JP5018807B2 - Decryption device - Google Patents

Description

  The present disclosure generally relates to a decoding apparatus, and more particularly to an apparatus for decoding a low density parity check code.

Error correction techniques are generally used in data communications such as networks and broadcasts.
The error correction technique can estimate and restore data transmitted only on the receiving side even if an error occurs in the sign bit of the received data. By using an error correction technique, it is possible to correct an error of a code bit generated due to the influence of disturbance on the transmission path.

  As a technique having a good error correction capability, a technique called a turbo code has been developed and used. In recent years, a low density parity check (LDPC) code has been studied as a technique capable of extracting performance equal to or higher than that of a turbo code, and research is being conducted in various places. Non-patent literatures 1 and 2 disclose LDPC code decoding techniques.

  Although the use of the LDPC code enables high-performance error correction, a large amount of calculation is required to decode the LDPC code. A decoding device that performs such a large amount of calculation has a problem that power consumption increases. Moreover, there is a problem that the circuit scale becomes large in order to execute a large amount of calculations.

International Publication No. WO2007 / 007801 Pamphlet JP 2008-48207 A JP 2004-343170 A JP 2003-198383 A

Jinghu Chen at al, “Reduced-Complexity Decoding of LDPC Codes”, IEEE Transactions on communications, 2005 vol.53.No.8 p1288-1289 Matsumoto Wataru, Sakai Atsushi, Yoshida Hideo, “Circular Approximation MIN Algorithm”, IEICE Technical Report RCS, Wireless Communication System, RCS2005-40, Vol.105, No.196, pp. 1-6

  In view of the above, there is a demand for a decoding device that reduces power consumption in decoding received signals that have been subjected to low-density parity check encoding. It is also desirable to reduce the circuit scale.

  A decoding device that decodes a low-density parity check-encoded received signal received via a transmission line includes: a column calculation unit that performs column processing using the received signal as input; and row processing based on a δ-min algorithm In an algorithm in which a logarithmic likelihood ratio is obtained from a received signal by including the noise variance of the transmission channel, and the noise variance value is set to a specific value, the state of the transmission channel is used as an input, and the state of the transmission channel A row calculation unit that performs a row process for performing correction using a correction term having a correction value corresponding to the correction value, and according to the state of the transmission line, either a correction value according to the state of the transmission line or zero A value is selected, and the correction term used in the row processing is set to the selected value.

  Further, a decoding device that decodes a low-density parity check-encoded received signal received via a transmission path includes: a column calculation unit that executes a column process with a log likelihood ratio calculated from the received signal as an input; A row calculation unit that performs row processing according to a δ-min algorithm that performs correction using a correction term having a calculated correction value, and sets any one of the correction value and zero according to the state of the transmission path. Selecting, and setting the correction term used in the row processing to the selected value.

  According to at least one embodiment of the present disclosure, since a zero value is selected as a correction value according to the state of the transmission line, calculation of an unselected correction value is unnecessary, and power consumption in the correction value calculation circuit is reduced. It becomes possible. In addition, since it is not necessary to calculate correction values that are not selected, it is possible to reduce the size of a table used for correction value calculation.

It is a figure which shows the structure of the 1st Example of a decoding apparatus. It is a figure which shows the structural example of a selector. It is a figure which shows an example of a structure of the minimum value calculation part. It is a flowchart which shows the line process of a line calculating part. It is a figure for demonstrating the setting of the threshold value used with an operation control part and a selector. It is a figure which shows the structure of the 2nd Example of a decoding apparatus.

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

  Several methods have been proposed as methods for reducing the amount of calculation when decoding an LDPC code. There is a min-sum BP (Belief-Propagation) method as a method that greatly reduces the error correction capability but greatly reduces the amount of calculation. As a method for improving the characteristics of the Min-sum BP method, a method such as the offset min-sum BP method has been proposed. However, it is necessary to set an optimal variable, and there is a problem that this setting is difficult. Therefore, a method has been proposed in which the error correction capability is not deteriorated as much as the min-sum BP method and a variable need not be determined.

  In a basic algorithm for decoding an LDPC-encoded received signal, decoding is performed by repeatedly performing a logarithmic likelihood ratio LLR (Log Likelihood Ratio) that is probabilistic reliability information of the received signal. Do. Specifically, in a decoding apparatus that decodes a low-density parity check-encoded received signal received via a transmission path, a log likelihood ratio is calculated from the received signal, and the column operation unit calculates the log likelihood ratio. Perform column processing as input. The line processing unit executes line processing. These column processing and row processing are repeated, and decoding processing is performed by calculating and updating the LLR. A sum-product algorithm is known as such an algorithm. In addition, the min-sum algorithm that approximates the row processing of the sum-product algorithm, and further improves the performance over the min-sum algorithm by correcting the minimum value of the absolute value of the LLR for row processing to the optimum value using the correction term. A known δ-min algorithm is known.

の場合、

となる。ここで、Ｃは定数、ｙは受信信号、ａ，ｂは受信信号から求められる行演算部と列演算部の値で使用される確率の値、σ^２は伝送路の雑音の分散の値である。また上記の演算された補正値とは、数１のｍａｘ関数の中にある２項目のｍａｘ関数の値（ｍａｘ（Ｃ−||ａ|−|ｂ||／２，０））のことである。定数Ｃは、例えば、特許文献１に示されるように０．９であってよい。δ−ｍｉｎアルゴリズムは公知であり、その復号処理の具体的内容については、例えば特許文献１及び非特許文献２に詳しく記載されている。 As described in Non-Patent Document 2, the formula of the row calculation unit of the δ-min algorithm that performs correction using a correction term having a calculated correction value is an LLR calculation formula:

in the case of,

It becomes. Here, C is a constant, y is a received signal, a and b are values of probabilities used in the values of the row calculation unit and the column calculation unit obtained from the received signal, and σ ² is a noise variance value of the transmission path. is there. The calculated correction value is the value (max (C− || a | − | b || 2, 0)) of the two items of the max function in the max function of Equation 1. is there. The constant C may be 0.9 as shown in Patent Document 1, for example. The δ-min algorithm is publicly known, and specific contents of the decoding process are described in detail in Patent Document 1 and Non-Patent Document 2, for example.

  In the case where the transmission path is hardly changed in time by improving the above method, the log likelihood ratio (by calculating the variance value of noise indicating the state of the received signal and the transmission path as a specific value ( A method of reducing the amount of calculation to be converted into (LLR) is conceivable. In this method, the LLR value is determined from the received signal regardless of the noise variance value. Therefore, when the correction term used in the row calculation unit of LDPC decoding is obtained, the value of the noise variance of the transmission path is used so that the LLR is calculated. Instead of calculating the LLR, it is necessary to use the noise dispersion value of the transmission line in order to obtain the correction value in the row calculation unit of the LDPC code, but it is not necessary to perform the calculation in bit units as in the LLR. Therefore, the effect of reducing the amount of calculation can be obtained.

となる。この場合、補正値の算出の仕方が変更されるだけで、他の演算に関しては、非特許文献２と同様の演算で求めることができる。 When the calculation of LLR is simplified, the above equation for the row calculation unit is

It becomes. In this case, only the method of calculating the correction value is changed, and other calculations can be obtained by the same calculation as in Non-Patent Document 2.

In this case, in a decoding device that decodes a low-density parity check-encoded received signal received via a transmission path, a column calculation unit performs column processing with the received signal as an input. Further, the row processing unit executes the row processing of the above formula 3. This line process, in the row processing based on the [delta]-min algorithm, by including the variance sigma ² of the noise of the transmission path, certain variance sigma ² of the noise of the transmission path when obtaining the log likelihood ratio from the received signal The algorithm is replaced with a value (σ ² = 2 in the case of Equation 1). Specifically, in this row processing, the transmission path state σ ² is input, and a correction value max (C × σ ² / 2− || a | − | b || / 2) corresponding to the transmission path state. Correction is performed with a correction term having 0). By repeating these column processing and row processing, the scaled LLR is calculated and updated to perform decoding processing. As will be described later, in the embodiment of the decoding apparatus, a correction term used in row processing is selected by selecting either a correction value or zero according to the state of the transmission path according to the state of the transmission path. Is set to the selected value.

  FIG. 1 is a diagram illustrating a configuration of a first embodiment of a decoding apparatus in a case where a location where an element where a row and a column intersect is 1 is 3 in the row direction of the parity check matrix. The decoding device 10 in FIG. 1 includes a column calculation unit 11, a row calculation unit 12, and a RAM 13. The column calculation unit 11 receives the received signal, executes the column processing calculation, and stores the result of the column processing calculation in the RAM 13. Based on the values a and b stored in the RAM 13, the row calculation unit 12 executes the row processing calculation of Equation 3 and stores the result of the row processing calculation in the RAM 13. By repeating this column processing and row processing, the column calculation unit 11 outputs a decoding result (a reception signal that has been decoded and corrected for errors).

The row calculation unit 12 includes an operation control unit 21, a table 22, a correction value calculation unit 23, a zero output unit 24, a selector 25, a minimum value calculation unit 26, and a correction calculation unit 27. The table 22 receives, for example, the noise dispersion value σ ² of the transmission line as the state of the transmission line, and outputs a value k corresponding to the dispersion value σ ² . This value k is equivalent to C × σ ^2/2 in the equation (3). The correction value calculation unit 23 calculates max (−α × || a | − | b || k, 0) based on the values a and b read from the RAM 13 and the value k from the table 22. Here, α is −1/2, and the calculated value of the correction value calculation unit 23 is equal to the correction value max (C × σ ² / 2− || a | − | b || 2, 0).

The selector 25 receives, for example, the noise dispersion value σ ² of the transmission line as the state of the transmission line, selects the calculation value (correction value) or the value 0 of the correction value calculation unit 23 according to the state of the transmission line, and corrects the correction term. Output as t. Specifically, when the value indicating the state of the transmission path (the value corresponding to the magnitude of the noise) is lower than a predetermined threshold and the state of the transmission path is good, the selector 25 selects 0. In addition, when the value indicating the state of the transmission line (value corresponding to the magnitude of the noise) is equal to or greater than a predetermined threshold and the state of the transmission line is defective, the selector 25 calculates the correction value (correction value) of the correction value calculation unit 23. Value). This is because, in the row processing calculation of the row calculation unit 12 (the calculation of Equation 3), if the transmission path state is good, the decoding is not performed (that is, even if the correction value is zero). This is because the process converges and exhibits sufficient error correction capability.

  FIG. 2 is a diagram illustrating a configuration example of the selector 25. The selector 25 includes a comparator 31 and a selection circuit 32. The comparator 31 receives the value indicating the state of the transmission path and the threshold value, compares the value indicating the state of the transmission path with the threshold value, and outputs a selection signal having a signal level corresponding to the comparison result. The selection circuit 32 selects either the output calculation value of the correction value calculation unit 23 or the output zero value of the zero output unit 24 according to the signal level of the selection signal.

  Referring to FIG. 1 again, the minimum value calculation unit 26 calculates and outputs s = Min (| a |, | b |). The correction calculation unit 27 calculates Max (s−t, 0) based on the value s from the minimum value calculation unit 26 and the correction term t. This calculated value is the calculated value of Equation 3 when the correction term t is equal to the calculated value (corrected value) of the corrected value calculating unit 23. A value using the calculation result of the correction calculation unit 27 is stored in the RAM 13 as an output of the row calculation unit 12.

The operation control unit 21 receives, for example, the noise variance value σ ² of the transmission line as the state of the transmission line, and controls the operation of the table 22 and the correction value calculation unit 23 according to the state of the transmission line. Specifically, when the value indicating the state of the transmission path (value corresponding to the magnitude of the noise) is lower than a predetermined threshold and the state of the transmission path is good, the operation control unit 21 includes the table 22 and The operation of the correction value calculation unit 23 is stopped. When the value indicating the state of the transmission line (value corresponding to the magnitude of noise) is equal to or greater than a predetermined threshold and the state of the transmission line is defective, the operation control unit 21 includes the table 22 and the correction value calculation unit 23. To work. This is because the correction term is not used when the state of the transmission path is good, and the operations of the table 22 and the correction value calculation unit 23 are unnecessary. By stopping the operation of the table 22 and the correction value calculator 23, the power consumption can be reduced. Note that a large number of row calculation units 12 are actually provided, and a plurality of row calculation units 12 operate simultaneously in parallel. Therefore, the effect of reducing the power consumption by stopping the operation of the table 22 and the correction value calculation unit 23 is great.

  FIG. 3 is a diagram illustrating an example of the configuration of the minimum value calculation unit 26. The minimum value calculation unit 26 includes a comparator 33. The comparator 33 receives a value (for example, noise power) indicating the state of the transmission path and a threshold, compares the value indicating the state of the transmission path with the threshold, and outputs an operation control signal having a signal level corresponding to the comparison result. Output. The operations of the table 22 and the correction value calculation unit 23 are controlled according to the signal level (for example, HIGH / LOW) of the operation control signal.

  FIG. 4 is a flowchart showing the row processing of the row calculation unit 12. In step S1, the selector 25 of the row calculation unit 12 determines whether or not the noise variance of the transmission line is smaller than the threshold value. If the noise variance of the transmission line is smaller than the threshold value, 0 is output as the correction term t by the selector 25 of the row calculation unit 12 in step S2. If the noise variance of the transmission line is greater than or equal to the threshold value, an optimum value k corresponding to the noise variance of the transmission line is output from the table 22 of the row calculation unit 12 in step S3. Further, in step S4, a correction value is calculated by the correction value calculation unit 23 of the row calculation unit 12, and this correction value is selected as a correction term t by the selector 25 and output. Finally, in step S5, the correction calculation unit 27 of the row calculation unit 12 calculates the update value Max (s−t, 0) using the correction term t.

Note that the index indicating the state of the transmission path that is an input to the operation control unit 21 and the selector 25 is not limited to the noise dispersion value σ ² of the transmission path. For example, CNR (carrier-to-noise ratio) or the like may be used as an index indicating the state of the transmission path. Further, instead of the table 22, for example, an arithmetic circuit that calculates an output value based on an input value may be used.

  FIG. 5 is a diagram for explaining setting of threshold values used in the operation control unit 21 and the selector 25. The graph of FIG. 5 shows the state of the transmission path on the horizontal axis, and the vertical axis shows the error probability of the signal after the decoding apparatus 10 decodes and corrects the received signal. The characteristic curve 41 shows the error probability of the decoded signal obtained by the decoding apparatus 10 when the correction term t is 0, that is, when the transmission path state is good and the selector 25 selects the output 0 of the zero output unit 24. The values are shown for the value of the state of each transmission line. In the characteristic curve 42, when the correction term t is not 0, that is, when the state of the transmission path is bad and the selector 25 selects the correction value calculated by the correction value calculation unit 23, the decoded signal obtained by the decoding apparatus 10 is obtained. The error probability values are shown for the state values of the respective transmission lines.

  A dotted line 43 extending in the horizontal direction indicates a certain error probability value, and this error probability value is performance required for the decoding apparatus 10. The error probability of the decoded signal of the decoding apparatus 10 needs to exceed the required performance indicated by the dotted line 43 with respect to a wide range of transmission path states including cases where the transmission path state is good and bad. There is. In the graph of FIG. 5, since the error probability on the vertical axis is higher on the upper side and lower on the lower side, the error is less below the dotted line 43 and exceeds the required performance. For this purpose, for example, a threshold value may be set to a value indicated by a dotted line 44 extending in the vertical direction. In this case, the characteristic curve 41 is used on the right side of the dotted line 44 indicating the threshold value, and the characteristic curve 42 is used on the left side of the dotted line 44 indicating the threshold value. Therefore, the maximum power reduction effect can be exhibited while satisfying the required performance for a wide range of transmission path conditions.

In FIG. 5, since the characteristic curve 41 is used on the right side of the dotted line 44 indicating the threshold value, a value k corresponding to the variance value σ ² needs to be stored in the table 22 for the area on the right side of the dotted line 44. There is no. Therefore, the size of the table 22 can be reduced as compared with the configuration in which the characteristic curve 42 is always used, that is, the configuration in which the correction term t that is not always 0 is used. As described above, since a large number of row calculation units 12 are provided, the effect of reducing the circuit scale by reducing the size of the table 22 (reducing the memory circuit scale) is great.

  FIG. 6 is a diagram illustrating the configuration of a second embodiment of the decoding device having the same parity check matrix as FIG. The decoding device 10A in FIG. 6 includes a column calculation unit 11A, a row calculation unit 12A, and a RAM 13A. The column calculation unit 11A receives the log likelihood ratio LLR obtained from the received signal, executes the column processing calculation, and stores the result of the column processing calculation in the RAM 13A. Based on the values a and b stored in the RAM 13A, the row calculation unit 12A executes the row processing calculation of Formula 2 and stores the result of the row processing calculation in the RAM 13A. By repeating this column processing and row processing, the column calculation unit 11 outputs a decoding result (a reception signal that has been decoded and corrected for errors).

  The row calculation unit 12A includes an operation control unit 21, a constant output unit 22A, a correction value calculation unit 23A, a zero output unit 24, a selector 25, a minimum value calculation unit 26, and a correction calculation unit 27. In FIG. 6, the same components as those in FIG. 1 are referred to by the same numerals, and a description thereof will be omitted as appropriate.

  The constant output unit 22A outputs a constant value k. This value k corresponds to C in Equation 2 above. The correction value calculation unit 23A calculates max (−α ′ × || a | − | b || + k, 0) based on the values a and b read from the RAM 13A and the value k from the constant output unit 22A. To do. Here, α ′ is −1/2, and the calculated value of the correction value calculation unit 23 is equal to the correction value max (C− || a | − | b || 2, 0).

The selector 25 receives, for example, the noise dispersion value σ ² of the transmission line as the state of the transmission line, selects the calculation value (correction value) or the value 0 of the correction value calculation unit 23 according to the state of the transmission line, and corrects the correction term. Output as t. The minimum value calculation unit 26 calculates and outputs s = Min (| a |, | b |). The correction calculation unit 27 calculates Max (s−t, 0) based on the value s from the minimum value calculation unit 26 and the correction term t. This calculated value is the calculated value of Formula 2 when the correction term t is equal to the calculated value (corrected value) of the corrected value calculating unit 23. The operations of the selector 25, the minimum value calculator 26, and the correction calculator 27 are the same as in the first embodiment. The calculation result of the correction calculation unit 27 is stored in the RAM 13A as the output of the row calculation unit 12A.

The operation control unit 21 receives, for example, the noise dispersion value σ ² of the transmission line as the state of the transmission line, and controls the operation of the correction value calculation unit 23A according to the state of the transmission line. Specifically, when the value indicating the state of the transmission line (value corresponding to the magnitude of noise) is lower than a predetermined threshold and the state of the transmission line is good, the operation control unit 21 calculates a correction value. The operation of the unit 23A is stopped. If the value indicating the state of the transmission line (value corresponding to the magnitude of noise) is equal to or greater than a predetermined threshold and the state of the transmission line is defective, the operation control unit 21 operates the correction value calculation unit 23A. . Power consumption can be reduced by stopping the operation of the correction value calculation unit 23A. Note that a large number of row calculation units 12A are actually provided, and a plurality of row calculation units 12A operate simultaneously in parallel. Therefore, the effect of reducing power consumption by stopping the operation of the correction value calculation unit 23A is great.

  As mentioned above, although this invention was demonstrated based on the Example, this invention is not limited to the said Example, A various deformation | transformation is possible within the range as described in a claim.

DESCRIPTION OF SYMBOLS 10 Decoding apparatus 11 Column calculating part 12 Row calculating part 13 RAM
21 Operation Control Unit 22 Table 23 Correction Value Calculation Unit 24 Zero Output Unit 25 Selector 26 Minimum Value Calculation Unit 27 Correction Operation Unit

Claims (5)

A decoding device that decodes a low-density parity check-encoded received signal received via a transmission line,
A column operation unit that performs column processing with the received signal as an input;
In a line processing based on the δ-min algorithm, when the logarithmic likelihood ratio is obtained from the received signal by including the noise variance of the transmission channel, an algorithm in which the value of the noise variance is a specific value is used. And a row operation unit that executes a row process for performing correction using a correction term having a correction value corresponding to the state of the transmission line, and according to the state of the transmission line according to the state of the transmission line. A decoding apparatus, wherein either a correction value or zero is selected, and the correction term used in the row processing is set to the selected value.   The decoding apparatus according to claim 1, wherein the row calculation unit includes a circuit for obtaining the correction value, and stops the operation of the circuit when the selected value is zero.   3. The decoding apparatus according to claim 2, wherein the circuit includes a table for inputting the state of the transmission path and outputting the correction value. A decoding device that decodes a low-density parity check-encoded received signal received via a transmission line,
A column calculation unit that executes a column process with the log likelihood ratio calculated from the received signal as an input;
A row calculation unit that performs row processing by a δ-min algorithm that performs correction using a correction term having a calculated correction value, and selects either the correction value or zero according to the state of the transmission path And the correction term used in the row processing is set to the selected value.   5. The decoding apparatus according to claim 4, wherein the row calculation unit includes a circuit that calculates the correction value, and stops the operation of the circuit when the selected value is zero.

Images