite state type code conversion table of FIG.

  FIG. 35 shows the number of states at the time of encoding, the codeword constraint length at the time of decoding in the code conversion table of 7 states shown in FIGS. 20 to 22 and the code conversion table of 8 states shown in FIG. It is the result of calculating the error propagation length at the time of decoding.

  The encoding method using the 7-state code conversion table shown in FIGS. 20 to 22 and the 8-state code conversion table shown in FIG. 33 is apparent from a comparison between FIG. 35 and FIG. Compared with the conventional encoding method using the code conversion tables of FIG. 7, FIG. 8, and FIG. 9, the number of states at the time of encoding, the code word constraint length at the time of decoding, and the error propagation length at the time of decoding Can be greatly improved.

  Note that any method may be used as a decoding method for decoding a code using the encoding method described above. As the decoding method, for example, as is apparent from the fact that the codeword constraint length at the time of decoding is obtained in FIG. 35, general sliding block decoding or the like may be used. As another decoding method, for example, by utilizing the relatively small number of code states in the encoding method of the present invention, for example, soft decision decoding using a trellis diagram according to a finite state type code conversion table Etc. may be used.

For example, when sliding block decoding is performed on a code encoded in accordance with the code conversion table of FIG. 21 or FIG. 33, an input code word string of 4 blocks (12 bits) is [b ₀ , b ₁ , b ₂ , b ₃ , b ₄ , b ₅ , b ₆ , b ₇ , b ₈ , b ₉ , b ₁₀ , b ₁₁ ], the 2-bit information word [a ₀ , a ₁ ] According to the following Bool algebraic logic expressions (4) and (5), it can be decoded with a very simple circuit.

However, when x = 0, both codes represented by FIG. 33 or FIG. 21 can be decoded, and when x = 1, the codes represented by FIG. 21 can be respectively decoded. B ₁₁ is not used in sliding block decoding. At the time of decoding, the DC control bits inserted at the time of encoding are usually discarded.

  The series of processes described above can also be executed by software. The software is a computer in which the program constituting the software is incorporated in dedicated hardware, or various functions can be executed by installing various programs, for example, a general-purpose personal computer For example, it is installed from a recording medium.

  As shown in FIG. 16, the recording medium for recording the program for executing the above-described series of processing is distributed to provide the program to the user separately from the computer, and is a magnetic disk 141 (on which the program is recorded). Flexible disk), optical disk 142 (including compact disk-read only memory (CD-ROM), DVD (digital versatile disk)), magneto-optical disk 143 (including MD (mini-disk) (trademark)), or It is constituted by a package medium composed of a semiconductor memory 144 or the like.

  Further, in the present specification, the step of describing the program recorded in the recording medium is not limited to the processing performed in chronological order according to the described order, but may be performed in parallel or It also includes processes that are executed individually.

It is a block diagram which shows the structure of the conventional recording / reproducing apparatus. It is a figure explaining the look-ahead type code conversion table of a (1,7) RLL code. It is a figure explaining the look-ahead type code conversion table of a (1,7) RLL code. It is a figure for demonstrating the finite state type | mold 5-state code conversion table which converted the code conversion table of FIG. 2 and FIG. It is a state transition diagram of 8 states giving a (1, 7) RLL code. It is a figure for demonstrating the finite state type | mold 4-state code conversion table of a (1,7) RLL code. It is a figure explaining the look-ahead type | mold code conversion table | surface using the parity preservation type | mold codeword allocation of a (1,7) RLL code. It is a figure explaining the look-ahead type | mold code conversion table | surface using the parity preservation type | mold codeword allocation of a (1,7) RLL code. It is a figure explaining the code conversion table | surface at the time of violation added to the code conversion table | surface of FIG. 7 and FIG. 8, and restrict | limits the maximum continuous number of the minimum run to six. It is a figure explaining the number of states of a conventional (1,7) RLL code, codeword constraint length, and error propagation length. It is a block diagram which shows the structure of the recording / reproducing apparatus to which this invention is applied. It is a block diagram which shows the structure of the encoding part of FIG. 13 is a flowchart for describing an encoding process executed by an encoding unit in FIG. 12. It is a figure for demonstrating the state of 2 series in the encoding process which the encoding part of FIG. 12 performs. It is a figure for demonstrating the 7 state finite state table | surface of a (1,7) RLL code | symbol. It is a block diagram which shows the structure of a personal computer. It is a flowchart explaining a code conversion table search process. It is a figure explaining the codeword allocation number which satisfy | fills the conditions for every block length. It is a figure explaining the type classification | category of codeword allocation which satisfy | fills conditions satisfy | fill regardless of block length. It is a figure for demonstrating the 1st example of the 7 state code conversion table of the finite state type | mold of a (1,7) RLL code which satisfy | fills conditions. It is a figure for demonstrating the 2nd example of the 7 state code conversion table of the finite state type | mold of a (1,7) RLL code which satisfy | fills conditions. It is a figure for demonstrating the 3rd example of the 7 state code conversion table of the finite state type | mold of a (1,7) RLL code which satisfy | fills conditions. It is a figure for demonstrating the state of 2 series in the encoding process which the encoding part of FIG. 12 performs. It is a figure for demonstrating the look-ahead type | mold code conversion table of the (1,7) RLL code which satisfy | fills the conditions converted from the code conversion table of FIG. It is a figure for demonstrating the look-ahead type code conversion table of the (1,7) RLL code which satisfy | fills the conditions converted from the code conversion table of FIG. FIG. 23 is a diagram for describing a look-ahead type code conversion table of a (1,7) RLL code that satisfies the conditions converted from the code conversion table of FIG. 22; It is a figure explaining DC control redundancy dependence of electric power spectrum density. It is a figure explaining DC control redundancy dependence of electric power spectrum density. It is a figure explaining the maximum continuous number and Shannon capacity of the minimum run. It is a state transition diagram of 18 states of a (1, 7) RLL code that limits the maximum number of consecutive runs of a minimum run to 5. It is a figure for demonstrating the 8-state finite state table | surface of a (1, 7) RLL code which restrict | limits the maximum continuous number of the minimum run to five. It is a figure explaining the codeword allocation number by the type classification | category of the codeword allocation which satisfy | fills conditions satisfy | fill regardless of block length. It is a figure for demonstrating the finite state type | mold 8-state code conversion table of the (1,7) RLL code which restrict | limits the maximum continuous number of the minimum run to five. It is a figure for demonstrating the code conversion table | surface at the time of violation added to the basic code conversion table | surface of FIG. 25 in order to restrict | limit the maximum continuous number of the minimum run to five. The number of states at the time of encoding, the codeword constraint length at the time of decoding, and the error at the time of decoding in the code conversion table for the seven states shown in FIGS. 20 to 22 and the code conversion table for the eight states shown in FIG. It is a figure for demonstrating propagation length.

Explanation of symbols

  31 recording / reproducing apparatus, 41 encoding apparatus, 51 encoding section, 101 DC control bit insertion section, 102, 103 m / n conversion section, 106 code selection signal generation section, 107 code selection section

Claims (6)

In an encoding device that continuously converts an m-bit information word into an n-bit code word,
encoding means for continuously converting the m-bit information word to the n-bit codeword;
The encoding means executes an encoding process based on an encoding rule having a minimum run limit of 1, a maximum run limit of 7, and a maximum continuous number of minimum runs of the codeword of 3 to 5. An encoding device characterized by: The encoding rule used by the encoding means has a state number of 8 when expressed using a code conversion table and a maximum continuous number of the minimum run of 5. Encoding device. The encoding apparatus according to claim 1, wherein the encoding unit performs code conversion so that a codeword constraint length at the time of decoding is 4 blocks. In an encoding method of an encoding device for continuously converting an m-bit information word into an n-bit code word,
Based on a coding rule in which the minimum run limit is 1, the maximum run limit is 7, and the maximum number of consecutive runs of the code word is 3 to 5, the m-bit information word is converted to the n-bit information word. An encoding method comprising: an encoding step of continuously converting codewords. A program for causing a computer to execute a process of continuously converting an m-bit information word into an n-bit code word,
Based on a coding rule in which the minimum run limit is 1, the maximum run limit is 7, and the maximum number of consecutive runs of the code word is 3 to 5, the m-bit information word is converted to the n-bit information word. A computer-readable recording medium on which a program for causing a computer to execute a process including an encoding step of continuously converting into a code word is recorded. A program for causing a computer to execute a process of continuously converting an m-bit information word into an n-bit code word,
Based on a coding rule in which the minimum run limit is 1, the maximum run limit is 7, and the maximum number of consecutive runs of the code word is 3 to 5, the m-bit information word is converted to the n-bit information word. The program for making a computer perform the process characterized by including the encoding step which converts into a codeword continuously.

Images