JP2012209632A - Minimum value detection circuit and minimum value detection method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a minimum value detection circuit and a minimum value detection method that can quickly detect a minimum value and the next minimum value, which is next to the minimum value, from a plurality of data values.SOLUTION: A first data value and a second data value are compared in magnitude and the smaller and larger are regarded as first and second values, respectively, and a third data value and a fourth data value of individual data values are compared in magnitude and the smaller and larger are regarded as third and fourth values, respectively. The smaller of the first and third values is regarded as a minimum value of the first to forth data values. The larger of the first and third values is regarded as a first next minimum value candidate, and if the first value is smaller than the third value, the second value is regarded as a second next minimum value candidate, while if the first value is larger than the third value, the fourth value is regarded as a second next minimum value candidate. The smaller of the first and second next minimum value candidates is the next minimum value of the first to fourth data values.

Description

  The present invention relates to a minimum value detection circuit and a minimum value detection method for detecting a minimum value and a next minimum value next to the minimum value from a data series having a finite number of data.

  In recent years, a low density parity check code proposed by Gallager, so-called LDPC (low-density parity-check code), has attracted attention as an error correction code employed when constructing a data communication system. In the encoding process by LDPC, a certain number of bits of logical level 1 are included in each row and each column of information data to be transmitted, and a parity check matrix having logical level 0 is generated for the others. Is added to the information data as code data. Decoding of such code data is performed by message passing by probability propagation on a so-called Tanner graph having a shape in which a variable node corresponding to each bit of the code word and a check node corresponding to a parity check sum are connected by branches. Implemented based on algorithm. As a message passing algorithm, a Sum-Product algorithm and a Min-Sum algorithm are known.

  As one of the Sum-Product algorithms, a parity check is performed while updating the external value log ratio αmn according to the Gallager function f based on the log likelihood ratio λn and the prior value log ratio βmn corresponding to the received symbol. It is known (see paragraphs [0036] to [0061] of Patent Document 1). In the decoding process based on the Sum-Product algorithm, the prior value log ratio βmn is updated according to the external value logarithmic ratio αmn described above, and a plurality of codes are used by using the external value logarithmic ratio αmn and the log likelihood ratio λn. Generated primary estimated words are generated, and parity check of these primary estimated words is performed. If the syndrome does not become 0 by the parity check, it is determined that there is an error in the primary estimation word, and the above processing is repeatedly executed. When the number of repetitions of this process reaches a predetermined value, the primary estimated word at that time is output as the final decoded word.

  In addition, a Min-Sum algorithm has been proposed as a process for realizing a process according to the Gallager function f using only a simple mathematical expression approximating a function used in the Sum-Product algorithm. As one of the Min-Sum algorithms, an external value logarithmic ratio αmn is set to a minimum value among a plurality of variable values represented by the Gallagher function f and a minimum value (hereinafter referred to as the next minimum value) next to the minimum value. (Refer to the description of paragraphs [0063] to [0082] of Patent Document 1). Accordingly, in order to decode LDPC data using this Min-Sum algorithm, a minimum value detection circuit for detecting the minimum value and the next minimum value from a series of a plurality of variable values is required.

  In this case, in order to detect the next minimum value from a series of variable values, first, the minimum value is detected from the series of variable values (first stage), and the variable value corresponding to the minimum value is masked. In this state, the minimum value is detected again from the series of variable values (second stage).

  Therefore, every time the absolute value Rm of the external value logarithmic ratio αmn is updated, the above-described two-stage minimum value detection process must be executed, which causes a problem that the error correction process takes time.

JP 2007-306495 A

  The present invention has been made to solve the above-described problem, and is a minimum capable of detecting a minimum value and a next minimum value next to the minimum value among a plurality of data values at high speed. A value detection circuit and a minimum value detection method are provided.

  A minimum value detection circuit according to the present invention is a minimum value detection circuit that detects a minimum value and a next minimum value next to the minimum value from among a plurality of data values, and includes a first value of the data values. A first comparison unit that compares the data value of the first data value with the second data value and sets a smaller value as a first value and a larger value as a second value; and a third data value of the data values; A second comparison unit that compares the fourth data value with the third value and sets the smaller value as the third value and the larger value as the fourth value, and the smaller of the first value and the third value. The minimum value among the first to fourth data values is set, and the larger one of the first value and the third value is set as a first next minimum value candidate, and the first value is the third value. If it is smaller, the second value is set as the second next minimum candidate, while if the first value is larger than the third value, the fourth value is changed to the second value. A determination unit that determines a minimum value candidate and sets the smaller one of the first and second next minimum value candidates as the next minimum value in the first to fourth data values. .

  A minimum value detection method according to the present invention is a minimum value detection method for detecting a minimum value and a next minimum value next to the minimum value from among a plurality of data values, respectively. The first data value and the second data value are compared in magnitude, and the smaller one is the first value, the larger one is the second value, and the third data value and the fourth data value of each of the data values are the fourth value. A first step in which the smaller value is compared with the third data value and the smaller value is the third value and the larger value is the fourth value, and the smaller one of the first value and the third value is the first value. The minimum value in the fourth data value is set, and the larger one of the first value and the third value is set as the first next minimum value candidate, and the first value is smaller than the third value. In this case, the second value is set as a second next minimum candidate, and when the first value is greater than the third value, the fourth value is set to the second value. A second step of setting a smaller one of the first and second next minimum value candidates as a next minimum value in the first to fourth data values; Have.

  In the present invention, the first data value and the second data value are compared in magnitude, and the smaller one is set as the first value, the larger one is set as the second value, and the third data value and the fourth data value are compared. Comparing the data values with each other, the smaller one is the third value, the larger one is the fourth value, and the smaller one of the first and third values is the smallest of the first to fourth data values. Value. Further, the larger one of the first value and the third value is set as the first next minimum value candidate, and when the first value is smaller than the third value, the second value is set as the second next minimum value. On the other hand, if the first value is larger than the third value, the fourth value is set as the second next minimum value candidate, and the first next minimum value candidate and the second next minimum value candidate are smaller. Is the next minimum value that is the next smallest value after the minimum value in the first to fourth data values.

  According to such minimum value detection processing, the minimum value is detected from each of the data values, and the minimum value is detected again from each of the data values in a state where the data value corresponding to the minimum value is masked. Compared to the conventional minimum value detection circuit that sets the minimum value, the minimum value and the next minimum value can be detected from a plurality of data values at high speed.

It is a block diagram which shows an example of a structure of the minimum value detection circuit by this invention. It is a block diagram which shows the internal structure of a comparator selector. It is a block diagram which shows the internal structure of the minimum value order minimum value determination part. It is a block diagram which shows another example of a structure of the minimum value detection circuit. FIG. 5 is a diagram showing a modification of the minimum value detection circuit shown in FIG. 4.

  In the present invention, the first data value and the second data value are compared in size, and the smaller one is set as the first value, and the larger one is set as the second value. The data value and the fourth data value are compared in size, and the smaller value is the third value, and the larger value is the fourth value. At this time, the smaller one of the first value and the third value is set as the minimum value in the first to fourth data values. Further, the larger one of the first value and the third value is set as the first next minimum value candidate, and when the first value is smaller than the third value, the second value is set as the second next minimum value. On the other hand, if the first value is larger than the third value, the fourth value is set as the second next minimum value candidate. At this time, the smaller one of the first and second next minimum value candidates is set as the next minimum value in the first to fourth data values.

  FIG. 1 is a block diagram showing a basic configuration of a minimum value detection circuit according to the present invention.

  In FIG. 1, both comparator selectors 11 and 12 have an internal configuration as shown in FIG.

  In FIG. 2, the comparator 101 compares the value of the data supplied to the input terminal a with the value of the data supplied to the input terminal b, and generates a comparison result signal CM1 corresponding to the size comparison result. Supply to selector 102. The selector 102 receives the data supplied to the input terminal a when the comparison result signal CM1 indicating that the data value supplied to the input terminal a is smaller of the input terminals a and b is supplied. The value is output as the temporary minimum value M1, and the data value supplied to the input terminal b is output as the temporary minimum value M2. On the other hand, when the comparison result signal CM1 indicating that the data value supplied to the input terminal b is smaller among the input terminals a and b is supplied, the selector 102 is supplied to the input terminal b. The output data value is output as the provisional minimum value M1, and the data value supplied to the input terminal a is output as the provisional next minimum value M2. When the data values supplied to the input terminals a and b are the same value, the selector 102 outputs the values as the temporary minimum value M1 and the temporary next minimum value M2.

With this configuration, the comparator selector 11 supplies the smaller _one of the data values X ₀ and X ₁ as the provisional minimum value M 1 _{A to} the input terminal _{A 1} of the minimum value / next minimum value determination unit 21 and the data value X ₀ and supplies it to the input terminal A2 of the minimum and the next minimum value determining unit 21 a large becomes better among X ₁ as a provisional next minimum value M2 _a.

The comparator selector 12 supplies the smaller one of the data values X ₂ and X ₃ as the provisional minimum value M 1 _{B to} the input terminal _{B 1} of the minimum value / next minimum value determination unit 21 and the data values X ₃ and X _3. supplied to the input terminal B2 of the minimum and the next minimum value determining unit 21 a large becomes better among the provisional next minimum value M2 _B.

  The minimum value / next minimum value determination unit 21 has an internal configuration as shown in FIG.

In FIG. 3, the comparator 201 compares the data value supplied to the input terminal A1 with the data value supplied to the input terminal B1, and selects the comparison result signal CM2 corresponding to the size comparison result with the selector 203. To each of -205. The selector 203 is supplied to the input terminal A1 when the comparison result signal CM2 indicates that the data value supplied to the input terminal A1 is smaller than the data value supplied to the input terminal B1. the data value, and outputs the smallest becomes the value in the four data values X ₀ to X ₄ as temporary minimum value M1 shown tentatively. On the other hand, if the comparison result signal CM2 indicates that the data value supplied to the input terminal B1 is smaller than or equal to the data value supplied to the input terminal A1, the selector 203 Outputs the data value supplied to the input terminal B1 as the provisional minimum value M1.

That is, the selector 203 shows the smallest value among the data values X _{0 to} X ₃ whichever is smaller between the data value supplied to the input terminal A 1 and the data value supplied to the input terminal B 1. It is output as the minimum value M1.

  When the comparison result signal CM2 indicates that the data value supplied to the input terminal A1 is greater than or equal to the data value supplied to the input terminal B1, the selector 204 follows the data value supplied to the input terminal A1. The minimum value candidate M2a is supplied to each of the comparator 202 and the selector 206. On the other hand, when the comparison result signal CM2 indicates that the data value supplied to the input terminal B1 is larger than the data value supplied to the input terminal A1, the selector 204 is connected to the input terminal B1. The supplied data value is supplied to each of the comparator 202 and the selector 206 as the next minimum value candidate M2a.

  That is, the selector 204 supplies the larger one of the data value supplied to the input terminal A1 and the data value supplied to the input terminal B1 to the comparator 202 and the selector 206 as the next minimum value candidate M2a. To do.

  When the comparison result signal CM2 indicates that the data value supplied to the input terminal B1 is less than or equal to the data value supplied to the input terminal A1, the selector 205 sets the data value supplied to the input terminal B2 to the selector 205. The next minimum value candidate M2b is supplied to each of the comparator 202 and the selector 206. On the other hand, when the comparison result signal CM2 indicates that the data value supplied to the input terminal B1 is larger than the data value supplied to the input terminal A1, the selector 205 is supplied to the input terminal A2. The obtained data value is supplied to each of the comparator 202 and the selector 206 as the next minimum value candidate M2b.

  The comparator 202 compares the next minimum value candidate M2a with the next minimum value candidate M2b, and supplies a comparison result signal CM3 indicating the result of the size comparison to the selector 206.

  When the comparison result signal CM3 indicates that the next minimum value candidate M2b is smaller than the next minimum value candidate M2a, the selector 206 selects the next minimum value candidate M2b from the next minimum value candidates M2a and M2b. This is selected and output as the provisional next minimum value M2. On the other hand, when the comparison result signal CM3 indicates that the next minimum value candidate M2a is smaller than the next minimum value candidate M2b, the selector 206 selects the next minimum value candidate M2a from the next minimum value candidates M2a and M2b. Is selected and output as the provisional next minimum value M2.

That is, the selector 206, the one to be small within the next minimum value candidate M2a and next minimum value candidate M2b, provisional next showing tentatively small becomes value to the second among the data values X ₀ to X ₃ It is output as the minimum value M2.

  As described above, the minimum value / next minimum value determination unit 21 based on the magnitude relationship between the data values supplied to the input terminals A1, A2, B1 and B2, as described below, the provisional minimum value M1 and the next minimum value candidate. Each of M2a and M2b and provisional next minimum value M2 is detected.

M1: the smaller of A1 and B1 M2a: the larger of A1 and B1 M2b: A2 if A1 <B1, B2 if A1 > B1
M2: The smaller of M2a and M2b

Here, as shown in FIG. 1, the provisional minimum value M1 _A and provisional next minimum supplied from the comparator selector 11 are respectively input to the input terminals A1, A2, B1, and B2 of the minimum value / next minimum value determination unit 21. Each of the value M2 _A , the provisional minimum value M1 _B and the provisional next minimum value M2 _B supplied from the comparator selector 12 is supplied as follows.

A1: Provisional minimum value M1 _A
A2: Provisional minimum value M2 _A
B1: Provisional minimum value M1 _B
B2: Provisional next minimum value M2 _B

In this case, temporary minimum value M1 _A is small becomes better values for of the data value X ₀ and X _1, provisional next minimum value M2 _A in large becomes better value of the data value X ₀ and X ₁ is there. Also, temporary minimum value M1 _B is small becomes better values for of the data value X ₂ and X _3, the provisional next minimum value M2 _B is large becomes better value of the data value X ₂ and X ₃ .

Therefore, the minimum value / next minimum value determination unit 21 outputs the temporary minimum value M1 temporarily indicating the smallest value among the data values X _{0 to} X ₃ as the minimum value as it is, and also outputs the data value X ₀ to the provisional next minimum value M2 that indicates the small becomes value to the 2nd of the X ₃ directly outputs the next minimum value.

As described above, in the minimum value detection circuit shown in FIG. 1, first, the first comparison unit (11) first and second data values (X ₀ , X ₁ ) of each of the plurality of data values are compared with each other. Are compared to each other, and the smaller one is the first value (M1 _A ), and the larger one is the second value (M2 _A ). Further, the second comparison unit (12) compares the third and fourth data values (X ₂ , X ₃ ) with each other, the smaller one is the first value (M1 _B ), and the larger one is the first value. It is assumed to be binary (M2 _B ). Next, the minimum value / next minimum value determination unit (21) sets the smaller one of the first value and the third value as the minimum value in the first to fourth data values. Further, the minimum value / next minimum value determination unit (21) sets the larger one of the first value and the third value as the first next minimum value candidate (M2a), and the first value is the third value. When the value is smaller than the value, the second value is set as the second next minimum value candidate (M2b), while when the first value is larger than the third value, the fourth value is set as the second next minimum value candidate ( M2b). At this time, in the minimum value / next minimum value determination unit (21), the smaller of the first next minimum value candidate (M2a) and the second next minimum value candidate (M2b) is the first to fourth. It is the next minimum value in the data value.

Here, each comparator selector in the minimum value detection circuit shown in FIG. 1 includes a two-input comparator (101) for one stage and a two-input selector (102) for one stage as shown in FIG. Further, the minimum value / next minimum value determination unit includes two stages of 2-input comparators (201, 202) and two stages of 2-input selectors (203 to 206) as shown in FIG. At this time, if the processing time consumed by the two-input comparators (101, 201, 202) alone is T _C , and the processing time spent by the two-input selectors (102, 203-206) alone is T _S , the comparator selector 11, 12, the processing time _(T C ₊ T S), minimum and next minimum value determining unit 21 in the processing time _(2T C + 2T _S) is spent. That is, in the minimum value detection circuit shown in FIG. 1, the minimum value and the next minimum value are obtained by spending the processing time (3T _C + 3T _S ) for the four input data values X _{0 to} X ₃ . It is.

In contrast, in the conventional minimum value detecting circuit, first, detects a minimum value from the data value X ₀ to X _3, again the data value X ₀ to X while masking the data value X corresponding to this minimum value _The minimum value is detected from ₃ and this is set as the next minimum value. That is, in the conventional minimum value detecting circuit, spent during the masking time _(2T C + 2T _S), the data value X corresponding to the minimum value in detecting the minimum value from the data value X ₀ to X ₃ The total time (4T _C + 4T _S + T) of the processing time T _M to be detected and the processing time (2T _C + 2T _S ) when the minimum value is detected again from the data values X _{0 to} X ₃ while masking the data value X _The minimum value and the next minimum value are obtained through _M ).

Therefore, according to the minimum value detection circuit shown in FIG. 1, it is possible to detect the minimum value and the next minimum value from the four data values X _{0 to} X _{3 at} a higher speed than the conventional one.

In the minimum value detection circuit shown in FIG. 1, the minimum value and the next minimum value are detected from the four data values X _{0 to} X _3. By combining a plurality of these values, 4 · 2 ⁿ It is also possible to construct a minimum value detection circuit that detects a minimum value and a next minimum value from ⁻¹ (n is a natural number) data values.

FIG. 4 is a block diagram showing a configuration of a minimum value detection circuit that detects the minimum value and the next minimum value from the 16 data values X _{0 to} X ₁₅ .

  In FIG. 4, each of comparator selectors 11-18 has the same internal configuration, that is, each has the internal configuration shown in FIG. Further, each of the minimum value / next minimum value determination units 21 to 27 has the same internal configuration, that is, each has the internal configuration as shown in FIG.

The comparator selector 11 supplies the smaller _one of the data values X ₀ and X ₁ as the provisional minimum value M 1 _{A to} the input terminal _{A 1} of the minimum value / next minimum value determination unit 21 and the data values X ₀ and X _1. supplied to the input terminal A2 of the minimum and the next minimum value determining unit 21 a large becomes better among the provisional next minimum value M2 _a. Comparator selector 12 supplies to the input terminal B1 of the minimum and the next minimum value determining unit 21 towards made small among the data values X ₂ and X ₃ as temporary minimum value M1 _B, the data value X ₂ and X ₃ supplied to the input terminal B2 of the minimum and the next minimum value determining unit 21 a large becomes better among the provisional next minimum value M2 _B. Comparator selector 13 supplies to the input terminal A1 of the minimum and the next minimum value determining unit 22 towards made small among the data values X ₄ and X ₅ as temporary minimum value M1 _C, data value X ₄ and X ₅ supplied to the input terminal A2 of the minimum and the next minimum value determining unit 22 a large becomes better among the provisional next minimum value M2 _C. The comparator selector 14 supplies the smaller one of the data values X ₆ and X ₇ as the provisional minimum value M 1 _{D to} the input terminal B 1 of the minimum value / next minimum value determination unit 22, and the data values X ₆ and X _77. supplied to the input terminal B2 of the minimum and the next minimum value determining unit 22 a large becomes better among the provisional next minimum value M2 _D. Comparator selector 15 supplies to the input terminal A1 of the minimum and the next minimum value determining unit 23 to those who small among the data values X ₈ and X ₉ as temporary minimum value M1 _E, data value X ₈ and X ₉ supplied to the input terminal A2 of the minimum and the next minimum value determining unit 23 a large becomes better among the provisional next minimum value M2 _E. Comparator selector 16 supplies to the input terminal B1 of the minimum and the next minimum value determining unit 23 to those who small among the data values _{X 10} and _{X 11} as temporary minimum value M1 _F, the data value _{X 10} and _{X 11} supplied to the input terminal B2 of the minimum and the next minimum value determining unit 23 a large becomes better among the provisional next minimum value M2 _F. Comparator selector 17 supplies the data value _{X 12} and the input terminal A1 of the minimum and the next minimum value determining unit 24 small becomes better within a temporary minimum value M1 _G of _{X 13,} data value _{X 12} and _{X 13} supplied to the input terminal A2 of the minimum and the next minimum value determining unit 24 a large becomes better among the provisional next minimum value M2 _G. The comparator selector 18 supplies the smaller one of the data values X ₁₄ and X ₁₅ as the provisional minimum value M 1 _{H to} the input terminal B 1 of the minimum value / next minimum value determination unit 24, and the data values X ₁₄ and X _15. supplied to the input terminal B2 of the minimum and the next minimum value determining unit 24 a large becomes better among the provisional next minimum value M2 _H.

  Each of the minimum value / next minimum value determination units 21 to 27 is based on the magnitude relationship between the data values supplied to the input terminals A1, A2, B1 and B2, as follows: M2a and M2b, and provisional next minimum value M2 are obtained.

M1: the smaller of A1 and B1 M2a: the larger of A1 and B1 M2b: A2 if A1 <B1, B2 if A1 > B1
M2: The smaller of M2a and M2b

As shown in FIG. 4, each of the input terminals A1, A2, B1, and B2 of the minimum value / next minimum value determination unit 21 has a provisional minimum value M1 _A and a provisional next minimum value M2 _A supplied from the comparator selector 11. Each of the provisional minimum value M1 _B and provisional next minimum value M2 _B supplied from the comparator selector 12 is supplied as follows.

A1: Provisional minimum value M1 _A
A2: Provisional minimum value M2 _A
B1: Provisional minimum value M1 _B
B2: Provisional next minimum value M2 _B

In this case, temporary minimum value M1 _A is small becomes better values for of the data value X ₀ and X _1, provisional next minimum value M2 _A in large becomes better value of the data value X ₀ and X ₁ is there. Also, temporary minimum value M1 _B is small becomes better values for of the data value X ₂ and X _3, the provisional next minimum value M2 _B is large becomes better value of the data value X ₂ and X ₃ .

Therefore, the minimum value / next minimum value determination unit 21 sets the smallest value among the data values X _{0 to} X ₃ as the provisional minimum value M1 _J , and sets the second smallest value as the provisional next minimum value M2 _J. These are supplied to the input terminals A1 and A2 of the minimum value / next minimum value determination unit 25, respectively.

Further, as shown in FIG. 4, the provisional minimum value M1 _C and provisional next minimum supplied from the comparator selector 13 are respectively input to the input terminals A1, A2, B1, and B2 of the minimum value / next minimum value determination unit 22. Each of the value M2 _C , the provisional minimum value M1 _D and the provisional next minimum value M2 _D supplied from the comparator selector 14 is supplied as follows.

A1: Provisional minimum value M1 _C
A2: Provisional next minimum value M2 _C
B1: Provisional minimum value M1 _D
B2: Provisional next minimum value M2 _D

In this case, temporary minimum value M1 _C is small becomes better values for of the data value X ₄ and X _5, provisional next minimum value M2 _C at atmospheric made towards the value of the data value X ₄ and X ₅ is there. Also, temporary minimum value M1 _D is small becomes better values for of the data value X ₆ and X _7, provisional next minimum value M2 _D is large becomes better value of the data value X ₆ and X ₇ .

Therefore, the minimum value / next minimum value determination unit 22 sets the smallest value among the data values X _{4 to} X ₇ as the provisional minimum value M1 _K and the second smallest value as the provisional next minimum value M2 _K. These are supplied to the input terminals B1 and B2 of the minimum value / next minimum value determination unit 25, respectively.

Further, as shown in FIG. 4, the provisional minimum value M1 _E and provisional order minimum supplied from the comparator selector 15 are respectively input to the input terminals A1, A2, B1, and B2 of the minimum value / next minimum value determination unit 23. The value M2 _E , the provisional minimum value M1 _F and the provisional next minimum value M2 _F supplied from the comparator selector 16 are supplied as follows.

A1: Provisional minimum value M1 _E
A2: Provisional next minimum value M2 _E
B1: Provisional minimum value M1 _F
B2: Provisional next minimum value M2 _F

In this case, temporary minimum value M1 _E is small becomes better values for of the data value X ₈ and X _9, provisional next minimum value M2 _E is large becomes better value of the data value X ₈ and X ₉ is there. Also, temporary minimum value M1 _F is small becomes better values for of the data value _{X 10} and _{X 11,} provisional next minimum value M2 _F is large become better value of the data value _{X 10} and _{X 11} .

Therefore, the minimum value / next minimum value determination unit 23 sets the smallest value among the data values X _{8 to} X ₁₁ as the provisional minimum value M1 _L , and sets the second smallest value as the provisional next minimum value M2 _L. These are supplied to the input terminals A1 and A2 of the minimum value / next minimum value determination unit 26, respectively.

Further, as shown in FIG. 4, the provisional minimum value M1 _G and provisional order minimum supplied from the comparator selector 17 are respectively input to the input terminals A1, A2, B1, and B2 of the minimum value / next minimum value determination unit 24. Each of the value M2 _G , the provisional minimum value M1 _H and the provisional next minimum value M2 _H supplied from the comparator selector 18 is supplied as follows.

A1: Provisional minimum value M1 _G
A2: Provisional next minimum value M2 _G
B1: Provisional minimum value M1 _H
B2: Provisional minimum value M2 _H

In this case, temporary minimum value M1 _G is small becomes better values for of the data value _{X 12} and _{X 13,} provisional next minimum value M2 _G is large becomes better value of the data value _{X 12} and _{X 13} is there. Also, temporary minimum value M1 _H is small becomes better values for of the data value _{X 14} and _{X 15,} provisional next minimum value M2 _H is large becomes better value of the data value _{X 14} and _{X 15} .

Therefore, the minimum value / next minimum value determination unit 24 sets the smallest value among the data values X _{12 to} X ₁₅ as the provisional minimum value M1 _M , and sets the second smallest value as the provisional next minimum value M2 _M. These are supplied to the input terminals B1 and B2 of the minimum value / next minimum value determination unit 26, respectively.

Each of the input terminals A1, A2, B1, and B2 of the minimum value / next minimum value determination unit 25 has a provisional minimum value M1 _J and a provisional next minimum value M2 _J supplied from the minimum value / next minimum value determination unit 21, Each of the provisional minimum value M1 _K and the provisional next minimum value M2 _K supplied from the minimum value / next minimum value determination unit 22 is supplied as follows.

A1: Provisional minimum value M1 _J
A2: Provisional minimum value M2 _J
B1: Provisional minimum value M1 _K
B2: Provisional next minimum value M2 _K

In this case, temporary minimum value M1 _J is the smallest composed value of the data value _X 0 to X _3, provisional next minimum value M2 _J is small in the second-of the data value _X 0 to X ₃ Value. Also, temporary minimum value M1 _K is the smallest composed value of the data value _X 4 to X _7, provisional next minimum value M2 _H becomes small to the second among the data value _X 4 to X ₇ value It is.

Therefore, the minimum value / next minimum value determination unit 25 sets the smallest value among the data values X _{0 to} X ₇ as the provisional minimum value M1 _N , and sets the second smallest value as the provisional next minimum value M2 _N. These are supplied to the input terminals A1 and A2 of the minimum value / next minimum value determination unit 27, respectively.

Each of the input terminals A1, A2, B1, and B2 of the minimum value / next minimum value determination unit 26 has a provisional minimum value M1 _L and a provisional next minimum value M2 _L supplied from the minimum value / next minimum value determination unit 23, Each of provisional minimum value M1 _M and provisional next minimum value M2 _M supplied from minimum value / next minimum value determination unit 24 is supplied as follows.

A1: Provisional minimum value M1 _L
A2: Provisional next minimum value M2 _L
B1: Provisional minimum value M1 _M
B2: Provisional next minimum value M2 _M

At this time, the provisional minimum value M1 _L is the smallest value among the data values X _{8 to} X ₁₁ , and the provisional minimum value M2 _L is the second smallest among the data values X _{8 to} X _11. Value. Also, temporary minimum value M1 _M is the smallest composed value of the data value _X 12 _{to X 15,} provisional next minimum value M2 _M becomes small to the second among the data value _X 12 _{to X 15} value It is.

Therefore, the minimum value / next minimum value determination unit 26 sets the smallest value among the data values X _{8 to} X ₁₅ as the provisional minimum value M1 _N , and sets the second smallest value as the provisional next minimum value M2 _N. These are supplied to the input terminals B1 and B2 of the minimum value / next minimum value determination unit 27, respectively.

The input terminals A1, A2, B1, and B2 of the minimum value / next minimum value determination unit 27 have provisional minimum values M1 _N and provisional next minimum values M2 _N supplied from the minimum value / next minimum value determination unit 25, Each of provisional minimum value M1 _P and provisional next minimum value M2 _P supplied from minimum value / next minimum value determination unit 26 is supplied as follows.

A1: Provisional minimum value M1 _N
A2: Provisional next minimum value M2 _N
B1: Provisional minimum value M1 _P
B2: Provisional next minimum value M2 _P

At this time, the provisional minimum value M1 _N is the smallest value among the data values X _{0 to} X ₇ , and the provisional next minimum value M 2 _N is the second smallest among the data values X _{0 to} X _7. Value. Also, temporary minimum value M1 _P is the smallest composed value of the data value _X 8 _{to X 15,} provisional next minimum value M2 _P becomes small to the second among the data value _X 8 _{to X 15} value It is.

As a result, the minimum value / next minimum value determination unit 27 outputs the temporary minimum value M1 indicating the smallest value among the data values X _{0 to} X ₁₅ as the final minimum value, and also outputs the data value X ₀ to the provisional next minimum value M2 that indicates the small becomes value to the second among X ₁₅ to output as the final next minimum.

Here, in the minimum value detection circuit shown in FIG. 4, the processing time (T _C + T _S ) in the comparator selectors 11 to 18 and the processing time (2T _C + 2T _S ) in the minimum value / next minimum value determination units 21 to 24 are minimum. The value / next minimum value determination units 25 and 26 spend processing time (2T _C + 2T _S ), and the minimum value / next minimum value determination unit 27 spends processing time (2T _C + 2T _S ). Therefore, in the minimum value detection circuit shown in FIG. 4, the minimum value and the next minimum value can be obtained by spending the processing time (7T _C + 7T _S ) for the 16 input data values X _{0 to} X ₁₅ . It is.

In contrast, in the conventional minimum value detecting circuit, first, detects a minimum value from the data value X ₀ to X _15, again the data value X ₀ to X while masking the data value X corresponding to this minimum value _The minimum value is detected from ₁₅ and is set as the next minimum value. That is, in the conventional minimum value detecting circuit, spent during the masking time _(4T C + 4T _S), the data value X corresponding to the minimum value in detecting the minimum value from the data value X ₀ to X ₁₅ a processing time _{T M} which again data value while masking the data value X _X 0 _{to X 15} processing time for detecting a minimum value from among _(4T C + 4T _S) and the total time of _(8T C + 8T S + T _The minimum value and the next minimum value are obtained through _M ).

Therefore, according to the minimum value detection circuit shown in FIG. 4, it is possible to detect the minimum value and the next minimum value from among the 16 data values X _{0 to} X _{15 at} a higher speed than the conventional one. .

  Therefore, by mounting the minimum value detection circuit shown in FIG. 1 or FIG. 4 on an LDPC decoder that employs the Min-Sum algorithm, it is possible to speed up the decoding process.

That is, in the LDPC decoder employing the Min-Sum algorithm, parity check is performed while updating the external value log ratio according to the Gallgarger function using the log likelihood ratio and the prior value log ratio in consideration of noise information in the received symbol. . At this time, in such an LDPC decoder, the external value logarithm ratio is approximated by the minimum value and the next minimum value among a plurality of variable values represented by the Gallagher function. That is, a minimum value detection circuit as shown in FIG. 1 or FIG. 4 is used to detect the minimum value and the next minimum value from a plurality of variable values represented by the Gallagher function. For example, when the number of variable values represented by the Gallagher function is 16, the external value is based on the minimum value and the next minimum value detected by the minimum value detection circuit having the structure shown in FIG. The log ratio is obtained. When the number of variable values represented by the Gallagher function is 9 to 15, the minimum value detection circuit having the structure shown in FIG. One of the variable values is commonly supplied to the terminal a or b. For example, when the number of variable values represented by the Gallagher function is 12, as shown in FIG. 5, the data values X _{0 to} X ₁₁ corresponding to the variable values are respectively compared with the comparator selectors 11 to 11. 16 is supplied to each of the input terminals a and b, the remaining comparators selectors 17 and 18 each of the input terminals a and b, and supplies the data value X ₁₁ in common.

  In the above-described embodiment, the operation when the detection processing of the minimum value and the next minimum value is realized by hardware as shown in FIGS. 1 to 5 has been described. However, the operation shown in FIGS. You may make it implement | achieve in. However, by realizing the hardware as shown in FIGS. 1 to 5, high-speed processing can be performed as compared with the case where software is constructed.

11-18 Comparator selectors 21-27 Minimum value / next minimum value determination units 101, 201, 202 Comparator 102, 203-206 selector

Claims (6)

A minimum value detection circuit for detecting a minimum value and a next minimum value next to the minimum value among a plurality of data values,
A first comparison unit that compares the first data value and the second data value of the data values, and sets a smaller value as a first value and a larger value as a second value;
A second comparison unit that compares the third data value and the fourth data value of the data values and sets a smaller value as a third value and a larger value as a fourth value;
The smaller one of the first value and the third value is set as the minimum value in the first to fourth data values, and the larger one of the first value and the third value is the first value. If the first value is smaller than the third value, the second value is the second next minimum candidate, while the first value is greater than the third value. In this case, the fourth value is set as the second next minimum value candidate, and the smaller of the first next minimum value candidate and the second next minimum value candidate is the first to fourth data. And a determination unit for determining a next minimum value in the value.   2. The minimum value detection circuit according to claim 1, wherein the data values are a plurality of variable values represented by a Gallagher function in a decoder that decodes a low density parity check code according to a Min-Sum algorithm. A minimum value detection circuit that detects a minimum value and a next minimum value next to the minimum value from among 4n (n is an integer of 2 or more) data values,
The 4n data values are grouped into two 2n groups, and the data values are compared for each group. The smaller one is the provisional minimum value, and the larger one is the provisional next minimum value. 2n comparators to output,
The 2n comparison units are grouped into two n groups, and for each group, the provisional minimum value and the provisional minimum value output from one of the comparison units are respectively set to a first value and a first value. The temporary minimum value and the temporary next minimum value output from the other of the comparison unit are set as a third value and a fourth value, respectively, and the minimum value and the next value are selected from the first value to the fourth value. N first-stage determination units that determine a minimum value and relay-output the minimum value and the next minimum value as a temporary minimum value and a temporary next minimum value, respectively;
The n first-stage determination units are grouped into two n / 2 groups, and the provisional minimum value and the provisional minimum value output from one of the first-stage determination units are respectively determined for each group. 1st value and 2nd value, and the provisional minimum value and the provisional next minimum value output from the other of the first stage determination unit are respectively set as a 3rd value and a 4th value. N / 2 next-stage determination units that determine the minimum value and the next minimum value from the above, and relay output the minimum value and the next minimum value as the temporary minimum value and the temporary next minimum value, respectively,
Each of the first stage determination unit and the next stage determination unit sets the smaller one of the first value and the third value as the provisional minimum value, and among the first value and the third value, The larger one is set as the first next minimum value candidate, and when the first value is smaller than the third value, the second value is set as the second next minimum value candidate, while the first value is When the value is larger than the third value, the fourth value is set as the second next minimum value candidate, and the smaller of the first next minimum value candidate and the second next minimum value candidate is the provisional value. A minimum value detection circuit characterized by having a next minimum value.   3. The minimum value detection circuit according to claim 2, wherein the data value is a plurality of variable values represented by a Gallagher function in a decoder that decodes a low density parity check code according to a Min-Sum algorithm. A minimum value detection method for detecting a minimum value and a next minimum value next to the minimum value from a plurality of data values, respectively,
The first data value and the second data value of each of the data values are compared in magnitude, and the smaller one is set as the first value and the larger one is set as the second value. A first step in which a third data value and a fourth data value are compared in magnitude and the smaller value is the third value and the larger data value is the fourth value;
The smaller one of the first value and the third value is set as the minimum value in the first to fourth data values, and the larger one of the first value and the third value is the first value. If the first value is smaller than the third value, the second value is the second next minimum candidate, while the first value is greater than the third value. In this case, the fourth value is set as the second next minimum value candidate, and the smaller of the first next minimum value candidate and the second next minimum value candidate is the first to fourth data. And a second step of setting a next minimum value in the value.   6. The minimum value detecting method according to claim 5, wherein the data value is a plurality of variable values represented by a Gallagher function in a decoder that decodes a low density parity check code according to a Min-Sum algorithm.

Images