JP2012175564A - Decoding apparatus, encoding apparatus, decoding method and encoding method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a technique for reducing the computational complexity of a decoding process while keeping decoding characteristics intact.SOLUTION: A decoding section 28 inputs data LDPC-coded with a check matrix via a channel. A channel state estimation section 46 estimates a state of the channel from the input data. A reduction number decision section 48 and a check matrix modification section 50 modify the size of the check matrix in accordance with the estimated state of the channel. A frame data number adjustment section 42 adjusts the number of parity bits in the input data in accordance with the modified size of the check matrix. An LDPC decoding section 44 executes a decoding processing using the modified check matrix on the adjusted data.

Description

  The present invention relates to a decoding technique and an encoding technique, and more particularly, to a decoding apparatus, an encoding apparatus, a decoding method, and an encoding method that should use LDPC encoding.

  In recent years, LDPC (Low Density Parity Check Code) has attracted attention as an error correction code having strong error correction capability even in a low S / N transmission path, and is applied in many fields. In LDPC, data is encoded by an encoding matrix generated on the transmission side based on a sparse check matrix. Here, a sparse check matrix is a matrix having 1 or 0 elements and a small number of 1s. On the other hand, on the receiving side, data decoding and parity check are performed based on the check matrix. In particular, decoding performance is improved by iterative decoding using the BP (Belief Propagation) method or the like.

  In this decoding, a check node process for decoding in the row direction and a variable node process for decoding in the column direction are repeatedly executed for a portion where the matrix element of the parity check matrix is “1”. Therefore, the amount of decoding processing is determined by the number of check matrices and the number of matrix elements “1”. Here, sum-product decoding using a Gallager function or a hyperbolic function is known as one of the check node processes. A decoding method that simplifies sum-product decoding is min-sum decoding. One of LDPC parity check matrices is a LDGM parity check matrix. The parity check matrix of the LDGM structure is a special matrix that can be encoded without obtaining a generator matrix from the parity check matrix.

  Specifically, examples of the matrix of the portion that is affected by the parity bit of the parity check matrix have a structure in which the upper right triangle is “0” and a structure in which the double diagonal portion is “0”. According to this structure, since the parity is calculated in order from the upper parity, the generator matrix and the check matrix are the same matrix. When the LDGM check matrix is used, the amount of calculation decreases if the number of portions of the matrix affected by the parity bits is reduced. On the other hand, since the coding rate increases according to the reduction amount, the decoding characteristics deteriorate. On the other hand, when the wireless device operates by battery drive, it is desired to extend the operation time. Therefore, it is necessary to reduce the amount of calculation. For example, decoding is performed after reconstructing a parity check matrix according to the state of the communication path (see, for example, Patent Document 1).

JP 2006-013714 A

  Even when the parity check matrix is reconstructed as in the background art, since the structure of the parity check matrix is changed for the purpose of improving decoding characteristics, the amount of calculation is not reduced. Therefore, it is desired to reduce the amount of decoding processing and reduce power consumption by changing the number of check matrices according to the state of the communication path. At that time, it is also desired to suppress the deterioration of the decoding characteristics and simplify the process of changing the number of check matrices.

  The present invention has been made in view of such circumstances, and an object of the present invention is to provide a technique for reducing the amount of decoding processing while suppressing deterioration in decoding characteristics.

  In order to solve the above problems, a decoding apparatus according to an aspect of the present invention is based on an input unit that inputs data that is LDPC-encoded by a parity check matrix via a communication channel, and data that is input in the input unit. , An estimation unit that estimates the state of the communication channel, a change unit that changes the size of the parity check matrix according to the state of the communication channel estimated by the estimation unit, and an input according to the size of the parity check matrix that is changed by the change unit An adjustment unit that adjusts the number of parity bits among the data input in the unit, and a decoding unit that performs a decoding process on the data adjusted in the adjustment unit based on the parity check matrix changed in the change unit, Is provided.

  According to this aspect, since the decoding process is executed based on the parity check matrix whose size is changed according to the estimated state of the communication channel, it is possible to reduce the calculation amount of the decoding process while suppressing the deterioration of the decoding characteristics.

  The data input in the input unit is LDPC-encoded with a check matrix having an LDGM structure, and the changing unit reduces the rows in order from the bottom row or the top row of the check matrix corresponding to the parity bit to be encoded last, The size of the parity check matrix may be changed by reducing the columns in order from the rightmost column or the leftmost column corresponding to the parity bit to be encoded last. In this case, since the parity check matrix having the LDGM structure is used, and the matrix is reduced in order from the bottom row or the top row, and the rightmost column or the leftmost column, decoding can be performed even if it differs from the parity check matrix used in the transmission apparatus. .

  The changing unit may change the size of the parity check matrix by deleting any row of the parity check matrix. In this case, since only the line is deleted, the processing can be simplified.

  Another aspect of the present invention is an encoding device. This apparatus changes the size of the check matrix of the LDGM structure according to the acquisition unit that acquires the estimation result of the state of the communication path with the receiving apparatus, and the estimation result of the state of the communication path acquired by the acquisition unit A changing unit, an encoding unit that performs LDPC encoding on the data using a check matrix whose size is changed by the changing unit, and an output unit that outputs the LDPC encoded data in the encoding unit to the receiving device via a communication path. . The changing unit reduces the rows in order starting from the bottom row or top row of the parity check matrix corresponding to the last encoded parity bit, and reduces the columns starting from the rightmost column or leftmost column corresponding to the last encoded parity bit. To change the size of the parity check matrix.

  According to this aspect, since the check matrix having a reduced size at the time of reception is also used for encoding, transmission efficiency can be improved.

  When retransmitting data, the encoding unit may use a check matrix having a size larger than the size of the check matrix before retransmission. In this case, at the time of retransmission, a check matrix having a size larger than that of the check matrix before retransmission is used, so that decoding characteristics can be improved.

  Yet another embodiment of the present invention is a decoding method. This method includes a step of inputting data LDPC-encoded by a check matrix through a communication channel, a step of estimating a state of the communication channel based on the input data, and a state of the estimated communication channel. Accordingly, the step of changing the size of the parity check matrix, the step of adjusting the number of parity bits in the input data according to the size of the changed parity check matrix, and the changed check for the adjusted data Performing a decoding process based on the matrix.

  Yet another embodiment of the present invention is an encoding method. The method includes a step of obtaining an estimation result of a state of a communication channel with a receiving device, a step of changing a size of a parity check matrix of an LDGM structure according to the obtained estimation result of a state of a communication channel, The method includes a step of LDPC encoding data using a parity check matrix changed and a step of outputting the LDPC encoded data to a receiving device via a communication path. The changing step reduces the rows in order from the bottom row or the top row of the parity check matrix corresponding to the parity bit to be encoded last, and the columns in order from the rightmost column or the leftmost column corresponding to the parity bit to be encoded last. By reducing the size of the check matrix.

  It should be noted that any combination of the above-described constituent elements and a conversion of the expression of the present invention between a method, an apparatus, a system, a recording medium, a computer program, etc. are also effective as an aspect of the present invention.

  According to the present invention, the amount of decoding processing can be reduced while suppressing deterioration in decoding characteristics.

It is a figure which shows the structure of the communication system which concerns on Example 1 of this invention. It is a figure which shows the test matrix used in the encoding part of FIG. It is a figure which shows the structure of the decoding part of FIG. It is a figure which shows the BER-CNR characteristic measured in the channel state estimation part of FIG. FIG. 4 is a diagram illustrating a data structure of a check matrix number reduction table in FIG. 3. It is a figure which shows the Tanner graph which represented typically the operation | movement of the LDPC decoding part of FIG. It is a figure which shows the outline | summary of the update of the external value ratio in the LDPC decoding part of FIG. It is a figure which shows the outline | summary of the update of the prior value ratio in the LDPC decoding part of FIG. It is a flowchart which shows the procedure of the decoding process by the decoding part of FIG. It is a figure which shows the data structure of the check matrix number reduction table which concerns on Example 2 of this invention. It is a flowchart which shows the procedure of the decoding process by the decoding part which concerns on Example 2 of this invention. It is a figure which shows the structure of the decoding part and encoding part which concern on Example 3 of this invention. 13 is a flowchart illustrating a procedure of encoding processing by the decoding unit and the encoding unit in FIG. 12. 14A to 14C are diagrams illustrating another parity check matrix used in the encoding unit in FIG.

Example 1
Before describing the present invention specifically, an outline will be given first. The first embodiment of the present invention performs iterative decoding based on a check matrix for a transmission apparatus that performs LDPC encoding and data encoded in the transmission apparatus (hereinafter referred to as “encoded data”). The present invention relates to a communication system including a receiving device. In particular, the receiving device executes a min-sum algorithm. As described above, the min-sum algorithm is realized by simple processing, but the decoding characteristics are likely to deteriorate. In order to reduce the calculation amount of the decoding process while suppressing the deterioration of the decoding characteristics, the communication system according to the present embodiment, particularly the receiving apparatus, is configured as follows.

  Note that the parity check matrix is formed with an LDGM structure. The receiver estimates the channel state with the transmitter, reduces the number of check matrixes when the channel state is good, and excludes the parity bits of the demodulated data corresponding to the deleted matrix part Perform a decoding operation. By reducing the number of check matrices in this way, the amount of computation for decoding processing is reduced. On the other hand, the reception apparatus suppresses deterioration of decoding characteristics by using the check matrix as it is when the channel state is deteriorated.

  FIG. 1 shows a configuration of a communication system 100 according to Embodiment 1 of the present invention. The communication system 100 includes a transmission device 10 and a reception device 12. The transmission device 10 includes an information data generation unit 20, an encoding unit 22, and a modulation unit 24. The receiving device 12 includes a demodulator 26, a decoder 28, and an information data output unit 30.

  The information data generation unit 20 acquires data to be transmitted and generates information data. The acquired data may be used as information data as it is. The information data generation unit 20 outputs the information data to the encoding unit 22. The encoding unit 22 inputs information data from the information data generation unit 20. The encoding unit 22 adds parity (hereinafter referred to as “LDPC parity”) based on a parity check matrix in LDPC to information data. Information data to which the LDPC parity is added corresponds to the encoded data described above. The encoding unit 22 outputs the encoded data to the modulation unit 24. FIG. 2 shows a parity check matrix used in the encoding unit 22. Here, in order to simplify the description, it is assumed that the parity check matrix has 7 rows and 18 columns, but is not limited thereto. As shown in the figure, the LDGM structure has an upper right triangle of “0”. In this case, columns 1 to 11 of column numbers are calculated as message bit portions, and columns 12 to 18 of column numbers are calculated as parity bit portions. Is done.

次に、符号化部２２は、検査行列の２行目の情報データとメッセージビット（ｍ１〜ｍ１１）の情報と既に導出したパリティビットｐ１の情報から、パリティビットｐ２を次のように導出する。

The encoding unit 22 first derives the parity bit p1 from the information data of the first row of the parity check matrix and the information of the message bits (m1 to m11). More specifically, the encoding unit 22 derives a parity bit so that an exclusive OR (XOR) of a message bit having a constituent element “1” and a parity bit is “0” in the check matrix. To do. As a result, the parity bit p1 is derived from the information data of the first row of the check matrix and the information of the message bits (m1 to m11) as follows.

Next, the encoding unit 22 derives the parity bit p2 from the information data of the second row of the parity check matrix, the information of the message bits (m1 to m11), and the information of the already derived parity bit p1, as follows.

このように右上三角の部分に「０」を配置させた検査行列の場合、パリティビットはｐ１、ｐ２、ｐ３、ｐ４、ｐ５、ｐ６、ｐ７の順に導出される。図１に戻る。 The encoding unit 22 repeats such processing, and finally from the information data of the seventh row of the check matrix, the information of the message bits (m1 to m11), and the information of the already derived parity bits (p1 to p6). The parity bit p7 is derived as follows.

In the case of a parity check matrix in which “0” is arranged in the upper right triangular portion in this way, the parity bits are derived in the order of p1, p2, p3, p4, p5, p6, and p7. Returning to FIG.

  The modulation unit 24 receives encoded data from the encoding unit 22. The modulation unit 24 modulates the encoded data. As a modulation method, PSK (Phase Shift Keying), FSK (Frequency Shift Keying), or the like is used. The modulation unit 24 transmits the modulated encoded data as a modulation signal. The demodulator 26 receives the modulated signal from the modulator 24 via a communication path, for example, a wireless transmission path. The demodulator 26 demodulates the modulated signal. Since a known technique may be used for demodulation, the description is omitted here. The demodulator 26 outputs a demodulation result (hereinafter referred to as “demodulated data”) to the decoder 28.

The decoder 28 receives the demodulated data from the demodulator 26. The decoding unit 28 repeatedly performs a decoding process using a parity check matrix in LDCP on the demodulated data. As the decoding process, for example, a min-sum algorithm is executed. The min-sum algorithm is executed in the following procedure.
1. Initialization: The prior value ratio is initialized and the maximum number of decoding iterations is set.
2. Check node processing: The external value ratio is updated in the row direction of the check matrix.
3. Variable node processing: The priori value ratio is updated in the column direction of the check matrix.
4). Calculate temporary estimated words.

  The decoding unit 28 outputs the decoding result (hereinafter referred to as “decoded data”) to the information data output unit 30. The information data output unit 30 inputs the decoded data from the decoding unit 28. The information data output unit 30 generates information data based on the decoded data. The decoded data may be used as information data as it is. The information data output unit 30 includes an outer code decoding unit, and may decode an outer code such as a CRC (Cyclic Redundancy Check), for example.

  This configuration can be realized in terms of hardware by a CPU, memory, or other LSI of any computer, and in terms of software, it can be realized by a program loaded in the memory, but here it is realized by their cooperation. Draw functional blocks. Accordingly, those skilled in the art will understand that these functional blocks can be realized in various forms by hardware only, software only, or a combination thereof.

  FIG. 3 shows the configuration of the decoding unit 28. The decoding unit 28 includes a frame data storage unit 40, a frame data number adjustment unit 42, an LDPC decoding unit 44, a channel state estimation unit 46, a reduction number determination unit 48, a check matrix change unit 50, a check matrix number reduction table 52, a check A matrix storage unit 54 is included.

  The frame data storage unit 40 receives demodulated data from the demodulator 26 (not shown). The demodulated data can be said to be data that has been subjected to LDPC encoding via a communication channel and that has been subjected to LDPC encoding using a check matrix having an LDGM structure. The frame data storage unit 40 detects a frame synchronization signal included in the demodulated data. The frame data storage unit 40 specifies a unit of a frame formed by the demodulated data based on the frame synchronization signal. For example, when the frame synchronization signal is arranged at the head portion of the frame and the frame period has a fixed length, the frame data storage unit 40 identifies the fixed length period as a frame after detecting the frame synchronization signal. The unit of LDPC encoding may be a frame. The frame data storage unit 40 temporarily stores the demodulated signal in units of frames.

  The communication path state estimation unit 46 receives demodulated data from the demodulation unit 26 (not shown). The communication channel state estimation unit 46 estimates the state of the communication channel (hereinafter referred to as “communication channel value”) based on the demodulated data. Here, CNR is measured as the channel value. Such a channel value corresponds to the amount of dispersion of demodulated data for one frame. FIG. 4 shows the BER-CNR characteristic measured by the channel state estimation unit 46. The basic parity check matrix in the figure corresponds to the case where 8-level FSK modulation / demodulation is performed at a coding rate R = 360/522 (number of parity bits = 162 bits). In contrast, the BER-CNR characteristics when the number of parity bits is reduced by 25% (41-bit reduction) and when reduced by 50% (81-bit reduction) are also shown. When the required quality is BER = 1.0E-3 or less, it is possible to reduce the number of variability bits in an area where the CNR is large. Returning to FIG.

  The check matrix number reduction table 52 is a table in which the matrix reduction number information of the check matrix with respect to the channel value is indicated. In the example of FIG. 4, since the basic check matrix should be used when the CNR is less than 17.5 dB, the number of matrix reductions is “0”. In addition, when the CNR is 17.5 dB or more and less than 19.0 dB, 25% reduction is possible, and the matrix reduction number is “41”. Furthermore, when the CNR is 19.0 dB or more, 50% reduction is possible, so the number of matrix reductions is “81”. A summary of this is a check matrix number reduction table 52. FIG. 5 shows the data structure of the check matrix number reduction table 52. The check matrix number reduction table 52 includes a channel value column 200, a check matrix reduction number column 202, and a check matrix reduction matrix number column 204. The channel value column 200 and the check matrix reduction number column 202 are as described above. Here, in order to simplify the description, the number of matrix reductions is set in three stages for the channel value, but the present invention is not limited to this. The check matrix reduction matrix number column 204 indicates the parity bit number, row number, and column number to be deleted with respect to the matrix reduction number. Returning to FIG.

  The reduction number determination unit 48 receives the channel value from the channel state estimation unit 46. Further, the reduction number determination unit 48 selects the matrix reduction number and the parity check matrix reduction matrix number corresponding to the channel value by referring to the parity check matrix number reduction table 52. The latter corresponds to specifying the parity bit number, row number, and column number to be deleted. The reduction number determining unit 48 outputs the parity bit number to be deleted to the frame data number adjusting unit 42, and outputs the row number and column number to be deleted to the check matrix changing unit 50. The parity check matrix storage unit 54 stores the parity check matrix illustrated in FIG.

  The check matrix changing unit 50 inputs the row number and column number to be deleted from the reduction number determining unit 48. The parity check matrix changing unit 50 deletes the matrix part corresponding to the row number and column number specified by the reduction number determining unit 48 from the parity check matrix stored in the parity check matrix storage unit 54. That is, the parity check matrix changing unit 50 changes the size of the parity check matrix by reducing the rows in order from the bottom row of the parity check matrix and reducing the columns in order from the rightmost column according to the estimated channel state. To do. The parity check matrix changing unit 50 outputs the parity check matrix to the LDPC decoding unit 44.

  The frame data number adjustment unit 42 inputs the parity bit number to be deleted from the reduction number determination unit 48. The frame data number adjustment unit 42 deletes the parity bit data corresponding to the parity bit number designated by the reduction number determination unit 48 from the demodulated data for one frame stored in the frame data storage unit 40. That is, the frame data number adjusting unit 42 adjusts the number of parity bits in the demodulated data according to the size of the check matrix changed by the check matrix changing unit 50. For example, if the matrix reduction number is “2”, the last encoded parity bit p7, the matrix in the 7th and 18th columns of the parity check matrix in which this parity bit p7 is calculated, and the second from the end Decoding processing is performed by reducing the 6th row and 17th column matrices of the parity bit p6 encoded in the check matrix and the parity check matrix in which the parity bit p6 is calculated. This is equivalent to decoding the encoded information at a lower encoding rate, so that the decoding process is performed normally, although the decoding characteristics deteriorate. The frame data number adjustment unit 42 outputs the demodulated data to the LDPC decoding unit 44.

  The LDPC decoding unit 44 performs a decoding process on the demodulated data whose number of parity bits has been adjusted by the frame data number adjusting unit 42 based on the parity check matrix whose size has been changed by the LDPC decoding unit 44. This is to perform LDPC decoding using a check matrix in which the number of matrices is reduced by a quantity corresponding to the channel state and demodulated data in which parity bits are reduced by a quantity corresponding to the channel state. Equivalent to. As described above, the min-sum algorithm is executed as LDPC decoding. In the min-sum algorithm, check node processing and variable node processing are executed alternately. FIG. 6 is a Tanner graph schematically showing the operation of the LDPC decoding unit 44. In the Tanner graph, b1 to b18 are called variable nodes, and c1 to c7 are called check nodes. Here, the number of variable nodes is n, and bn is the nth variable node. The number of check nodes is m, and cm is the mth check node. Also, message bits m1 to m11 and parity bits p1 to p7 of frame data are connected to the variable nodes b1 to b18. Returning to FIG.

In the check node process, the prior value ratio β is initialized at the beginning of the iterative decoding. Here, the demodulated data input to the LDPC decoding unit 44 is used as it is. Next, the check node processing unit 56 obtains the minimum value min | βmn ′ | of the absolute value of the prior value ratio. In the check node process, the external value ratio αmn from cm to bm is updated with the variable node connected to the check node. The calculation of αmn is performed as follows for all pairs (m, n) satisfying the check matrix Hmn = 1.
αmn = a (Πsign (βmn ′)) · min | βmn ′ | (4)
Here, n ′ is A (m) \ n: A (m) is a variable node set connected to the check node m, and \ n indicates a difference set not including n. Further, sign represents a signature function, and min | βmn ′ | represents selection of the absolute minimum value. Here, a is a normalization constant. FIG. 7 shows an outline of the update of the external value ratio in the LDPC decoding unit 44. The external value ratio α11 is derived from β11 ′. That is, in the check node process, the external value ratio is updated based on the prior value ratio. Returning to FIG. The minimum value min | βmn ′ | of the absolute value of the prior value ratio is derived for each iteration.

In the variable node process, the prior value ratio βmn from bn to cm is updated between αmn and a check node connected to the variable node. The calculation of βmn is performed as follows for all pairs (m, n) satisfying the check matrix Hmn = 1.
βmn = Σαm′n + λn (5)
Here, λn is equal to the input data yn. The input data yn corresponds to demodulated data from the demodulator 26. M ′ is B (n) \ m: B (n) is a check node set connected to the variable node n, and \ m is a difference set not including m. FIG. 8 shows an overview of updating the prior value ratio in the LDPC decoding unit 44. The prior value ratio β11 is derived from α1′1. That is, the variable node processing unit 58 updates the prior value ratio based on the external value ratio. Returning to FIG. The LDPC decoding unit 44 calculates a temporary estimated word after the check node process and the variable node process are repeated a predetermined number of times. Note that the LDPC decoding unit 44 may calculate a temporary estimated word as long as the parity check result is correct even before being repeated a predetermined number of times. The LDPC decoding unit 44 may output the temporary estimated word as a decoding result.

  The operation of the communication system 100 configured as above will be described. FIG. 9 is a flowchart showing the procedure of the decoding process by the decoding unit 28. The channel state estimation unit 46 measures the channel value (CNR) as the channel state by deriving the amount of dispersion of the received data while receiving data for one frame (S10). If data for one frame is being received (Y in S12), the process returns to step 10. If data for one frame is not being received (N in S12), the reduction number determination unit 48 reads the communication channel value (CNR) at the time of data reception (S14). The reduction number determining unit 48 reduces the number of check matrices if the CNR is not 19.0 dB or more as the first CNR value determination (N in S16) and the CNR is not 17.5 dB or more as the second CNR value determination (N in S18). A value less than 17.5 dB is read from the table 52 (S20). Here, the matrix reduction number is set to “0”, and the deleted parity number, the deleted row number, and the deleted column number are set to “none”.

  On the other hand, if the CNR is 17.5 dB or more as the second CNR value determination (Y in S18), the reduction number determination unit 48 reads a value of 17.5 dB or more from the check matrix number reduction table 52 (S22). Here, the matrix reduction number is “41”, the deleted parity number is “P162 to P122”, the deleted row number is “162 to 122”, and the deleted column number is “522 to 482”. Furthermore, if the CNR is 19.0 dB or more as the first CNR value determination (Y in S16), the reduction number determination unit 48 reads a value of 19.0 dB or more from the check matrix number reduction table 52 (S24). Here, the matrix reduction number is “81”, the deleted parity number is “P162 to P82”, the deleted row number is “162 to 82”, and the deleted column number is “522 to 442”. The reduction number determination unit 48 outputs the number list to be deleted read from the check matrix number reduction table 52 to the check matrix change unit 50 (S26). The parity check matrix changing unit 50 reduces the number of matrixes (size) of the parity check matrix according to the number list (S28). The frame data number adjusting unit 42 deletes the parity bits according to the number list (S30). The LDPC decoding unit 44 performs a decoding process using a parity check matrix and data in which the number of matrices (size) is reduced (S32).

  According to the embodiment of the present invention, the decoding process is performed based on the parity check matrix whose size is changed according to the estimated channel state, so that the size of the parity check matrix can be reduced as the channel state becomes better. . In addition, since the size of the check matrix is reduced, the amount of calculation can be reduced. Moreover, since the amount of calculation is reduced, power consumption can also be reduced. In addition, since the decoding process is executed based on the parity check matrix whose size is changed according to the estimated channel state, the size of the parity check matrix can be increased when the channel state deteriorates. Moreover, since the size of the parity check matrix is increased, it is possible to suppress deterioration of the decoding characteristics. In addition, since the decoding process is executed based on the parity check matrix whose size is changed according to the estimated channel state, it is possible to reduce the calculation amount of the decoding process while suppressing the deterioration of the decoding characteristics.

  In addition, the channel condition is estimated from the received data, and the number of check matrix reductions is determined from the result, and the decoding process is performed by changing the number of matrices, so that the encoding process on the transmission side is not changed. Power consumption can be suppressed only on the side. In addition, since the number of check matrices is reduced, the number of operations can be reduced by the number of 1 included therein. Further, by reducing the number of parity bits by 50%, the number of rows of the check matrix can be reduced by half, and the number of operations can be reduced by half. In addition, since a check matrix having an LDGM structure is used and the matrix is reduced in order from the bottom row and the rightmost column, decoding can be performed even if the check matrix is different from the check matrix used on the transmission side. In addition, since decoding is performed even if the check matrix is different from the check matrix used on the transmission side, a design change on the transmission side can be avoided.

(Example 2)
Next, a second embodiment of the present invention will be described. As in the first embodiment, the second embodiment of the present invention relates to a receiving apparatus that executes a decoding process by reducing the size of the parity check matrix when the communication channel state is good. In the first embodiment, the number of check matrixes is reduced in order to reduce the size of the check matrix. However, in the second embodiment, only the number of rows of the check matrix is reduced in order to reduce the size of the check matrix. In the case of the second embodiment, since the number of columns is not reduced, the process of deleting the parity bit from the demodulated data becomes unnecessary, and the amount of calculation can be reduced with a simpler configuration. The rows to be excluded from the check matrix may be arbitrarily determined, but in each column in which the message bit is calculated after deleting the row, it is determined so that at least one “1” always remains in the column element. Should be. The communication system 100 according to the second embodiment is the same type as that shown in FIG. 1, and the decoding unit 28 is the same type as that shown in FIG. Here, the difference will be mainly described.

  The check matrix number reduction table 52 is a table in which information on the number of rows reduced in the check matrix with respect to the channel value is indicated. The row reduction number information shown in the check matrix number reduction table 52 is also determined based on the BER-CNR characteristics shown in FIG. FIG. 10 shows a data structure of the check matrix number reduction table 52 according to the second embodiment of the present invention. The check matrix number reduction table 52 includes a channel value column 210, a check matrix row reduction number column 212, and a check matrix reduction row number column 214. The channel value column 210, the parity check matrix row reduction number column 212, and the parity check matrix reduction row number column 214 respectively correspond to the channel value column 200, the parity check matrix reduction number column 202, and the parity check matrix reduction matrix number column 204 in FIG. . Returning to FIG.

  The reduction number determination unit 48 receives the channel value from the channel state estimation unit 46. In addition, the reduction number determination unit 48 refers to the check matrix number reduction table 52 to obtain the number of lines to be reduced according to the channel value, and specifies the number of lines to be deleted. The reduction number determining unit 48 outputs the row number to be deleted to the check matrix changing unit 50. In the decoding unit 28 according to the second embodiment, the frame data number adjustment unit 42 does not adjust the number of parity bits. Therefore, the decoding unit 28 according to the second embodiment may not include the frame data number adjustment unit 42. The check matrix changing unit 50 inputs the row number to be deleted from the reduction number determining unit 48. The parity check matrix changing unit 50 deletes the row portion corresponding to the row number designated by the reduction number determining unit 48 from the parity check matrix stored in the parity check matrix storage unit 54. That is, the parity check matrix changing unit 50 changes the size of the parity check matrix by deleting an arbitrary row of the parity check matrix. The parity check matrix changing unit 50 outputs the parity check matrix to the LDPC decoding unit 44.

  FIG. 11 is a flowchart illustrating the procedure of the decoding process performed by the decoding unit 28 according to the second embodiment of the present invention. The channel state estimation unit 46 measures the channel value (CNR) as the channel state by deriving the distribution amount of the received data while receiving data for one frame (S50). If one frame of data is being received (Y in S52), the process returns to step 50. If data for one frame is not being received (N in S52), the reduction number determination unit 48 reads the communication channel value (CNR) at the time of data reception (S54). If the CNR is not 19.0 dB or more as the first CNR value determination (N in S56) and the CNR is not 17.5 dB or more as the second CNR value determination (N in S58), the reduction number determination unit 48 reduces the number of check matrices. A value less than 17.5 dB is read from the table 52 (S60). Here, the matrix reduction number is set to “0”, and the deleted row number is set to “none”.

  On the other hand, if the CNR is 17.5 dB or more as the second CNR value determination (Y in S58), the reduction number determination unit 48 reads a value of 17.5 dB or more from the check matrix number reduction table 52 (S62). Here, the matrix reduction number is set to “30”, and the deleted row numbers are set to “135... 126, 85... 76, 35. Furthermore, if the CNR is 19.0 dB or more as the first CNR value determination (Y in S56), the reduction number determination unit 48 reads a value of 19.0 dB or more from the check matrix number reduction table 52 (S64). Here, the number of matrix reductions is “60”, and the deleted row numbers are “135... 126, 110... 101, 85. .. 1 ”. The reduction number determination unit 48 outputs the number list to be deleted read from the check matrix number reduction table 52 to the check matrix change unit 50 (S66). The parity check matrix changing unit 50 reduces the number of rows (size) of the parity check matrix according to the number list (S68). The LDPC decoding unit 44 performs a decoding process using a parity check matrix and data with a reduced number of rows (size) (S70).

  According to the embodiment of the present invention, since the decoding process is executed after reducing the number of rows of the check matrix by a preset number according to the channel state, the amount of computation is suppressed while suppressing the deterioration of the decoding characteristics. Can be reduced. Moreover, since the amount of calculation is reduced, power consumption can be reduced. Further, since only the rows are deleted, the processing can be simplified. Moreover, since arbitrary rows can be reduced, the degree of freedom in design can be improved.

(Example 3)
Next, a third embodiment of the present invention will be described. In the first embodiment, when the communication channel state is good, the decoding process is executed with the size of the check matrix being reduced. The third embodiment of the present invention generates encoded data with a reduced number of parity bits by performing encoding processing using a check matrix with a reduced matrix size at the time of reception when transmitting data. Is a wireless device for transmission. For this reason, the wireless device according to the third embodiment has the functions of the transmission device and the reception device so far. Therefore, the wireless device includes each component included in the transmission device 10 and each component included in the reception device 12 in the communication system 100 of FIG. Here, the difference will be mainly described.

  FIG. 12 shows configurations of the decoding unit 28 and the encoding unit 22 according to the third embodiment of the present invention. The decoding unit 28 is configured in the same manner as in FIG. As described above, the parity check matrix changing unit 50 reduces the rows in order from the bottom row of the parity check matrix according to the obtained estimation result of the state of the communication channel, and reduces the columns in order from the rightmost column. Change the size of the structure check matrix. That is, if the state of the communication channel is good, the size of the check matrix is reduced. The parity check matrix changing unit 50 outputs the parity check matrix to the LDPC decoding unit 44 and also to the encoding unit 22.

  Encoding unit 22 inputs information data from information data generation unit 20 (not shown) and also inputs a check matrix from check matrix change unit 50. Note that the size of the check matrix may be changed depending on the estimation result of the communication path state. The encoding unit 22 performs LDPC encoding on the information data by adding the LDPC parity generated based on the parity check matrix to the information data. When the size of the check matrix is reduced, the number of bits of LDPC parity is reduced. This corresponds to an improvement in coding rate and transmission efficiency. Since the generation of the LDPC parity is as described above, the description thereof is omitted here. In order to cope with such a change in the coding rate, information on the coding rate may be included in a part of the information data. It is assumed that the coding rate of a section including information on the coding rate is a fixed value. At that time, a radio apparatus on the receiving side (not shown) sets the coding rate according to the information related to the coding rate, and executes LDPC decoding. The encoding unit 22 outputs the encoded data to a modulation unit 24 (not shown). Such encoded data is transmitted to another wireless device (not shown) via a communication path.

  When the input information data is retransmission data, the encoding unit 22 uses a parity check matrix whose size is not adjusted regardless of the parity check matrix input from the parity check matrix changing unit 50. This corresponds to using a parity check matrix with the lowest coding rate, and corresponds to transmitting all LDPC parities back to a normal parity check matrix. Therefore, it can be said that encoding section 22 uses a check matrix having a size larger than the size of the check matrix before retransmission when data is retransmitted.

  FIG. 13 is a flowchart illustrating the procedure of the encoding process performed by the decoding unit 28 and the encoding unit 22. If the data is not retransmitted data (N in S90), the encoding unit 22 reads the latest check matrix created at the time of reception, and encodes information data using the matrix (S92). On the other hand, if the data is retransmission data (Y in S90), the encoding unit 22 reads a normal check matrix and performs encoding processing of information data using the matrix (S94). The modulation unit 24 modulates the encoded data and transmits it (S96).

  According to the embodiment of the present invention, since the check matrix having a reduced size at the time of reception is also used for encoding, the number of bits of LDPC parity can be reduced. Further, since the number of bits of LDPC parity is reduced, transmission efficiency can be improved. Further, since the number of bits of LDPC parity is reduced, the amount of data transmission is reduced, and communication power can be reduced. Further, since a parity check matrix having a predetermined size is used at the time of retransmission regardless of the size reduced at the time of reception, decoding characteristics can be improved. In addition, since the decoding characteristics during retransmission are improved, the number of retransmissions can be reduced.

  In the above, this invention was demonstrated based on the Example. This embodiment is an exemplification, and it will be understood by those skilled in the art that various modifications can be made to the combination of each component and each processing process, and such modifications are also within the scope of the present invention. .

  In the embodiment of the present invention, since the communication system 100 is premised on a wireless communication system, the transmission device 10 and the reception device 12 are included in the wireless communication device. However, the present invention is not limited to this. For example, the communication system 100 may be based on a wired communication system. At that time, the transmission device 10 and the reception device 12 are included in the wired communication device. According to this modification, the present invention can be applied to various devices.

  In the embodiment of the present invention, the decoding unit 28 executes a min-sum algorithm. However, the present invention is not limited to this. For example, the decoding unit 28 may execute an algorithm different from the min-sum algorithm. An example of an algorithm different from the min-sum algorithm is the sum-product algorithm. According to this modification, various algorithms can be used for decoding.

  In the embodiment of the present invention, in the check matrix number reduction table 52, the number of reductions is set in three stages. However, the present invention is not limited to this. For example, in the check matrix number reduction table 52, the number of reductions may be set in four or more stages or in two stages. According to this modification, the degree of freedom in design can be improved.

  In the embodiment of the present invention, the encoding unit 22 uses the check matrix used at the time of reception. However, the present invention is not limited to this. For example, the encoding unit 22 may reduce the size of the parity check matrix separately from the parity check matrix used at the time of reception. In this case, the communication path state estimated in the communication target wireless apparatus is received from the communication target wireless apparatus, and the reduction number determining unit 48 and the check matrix changing unit 50 are based on the received communication path state. Determine a parity check matrix for use in encoding. This determination is made separately from the determination of the check matrix used for decoding. According to this modification, since the state of the communication path is received from the wireless device to be communicated, the accuracy of determining the check matrix can be improved.

  In the embodiment of the present invention, the check matrix of the LDGM structure is shown in FIG. That is, message bits are arranged in the left column of the parity check matrix, and parity bits are arranged in the right column of the parity check matrix. Here, the parity bits to be encoded last are arranged in 7 rows and 18 columns, and are arranged in the lowermost row and the rightmost column. However, the present invention is not limited to this. For example, the LDGM check matrix may have a different data structure. FIGS. 14A to 14C show other parity check matrices used in the encoding unit 22. The parity bits encoded last in the parity check matrix shown in FIGS. 14A to 14C are arranged at positions different from the parity bits encoded last in the parity check matrix shown in FIG. Has been. In FIG. 14A, the parity bits encoded last are arranged in the uppermost row and the rightmost column. In FIG. 14B, the parity bits encoded last are arranged in the bottom row and the leftmost column. In FIG. 14C, the parity bits encoded last are arranged in the uppermost row and the leftmost column.

  Furthermore, when the parity check matrix of FIG. 14A is used, the parity check matrix changing unit 50 reduces the rows in order from the top row of the parity check matrix and reduces the columns in order from the rightmost column, thereby checking the parity check matrix. Change the size of. When the parity check matrix in FIG. 14B is used, the parity check matrix changing unit 50 reduces the number of rows in order from the bottom row of the parity check matrix and reduces the size of the parity check matrix by reducing the columns in order from the leftmost column. To change. When the parity check matrix of FIG. 14C is used, the parity check matrix changing unit 50 reduces the size of the parity check matrix by reducing the rows in order from the top row of the parity check matrix and reducing the columns in order from the leftmost column. To change. That is, the parity check matrix changing unit 50 deletes in order from the row side and the column side where the parity bit to be encoded last is arranged. According to this modification, various types of LDGM structure check matrices can be used.

  DESCRIPTION OF SYMBOLS 10 Transmission apparatus, 12 Reception apparatus, 20 Information data generation part, 22 Encoding part, 24 Modulation part, 26 Demodulation part, 28 Decoding part, 30 Information data output part, 40 Frame data storage part, 42 Frame data number adjustment part, 44 LDPC decoding unit, 46 channel state estimation unit, 48 reduction number determination unit, 50 parity check matrix change unit, 52 parity check matrix number reduction table, 54 parity check matrix storage unit, 100 communication system.

Claims (7)

An input unit for inputting LDPC-encoded data by a check matrix through a communication path;
Based on the data input in the input unit, an estimation unit that estimates the state of the communication path;
A changing unit that changes the size of the check matrix according to the state of the communication path estimated by the estimating unit;
An adjustment unit that adjusts the number of parity bits in the data input in the input unit according to the size of the parity check matrix changed in the change unit;
A decoding unit that performs a decoding process on the data adjusted in the adjustment unit based on the parity check matrix changed in the change unit;
A decoding apparatus comprising: The data input at the input unit is LDPC encoded with a parity check matrix of LDGM structure,
The changing unit reduces the rows in order from the bottom row or the top row of the parity check matrix corresponding to the parity bit to be encoded last, and the columns in order from the rightmost column or the leftmost column corresponding to the parity bit to be encoded last. The decoding apparatus according to claim 1, wherein the size of the check matrix is changed by reducing the size.   The decoding apparatus according to claim 1, wherein the changing unit changes the size of the parity check matrix by deleting an arbitrary row of the parity check matrix. An acquisition unit for acquiring an estimation result of a state of a communication path with the receiving device;
A changing unit that changes the size of the parity check matrix of the LDGM structure according to the estimation result of the state of the communication path acquired by the acquiring unit;
An encoding unit that performs LDPC encoding of data using a parity check matrix whose size is changed in the changing unit;
An output unit that outputs data that has been LDPC encoded in the encoding unit to a receiving device via a communication path;
The changing unit reduces the rows in order from the bottom row or the top row of the parity check matrix corresponding to the parity bit to be encoded last, and the columns in order from the rightmost column or the leftmost column corresponding to the parity bit to be encoded last. An encoding apparatus, wherein the size of a check matrix is changed by reducing the size of the check matrix.   The encoding apparatus according to claim 4, wherein, when retransmitting data, the encoding unit uses a check matrix having a size equal to or larger than a size of a check matrix before retransmission. Inputting LDPC-encoded data by a check matrix via a communication path;
Estimating the state of the communication path based on the input data;
Changing the size of the check matrix according to the estimated channel state;
Adjusting the number of parity bits in the input data according to the size of the modified check matrix;
Performing a decoding process on the adjusted data based on the modified parity check matrix;
A decoding method comprising: Obtaining an estimation result of the state of the communication path with the receiving device;
Changing the size of the check matrix of the LDGM structure according to the obtained estimation result of the state of the communication path;
LDPC encoding the data with a resized check matrix;
Outputting LDPC-encoded data to a receiving device via a communication path,
The step of changing includes reducing the rows in order from the bottom row or the top row of the parity check matrix corresponding to the parity bit to be encoded last, and the columns from the rightmost column or the leftmost column corresponding to the parity bit to be encoded last. A coding method characterized in that the size of a parity check matrix is changed by reducing.

Images