JP2006314055A - Turbo decoding method and apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To select an optimal likelihood update method that corresponds to the encoding rate, and to improve error correction ability, without making a circuit complex, as compared with the case in which likelihood information is not updated. <P>SOLUTION: Likelihood information update parts 103 and 107 compare agreement and disagreement in each logical value level of communication path value likelihood information and a value after the fact likelihood information. In the case of the disagreement, outside value likelihood information to be exchanged among a plurality of element decoders is selectively switched and updated, according to an encoding rate in the error correction encoding. Thus, even in a communication system whose encoding rate is variable, the optimal likelihood updating method can be selected, according to the encoding rate, so that turbo decoding can be realized of high error correction ability at any encoding rate. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a turbo decoding method and apparatus that perform iterative decoding while exchanging external value likelihood information among a plurality of element decoders.

  The turbo code is attracting attention as a code that can obtain an error rate characteristic close to the Shannon limit, which is the limit of the channel coding method, and is also adopted in W-CDMA (Wideband Code Division Multiple Access) and has already been put into practical use. Yes. Hereinafter, turbo code encoding and decoding will be schematically described with reference to FIGS.

  FIG. 6 is a block diagram showing a general configuration of a turbo encoder. As shown in FIG. 6, the turbo encoder 600 basically includes element encoders 601 and 602 and an interleaver 603.

  In FIG. 6, the information data sequence d is output as it is as an encoded data sequence u, and is input to the element encoder 601 and the interleaver 603. Element encoder 601 convolutionally encodes information data sequence d and outputs it as parity data sequence y1. The interleaver 603 rearranges the order of the information data series d at random and gives it to the element encoder 602. Element encoder 602 convolutionally encodes the interleaved information data sequence and outputs it as parity data sequence y2. In this turbo encoder 600, 2 bits of parity data are added to 1 bit of information data, so that the coding rate R is 1/3. Note that a recursive systematic convolutional encoder is used for the element encoders 601 and 602.

  Next, FIG. 7 is a block diagram showing a general configuration of a turbo decoder. As shown in FIG. 7, the turbo decoder 700 includes an element decoder 701, a subtractor 702, an interleaver 703, an element decoder 704, a subtractor 705, deinterleavers 706 and 707, and a determiner 708. , And selector 709.

  In FIG. 7, U, Y1, and Y2 are channel values obtained by multiplying a received signal obtained by digitally demodulating a transmission signal (u, y1, y2) via a channel and a constant called a channel constant. is there. L1 and L2 hidden in FIG. 7 are generated by the turbo decoder and are referred to as prior values. L1 'and L2' are posterior probabilities that are the targets of hard decision in the determiner 708, and are referred to as posterior values. Le1 and Le2 are reliability information passed between the element decoder 701 and the element decoder 704, and are referred to as external values.

As shown in FIG. 7, the element decoders 701 and 704 receive the channel value and the prior value, and output the posterior value and the external value. This posterior value is expressed as the following formula (1).
Subsequent value = communication path value + prior value + external value (1)

  The upstream element decoder 701 performs decoding corresponding to the element encoder 601, and the downstream element decoder 704 performs decoding corresponding to the element encoder 602. An external value obtained as a result of decoding by the element decoder 701 is input to the element decoder 704 as a prior value. Similarly, an external value obtained as a result of decoding by the element decoder 704 is input to the element decoder 701 as a prior value.

  Accordingly, in FIG. 7, the preceding element decoder 701 receives the channel value (U, Y1) and the prior value (L1), and outputs the posterior value (L1 ′) and the external value (Le1). The downstream element decoder 704 receives the channel value (U, Y2) and the prior value (Le1), and outputs the posterior value (L2 ') and the external value (Le2). As described above, it is possible to improve the decoding characteristics by repeatedly performing decoding while updating the external value as reliability information between the element decoder 701 and the element decoder 704 and updating the posterior value. This is a feature of decoding.

  Now, the element decoder 701 that performs decoding corresponding to the element encoder 601 receives the received signal U of the encoded data sequence u and the received signal Y1 of the parity data sequence y1 as channel values. Further, the output L1 of the deinterleaver 706 is input via the selector 709 as a prior value. At the time of the first iterative decoding, the selector 709 connects the L1 input terminal of the element decoder 701 to the ground, gives the element decoder 701 an initial value of L1, and then connects the L1 input terminal of the element decoder 701 to the element decoder 701. The output terminal of the deinterleaver 706 is connected.

The element decoder 701 outputs the external value Le1 and the posterior value L1 ′ using the channel value and the prior value. Instead of delivering the external value Le1 itself, which is reliability information, to the element decoder 704, the result obtained by subtracting the a priori value L1 from the a posteriori value L1 ′ by the subtractor 702 is delivered via the interleaver 703. L1′−L1 delivered to the element decoder 704 is expressed by the following equation (2) from the above equation (1), and therefore, the communication channel value U and the external value Le1 are delivered.
L1′−L1 = U + Le1 (2)

In addition, the element decoder 704 that performs decoding corresponding to the element encoder 602 receives the encoded data sequence U interleaved by the interleaver 703 and the parity data sequence Y2 as channel values. Furthermore, a data series obtained by interleaving the external value Le1 in the interleaver 703 is input as the hidden previous value L2 in FIG. The element decoder 704 outputs the external value Le2 and the posterior value L2 ′ using the channel value and the prior value. Since the external value Le2 is expressed by the following formula (3) from the above formula (1), the external value Le2 is obtained by subtracting the output signal Le1 + U of the interleaver 703 from the posterior value L2 ′ by the subtractor 705. Is required.
Le2 = L2′−L2-U (3)

  The external value Le2 is deinterleaved by the deinterleaver 706, and then input to the element decoder 701 as a prior value L1, and repeatedly decoded. On the other hand, the posterior value L2 ′ output from the element decoder 704 is deinterleaved by the deinterleaver 707 and input to the determiner 708. The determiner 708 makes a hard decision on the deinterleaved posterior value L2 'and outputs a decoding result d'.

  By the way, in the element decoders 701 and 704, a log-MAP (Maximum A Posteriori probability) decoding method, a MAX-log-MAP decoding method, a SOVA (Soft Output Viterbi Algorithm) decoding method, or the like is used. Here, a turbo decoding method mainly using a MAX-log-MAP decoding algorithm will be described.

The processing procedure in turbo decoding is divided into calculation of backward probability β, calculation of forward probability α, and calculation of likelihood L (ue _k ). The backward probability β is obtained using Equation (4) in all states s. The forward probability α is obtained using Expression (5) in all states s. In addition, the likelihood L (ue _k ) is obtained using Expression (6).

In Equations (4) to (6), k is a time point, s is a state, γ _k (s _k−1 , s _k ) is a transition probability from state s _{k −1} to state s _k , and u _k is information at time k, ue _k is an estimate of the information _{u k.}

  As shown in equation (4), since the value of the time point k + 1 is used to calculate the backward probability β at the time point k, the backward probability β is calculated in the direction opposite to the time axis from the terminal to the starting edge. . On the other hand, as shown in the equation (5), since the value of the time point k−1 is used to calculate the forward probability α at the time point k, the forward probability α is increased in the forward direction with respect to the time axis from the start end to the end point. Calculated. FIG. 8 is an example of a trellis diagram including a start end and a termination end.

However, the calculations of the above formulas (4) to (5) are very complicated. For this reason, the following approximate expression (7) is used in the MAX-log-MAP decoding.

When Expression (4) and Expression (5) are modified using Expression (7), Expression (8) and Expression (9) are obtained, respectively, and when Expression (6) is further modified, Expression (10) can be obtained.

The likelihood information calculated in this way is repeatedly decoded while being passed between the two element decoders. In the determiner 708, the likelihood L (ue _k ) at the time point k calculated using the equation (8) with respect to the final likelihood information based on a predetermined iterative decoding stop criterion is L (ue _k ) ≧ When 0, the hard decision is made that the estimated value ue _k = bit 0, and when L (ue _k ) <0, the hard decision is made that the estimated value ue _k = bit 1.

Here, in such a turbo code decoding process, Patent Document 1 discloses a decoding method that further improves error correction capability in an environment where the signal-to-noise ratio of a communication channel is low. In Patent Document 1, the result of hard decision on the posterior values of a plurality of element decoders is compared, and if the hard decision results are equal, the external value is not updated, and if the hard decision results are different, the external value A method for performing the update is disclosed. According to this method, it is possible to reduce only the likelihood value including an error included in an external value passed between a plurality of element decoders, so that the error correction capability in iterative decoding is improved. As a result, the number of decoding iterations for obtaining predetermined decoding characteristics can be reduced, so that the decoding processing delay time can be shortened.
JP 2001-127647 A

  However, in the above-described conventional turbo decoding method, when the coding rate changes in the communication system in which the coding rate of the turbo code is adaptively variable, the optimum likelihood value cannot be updated according to the coding rate. However, there is a problem that the error correction capability is reduced by the likelihood update process.

  In addition, since not only the external value that is soft information but also the posterior value that is soft information must be passed to the information that is passed between multiple element decoders, the likelihood information is not updated There is also a problem that the complexity of the circuit increases as compared with the above.

  The present invention has been made in view of such a point, and can select an optimum likelihood value updating method according to the coding rate of the turbo code, and can be used in comparison with a case where the likelihood information is not updated. It is an object of the present invention to provide a turbo decoding method and a turbo decoding device that can improve error correction capability without complicating the above.

  In order to solve this problem, a turbo decoding method according to the present invention is a turbo decoding method in which iterative decoding is performed while exchanging external value likelihood information among a plurality of element decoders. Determining whether or not to update external value likelihood information exchanged between a plurality of element decoders by comparing coincidence / mismatch of logical value levels of value likelihood information and posterior value likelihood information I tried to do it.

  According to this method, since it is not necessary to newly add a signal to be passed between the element decoders, it is not necessary to complicate the system wiring for updating the likelihood information. Furthermore, a memory for temporarily buffering the posterior value to be passed is also unnecessary.

  In the turbo decoding method according to the present invention, in the above invention, when the external value likelihood information is updated, the turbo decoding method is selectively switched according to a coding rate in error correction coding.

  According to this method, external value likelihood information can be updated based on the coding rate even in a communication system in which the coding rate of the turbo code is variable.

  In the turbo decoding method according to the present invention, in the above invention, in the step of determining, the logical value levels of the prior value likelihood information, the channel value likelihood information, and the posterior value likelihood information do not match. The external value likelihood information is determined to be updated.

  According to this method, it is not necessary to complicate a circuit that performs update control of likelihood information.

  The turbo decoding device according to the present invention is a turbo decoding device that performs iterative decoding while exchanging external value likelihood information among a plurality of element decoders. Prior value likelihood information, channel value likelihood information, and Comparing the mismatch of the logical value levels of the posterior value likelihood information, and selecting the external value likelihood information to be exchanged between multiple element decoders according to the coding rate in error correction coding A likelihood information updating unit that switches and automatically updates is adopted.

  According to this configuration, the optimal likelihood update method can be selected according to the coding rate even in a communication system in which the coding rate is variable without complicating the configuration, so that turbo decoding with high error correction capability can be performed. realizable. In addition, since the error correction capability is improved, the number of decoding iterations for obtaining a predetermined decoding characteristic can be reduced, so that the decoding processing delay time can be shortened.

  According to the present invention, it is possible to select an optimal likelihood value update method according to the coding rate of the turbo code, and to perform error correction without complicating the circuit as compared with the case where the likelihood information is not updated. Ability can be improved.

  The essence of the present invention is that even in a communication system in which the coding rate of the turbo code is variable, an optimal likelihood update method can be selected according to the coding rate, and the error correction capability is high at any coding rate. It is to realize turbo decoding.

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

(One embodiment)
FIG. 1 is a block diagram showing a configuration of a turbo decoding device according to an embodiment of the present invention. The turbo decoding apparatus 100 shown in FIG. 1 includes an element decoder 101, a subtracter 102, a likelihood information update unit 103, an interleaver 104, an element decoder 105, a subtractor 106, a likelihood information update unit 107, and deinterleavers 108 and 109. And a determiner 110 and a selector 111.

  The turbo decoding device 100 has a code different from the configuration shown in FIG. 7 (conventional example), but will be described using the code shown in FIG. 7 to update the likelihood information for the element decoder 701 in the preceding stage. Unit 103 is provided between subtractor 702 and interleaver 703, and likelihood information updating unit 107 for subsequent element decoder 704 is provided between subtractor 705 and deinterleaver 706, and each is provided with a coding rate. Are given in parallel.

  As shown in FIG. 1, the likelihood information updating unit 103 subtracts the sign bit of each of the prior value L1, the posterior value L1 ′, and the channel value U of the element decoder 101 in addition to the coding rate. The output (L1′−L1) of the device 102 is input. In addition, in likelihood information updating section 107, in addition to the coding rate, output (Le1 + U) of interleaver 104, posterior value L2 ′ of element decoder 105, each sign bit of channel value U, and subtractor 106 Output (L2′-L2-U) is input.

  FIG. 2 is a block diagram illustrating a configuration example of the likelihood information updating units 103 and 107 illustrated in FIG. As shown in FIG. 2, the likelihood information update units 103 and 107 shown in FIG. 1 each include a likelihood update determination unit 201, a correction coefficient selection unit 202, a multiplier 203, and a selector 204.

  In FIG. 2, the likelihood update determination unit 201 determines whether or not to update likelihood information by comparing the coincidence / mismatch of the logical value levels of the input posterior value code, channel value code, and prior value code. . FIG. 3 is a diagram illustrating an example of a relationship table between the coding rate and the correction coefficient included in the correction coefficient selection unit 202 illustrated in FIG. The correction coefficient selecting unit 202 obtains a correction value for correcting the likelihood information by applying the input coding rate to the table shown in FIG. The subtractor 203 multiplies the correction value obtained by the correction coefficient selection unit 202 by the likelihood information (L1′-L1) or (L2′-L2-U), and supplies the result to one input of the selector 204. Likelihood information (L1'-L1) or (L2'-L2-U) is given to the other input of the selector 204. The selector 204 selects the output of the multiplier 203 as likelihood update information when the output level of the likelihood update determination unit 201 is “1” level. Further, when the output level of the likelihood update determination unit 201 is “0” level, the selector 204 selects the likelihood information (L1′-L1) or (L2′-L2-U) as the likelihood update information as it is. To do.

  Next, the operation will be described. The channel value (information sequence U and redundant sequence Y1), which is a signal to be decoded, is decoded by the element decoder 101 together with the a priori value L1 to obtain a posterior value L1 '. Since the initial value L1 in the iterative decoding is set to zero, it is connected to the ground side via the selector 111 during the first iterative decoding. The subtracter 102 subtracts the prior value L1 from the obtained posterior value L1 '. The likelihood information (L1′−L1) is equal to the external value Le1 + communication channel value U from the above-described equation (1).

  The likelihood information (L1′−L1) output from the subtractor 102 is input to the likelihood information update unit 103. The likelihood information update unit 103 compares the posterior value L1 'with the channel value U and the posterior value L1' with the sign bit of the prior value L1, and updates the likelihood information if even one sign is different. That is, the likelihood update determination unit 201 sets the output level to “1” level. On the other hand, the likelihood information update unit 103 compares the sign bits of the posterior value L1 ', the channel value U, and the prior value L1, and does not update the likelihood if all the sign bits are equal. That is, the likelihood update determination unit 201 sets the output level to “0” level.

  The likelihood information updating unit 103 updates the likelihood information by multiplying the input likelihood information (L1'-L1) by a likelihood correction coefficient having a value of 1 or less. As shown in FIG. 3, the likelihood correction coefficient has a coefficient value that differs depending on the coding rate, and is a coefficient of value 1 or less.

  The signals subjected to the likelihood update process by the likelihood information update unit 103 are interleaved by the interleaver 104, and the arrangement of the signals matches the channel value Y2 of the redundant sequence. The interleaved likelihood information L1'-L1 (= Le1 + U) and the channel value Y2 of the redundant sequence are decoded by the element decoder 105 to obtain a posterior value L2 '. The subtracter 106 subtracts the likelihood information L1'-L1 (= Le1 + U) from the element decoder 101 from the obtained posterior value L2 '. This external value Le1 is equal to the prior value L2 of the element decoder 105. Therefore, the output L2'-L2-U of the subtractor 106 is equal to the external value Le2 of the element decoder 105 according to the equation (1).

  The external value Le2 that is the output of the subtractor 106 is input to the likelihood information update unit 107. The likelihood information updating unit 107 compares the sign bits of the posterior value L2 ′, the channel value U, the posterior value L2 ′, and the prior value Le1, and updates the likelihood information if even one sign is different. . That is, the likelihood update determination unit 201 sets the output level to “1” level. On the other hand, the likelihood information updating unit 107 compares the sign bits of the posterior value L2 ', the channel value U, and the prior value Le1, and does not update the likelihood when all the sign bits are equal. That is, the likelihood update determination unit 201 sets the output level to “0” level.

  The likelihood information updating unit 107 updates the likelihood information by multiplying the input likelihood information (L2'-L2-U) by a likelihood correction coefficient having a value of 1 or less. The likelihood correction coefficient is a coefficient value different depending on the coding rate, and is a coefficient having a value of 1 or less. Here, the same correction coefficient table as the likelihood information updating unit 103 may be used, or a different correction coefficient table may be used.

  The signal subjected to the likelihood update process by the likelihood information update unit 107 is deinterleaved by the deinterleaver 108, and becomes the prior value L1 of the element decoder 101. In this way, one-time iterative decoding is performed.

  The a posteriori value output of the element decoder 105 obtained by repeating this iterative decoding a predetermined number of times is deinterleaved by the deinterleaver 109, and the decision unit 110 makes a hard decision to obtain a decoding result d '. Note that other iterative decoding stop norms may be used.

  In addition, as a result of comparing the posterior value and the prior value, and the sign bit of the posterior value and the channel value, if the mismatch is recognized, the likelihood information is updated. Alternatively, a method using only the comparison result may be used.

  In the case of the likelihood information updating unit 107, a likelihood update determination method is performed using a comparison result of the sign value of (Le1 + U), which is an added value of the external value L2 ′ and the prior value and the channel value. May be.

  FIG. 4 is a diagram illustrating a relational characteristic between the signal power to noise power ratio [Eb / N0] per bit and the bit error rate [BER] of the turbo decoding device illustrated in FIG. Similarly, FIG. 5 is a diagram showing a relational characteristic between the signal power to noise power ratio [Eb / N0] per bit and the block error rate [BLER] of the turbo decoding device shown in FIG. In any case, when the block size is 1000 and the number of iterations is 8, the conventional method (conv1) without likelihood information update, the conventional method (conv2) disclosed in Patent Document 1, and the method according to the present embodiment (prop) The result of having been verified at the coding rate 1/3 and the coding rate 3/4 is shown.

  As shown in FIGS. 4 and 5, at the coding rate of 1/3, the conventional method (conv2) disclosed in Patent Document 1 has better characteristics than the conventional method (conv1) without updating the likelihood information. As shown in the figure, on the contrary, the characteristics deteriorate at the coding rate of 3/4. On the other hand, the system (prop) according to the present embodiment shows good characteristics at any coding rate.

  As described above, according to the embodiment of the present invention, the likelihood information can be updated from the input / output signal and the coding rate of one element decoder. The likelihood update process can be realized without complication, and the optimal likelihood update method can be selected according to the coding rate even in a communication system in which the coding rate is variable, and at any coding rate Turbo decoding with high error correction capability can be realized.

  The present invention is useful for improving error correction capability in a communication system in which the coding rate of a turbo code is variable.

The block diagram which shows the structure of the turbo decoding apparatus which concerns on one embodiment of this invention The block diagram which shows the structural example of the likelihood information update part shown in FIG. The figure which shows an example of the relationship table of the encoding rate with which the correction coefficient selection part shown in FIG. 2 is provided, and a correction coefficient. The figure which shows the bit error rate characteristic obtained with the turbo decoding apparatus shown in FIG. The figure which shows the block error rate characteristic obtained with the turbo decoding apparatus shown in FIG. Block diagram showing a typical configuration example of a turbo encoder Block diagram showing general configuration of turbo decoder A diagram showing an example of a trellis diagram including the start and end points

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Turbo decoding apparatus 101 Element decoder 102 Subtractor 103 Likelihood information update part 104 Interleaver 105 Element decoder 106 Subtractor 107 Likelihood information update part 108 Deinterleaver 109 Deinterleaver 110 Determinator 111 Selector 201 Likelihood update determination part 202 Correction coefficient selection unit 203 Multiplier 204 Selector

Claims (4)

  In a turbo decoding method in which iterative decoding is performed while exchanging external value likelihood information among a plurality of element decoders, the logical value level of each of prior value likelihood information, channel value likelihood information, and posterior value likelihood information A turbo decoding method comprising: determining whether or not to update external value likelihood information exchanged between a plurality of element decoders by comparing coincidence / non-coincidence.   The turbo decoding method according to claim 1, wherein when updating the external value likelihood information, the external value likelihood information is selectively switched according to a coding rate in error correction coding.   In the determining step, when the logical value levels of the prior value likelihood information, the channel value likelihood information, and the posterior value likelihood information do not match, it is determined to update the external value likelihood information. The turbo decoding method according to claim 1.   In a turbo decoding device that performs iterative decoding while exchanging external value likelihood information among a plurality of element decoders, logical value levels of prior value likelihood information, channel value likelihood information, and posterior value likelihood information are respectively Likelihood information updating means for comparing coincidence / mismatch and selectively switching and updating external value likelihood information exchanged between a plurality of element decoders according to the coding rate in error correction coding A turbo decoding device comprising:

Images