JP2005286624A - Decoder and decoding method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To prevent deterioration in an error rate characteristic by obtaining a likelihood highly accurately. <P>SOLUTION: A noise variance operator 201 obtains the variance of quantization noises in A/D conversion, obtains the variance of Gaussian noises, and totals the variance of quantization noises in A/D conversion and the variance of Gaussian noises which are obtained to obtain a total variance value. A soft decision decoder 203 and a soft decision decoder 205 perform soft decision decoding by using soft decision decoding data, systematic bit data and parity bit data, and perform soft decision decoding while considering the total variation value of the variance value of the quantization noises and the variance of Gaussian noises as the variance of an input signal and the variance of external information likelihood in performing soft decision decoding. A hard decision section 208 performs hard decision for a result of repeating soft decision decoding. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a turbo decoding apparatus and a turbo decoding method that are widely applied particularly to the fields of mobile communication and satellite communication.

  In wireless communication, an error correction code for correcting an error in a received signal is a very important technique particularly when communication is performed on a transmission path in which an error is likely to occur. As one of the error correction codes, a turbo code has been attracting attention in recent years (for example, Patent Document 1).

  One of the features when decoding turbo-encoded data is that the decoding is repeatedly performed by two decoders. Specifically, the likelihood (a value indicating the probability of each decoding result) obtained from the decoding result of the first decoder is input to the second decoder, and the likelihood is estimated by the second decoder. Decoding is performed using the data and the received data (data including redundant bits for error detection). Further, the likelihood obtained from the decoding result of the second decoder is input again to the first decoder, and decoding is repeated. In this way, decoding accuracy is gradually increased by performing iterative decoding using the likelihood calculated from one decoder.

  A decoder corresponding to a turbo encoder is configured by combining two element decoders (element decoder 1 and element decoder 2). The decoder receives noise (additional white Gaussian noise), and receives received signal sequences corresponding to information bits and parity bits, respectively. In the element decoder 1, the reliability information transmitted from the element decoder 2 is used. Decoding processing is performed using a certain prior value, and an external value is output. The external value represents an increase in the reliability of the symbol by the element decoder. Data output from the element decoder 1 as an external value is rearranged by an interleaver and input to the element decoder 2 as a prior value. Similarly, the element decoder 2 receives a reception signal corresponding to information bits rearranged by the interleaver and a prior value, performs a decoding process, and outputs an external value. Then, the data output from the element decoder 2 as the external value is subjected to an operation for returning the rearrangement by the interleaver by the deinterleaver, and is input to the element decoder 1 as the prior value, and is repeatedly decoded. After repeating several to a dozen times, the element decoder 2 calculates a posterior value defined as a log posterior probability ratio, and performs a hard decision to output a final decision result.

FIG. 6 is a diagram illustrating deterioration of BER (Bit Error Rate) due to an error (offset) in SNR estimation of an input signal. FIG. 6 shows that the larger the SNR estimation error, the more the BER characteristic is degraded.
JP 2002-217748 A

  However, in the conventional decoding device and decoding method, since the error rate characteristic of turbo decoding depends on the input signal and the likelihood, the error rate characteristic does not deteriorate if the likelihood is obtained accurately, but the likelihood is not improved. There is a problem that the error rate characteristic deteriorates because it cannot be obtained with high accuracy.

  The present invention has been made in view of the above points, and an object of the present invention is to provide a decoding apparatus and a decoding method that can prevent deterioration of error rate characteristics by obtaining likelihood with high accuracy.

  The decoding apparatus according to the present invention includes a noise variance calculation unit that obtains a variance of Gaussian noise and a variance of quantization noise in received data including systematic bit data and parity bit data, and the Gaussian obtained by the noise variance calculation unit. Soft decision decoding means for repeatedly performing soft decision decoding on the received data based on a variance of noise and a variance of the quantization noise; and a soft decision after the soft decision decoding is repeated a predetermined number of times by the soft decision decoding means The structure which comprises the hard decision means which performs a hard decision with respect to a decoding result is taken.

  According to this configuration, by performing soft decision decoding in consideration of quantization noise, it is possible to obtain external likelihood data with high accuracy, and thus it is possible to prevent deterioration of error rate characteristics.

  In the decoding device according to the present invention, in the configuration, the soft decision decoding means calculates a total variance value of the variance of the Gaussian noise and the variance of the quantization noise as a variance in a signal before soft decision decoding for each soft decision decoding. And the soft decision decoding is repeatedly performed by estimating the variance of the external information likelihood indicating the reliability in the soft decoding output for each soft decision decoding.

  According to this configuration, in addition to the above effects, external likelihood data can be obtained with high accuracy by performing soft decision decoding in consideration of the total variance of the variance of Gaussian noise and the variance of quantization noise. .

  The receiving apparatus of the present invention employs a configuration including the decoding apparatus.

  According to this configuration, by performing soft decision decoding in consideration of quantization noise, it is possible to obtain external likelihood data with high accuracy, and thus it is possible to prevent deterioration of error rate characteristics.

  The decoding method of the present invention includes a step of obtaining variance of Gaussian noise and variance of quantization noise in received data comprising systematic bit data and parity bit data, and obtaining the obtained variance of Gaussian noise and variance of quantization noise. Repetitively soft-decision decoding the received data consisting of the systematic bit data and parity bit data based on the following, and performing a hard decision on the soft-decision decoding result after repeated soft-decision decoding a predetermined number of times; It was made to comprise.

  According to this method, since the external likelihood data can be obtained with high accuracy by performing soft decision decoding in consideration of quantization noise, it is possible to prevent deterioration of error rate characteristics.

  According to the decoding method of the present invention, in the method, the total variance value of the variance of the Gaussian noise and the variance of the quantization noise is estimated as a variance in a signal before soft decision decoding for each soft decision decoding, and soft decision decoding is performed. The soft decision decoding is repeatedly performed by estimating the variance of the external information likelihood indicating the reliability in each soft decoding output.

  According to this method, in addition to the above effect, the external likelihood data can be obtained with high accuracy by performing soft decision decoding in consideration of the total variance value of the variance of the Gaussian noise and the variance of the quantization noise. .

  According to the present invention, it is possible to prevent deterioration of error rate characteristics by obtaining the likelihood with high accuracy.

  The gist of the present invention is to repeatedly perform soft decision decoding on received data composed of systematic bit data and parity bit data based on variance of Gaussian noise and variance of quantization noise.

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

(Embodiment)
FIG. 1 is a diagram showing a communication system 100 according to an embodiment of the present invention. The communication system 100 includes a transmission device 120 and a reception device 121 that receives transmission data transmitted from the transmission device 120 via the channel 108.

  First, the configuration of the transmission device 120 will be described. Encoding section 101 performs turbo encoding on transmission data and outputs the result to rate matching section 102.

  The rate matching unit 102 increases or decreases the number of bits so that the size of the transmission data input from the encoding unit 101 fits in one frame, and outputs the result to the modulation unit 103.

  Modulation section 103 modulates the transmission data input from rate matching section 102 and outputs the result to inverse fast Fourier transform (hereinafter referred to as “IFFT”) section 104.

  IFFT section 104 performs orthogonal frequency division multiplexing on the transmission signal input from modulation section 103 and outputs the resultant signal to guard interval (hereinafter referred to as “GI”) insertion section 105.

  The GI insertion unit 105 inserts a GI into the transmission signal input from the IFFT unit 104 and outputs it to the digital / analog (hereinafter referred to as “D / A”) conversion unit 106.

  The D / A conversion unit 106 performs D / A conversion on the transmission signal input from the GI insertion unit and outputs the result to the RF unit 107.

  The RF unit 107 up-converts the transmission signal input from the D / A conversion unit 106 from the baseband frequency to the radio frequency and transmits the signal to the reception device 121 via the channel 108 that is a radio channel.

  Next, the configuration of receiving apparatus 121 will be described. The RF unit 109 down-converts the received signal received from the radio frequency to the baseband frequency, and outputs it to the analog / digital (hereinafter referred to as “A / D”) conversion unit 110.

  A / D conversion section 110 performs A / D conversion on the received signal input from RF section 109 and outputs the result to GI removal section 111. In addition, A / D conversion section 110 outputs dynamic range information, which is information on dynamic range at the time of A / D conversion, and bit number information, which is information on the number of bits in dynamic range, to decoding section 116.

  The GI removal unit 111 removes the GI from the received signal input from the A / D conversion unit 110 and outputs the GI to the fast Fourier transform (hereinafter referred to as “FFT”) unit 112.

  The FFT unit 112 performs FFT on the reception signal input from the GI removal unit 111 and outputs the result to the demodulation unit 113.

  Demodulation section 113 demodulates the received signal input from FFT section 112 and outputs the demodulated signal to rate dematching section 114 and SNR measurement section 115.

  Rate dematching section 114 restores the increased or decreased number of bits in the received data input from demodulation section 113 and outputs the result to decoding section 116.

  The SNR measurement unit 115 measures the SNR from the demodulation result input from the demodulation unit 113 and outputs SNR information that is information of the measured SNR to the decoding unit 116.

  Decoding section 116 repeatedly receives soft decision decoding of the received data input from rate dematching section 114 and then performs turbo decoding that performs hard decision to obtain received data. At this time, the decoding unit 116 obtains the variance of the quantization noise from the dynamic range information and the bit information input from the A / D conversion unit 110, and obtains a soft decision decoding result using the obtained variance of the quantization noise.

  Next, details of the configuration of the decoding unit 116 will be described with reference to FIG. FIG. 2 is a block diagram illustrating a configuration of the decoding unit 116.

  The noise variance calculation unit 201 is based on the dynamic range information and bit information input from the A / D conversion unit 110 and the SNR information input from the SNR measurement unit 115, and includes Gaussian noise including thermal noise and noise caused by fading. The variance is calculated in consideration of both the variance and the quantization noise variance, and the variance information, which is the obtained variance information, is output to the soft decision decoding section 203 and the soft decision decoding section 205.

  Interleaver 202 rearranges systematic bit data Xa input from rate dematching section 114 in a predetermined pattern, and outputs the result to soft decision decoding section 205.

  The soft decision decoding unit 203 performs a soft decision based on the soft decision decoding result input from the deinterleaver 206, the systematic bit data Xa input from the rate dematching unit 114, the parity bit data Xb, and the variance information input from the noise variance calculation unit 201. Decode and output to interleaver 204.

  Interleaver 204 rearranges the soft decision decoding data input from soft decision decoding section 203 in the same pattern as interleaver 202 and outputs the result to soft decision decoding section 205.

Soft decision decoding section 205 receives soft decision decoded data Z _k input from interleaver 204, systematic bit data Xa input from interleaver 202, parity bit data Xc input from rate dematching section 114, and noise variance calculation section 201. Soft-decision decoding is performed from the input distributed information, and output to the deinterleaver 206 and deinterleaver 207. Details of the soft decision decoding unit 205 will be described later.

The deinterleaver 206 returns the soft-decision decoded data Z _k input from the soft-decision decoding unit 205 to the array before being rearranged by the interleaver 202 and the interleaver 204, and outputs it to the soft-decision decoding unit 203.

The deinterleaver 207 returns the soft-decision decoded data Z _k input from the soft-decision decoding unit 205 to the array before being rearranged by the interleaver 202 and the interleaver 204, and outputs it to the hard-decision unit 208.

Hard decision section 208 makes a hard decision on soft decision decoded data Z _k input from deinterleaver 207 to obtain received data.

  Next, details of the configuration of the soft decision decoding unit 205 will be described with reference to FIG. FIG. 3 is a block diagram illustrating a configuration of the soft decision decoding unit 205.

The transition probability calculation unit 301 obtains a transition probability γ from the soft decision decoded data Z _k input from the interleaver 204, the systematic bit data Xa input from the rate dematching unit 114, and the unordered parity bit data Xb. Then, the transition probability calculation unit 301 outputs the obtained transition probability γ information to the forward probability calculation unit 302 and the backward probability calculation unit 303.

  The forward probability calculation unit 302 obtains the forward probability α from the information on the transition probability γ input from the transition probability calculation unit 301. Then, the forward probability calculation unit 302 outputs information on the obtained forward probability α to the joint probability calculation unit 305.

  The backward probability calculation unit 303 obtains the backward probability β from the information on the transition probability γ input from the transition probability calculation unit 301. Then, the backward probability calculation unit 303 outputs the obtained backward probability β information to the storage unit 304.

  The storage unit 304 stores information on the backward probability β input from the backward probability calculation unit 303, and outputs the stored backward probability β to the joint probability calculation unit 305 at a predetermined timing.

  The connection probability calculation unit 305 obtains the connection probability λ from the information of the forward probability α input from the forward probability calculation unit 302 and the information of the backward probability β input from the storage unit 304. Further, the joint probability calculation unit 305 obtains the external information likelihood W from the obtained joint probability λ. Then, the joint probability calculation unit 305 outputs information on the obtained external information likelihood W to the posterior probability calculation unit 306 and the deinterleaver 206. Here, the external information likelihood indicates the reliability of the soft output decoding result.

The posterior probability calculation unit 306 is based on the information on the external information likelihood W input from the joint probability calculation unit 305, the soft decision decoded data Z _k input from the interleaver 204, and the variance information input from the noise variance calculation unit 201. Λ (d _k ) is obtained, and information on the obtained posterior probability Λ (d _k ) is output to the deinterleaver 207. The configuration of the soft decision decoding unit 203 is the same as that in FIG. 3 except that the posterior probability Λ (d _k ) obtained by the posterior probability calculation unit 306 is discarded without being output, and the description thereof is omitted. . A method for obtaining the posterior probability Λ (d _k ) will be described later.

  Next, details of the configuration of the noise variance calculation unit 201 will be described with reference to FIG. FIG. 4 is a block diagram illustrating a configuration of the noise division calculation unit 201.

  The Gaussian noise variance calculation unit 401 obtains Gaussian noise variance from the SNR measurement result input from the SNR measurement unit 115 and outputs the obtained Gaussian noise variance value to the adder 403.

  The quantization noise variance calculation unit 402 obtains the quantization noise variance from the dynamic range information and the bit number information input from the A / D conversion unit 110 and outputs the obtained quantization noise variance value to the adder 403.

  Adder 403 is a soft-decision decoding unit using, as distribution information, an addition value obtained by adding the variance value of Gaussian noise input from Gaussian noise variance calculation unit 401 and the variance value of quantization noise input from quantization noise variance calculation unit 402 203 and the soft decision decoding unit 205.

Next, a method for obtaining the posterior probability Λ (d _k ) in the decoding unit 116 will be described.

  First, the transition probability calculation unit 301 uses the soft decision decoding result input from the interleaver 203, the systematic bit data Xa input from the rate dematching unit 114, and the unordered parity bit data Xb to To determine the transition probability γ.

  Next, the forward probability calculation unit 302 obtains the forward probability α using the equation (2) from the transition probability γ.

  Next, the backward probability calculation unit 303 obtains the backward probability β using the equation (3) from the transition probability γ.

  Next, the joint probability calculation unit 305 obtains the joint probability λ using the formula (4) from the forward probability α and the backward probability β.

Further, the joint probability calculation unit 305 obtains the external information likelihood W _k from the obtained joint probability λ using equation (5).

Then, the joint probability calculation unit 305 outputs the obtained external information likelihood W _k to the deinterleaver 206 and also outputs it to the posterior probability calculation unit 306.

On the other hand, the noise variance calculation unit 201 obtains the Gaussian noise variance σ _g ² of both the entire received data and the soft decision decoded data Z _k by the Gaussian noise variance calculation unit 401 from the SNR measurement result and Equation (6). .

ここで、Ａ：振幅
Ｒ：符号化率
Ｅ_ｂ／Ｎ_０：１ｂｐｓの信号を伝送するために必要な電力と１Ｈｚ当たりの熱雑音電力との比。

Here, A: Amplitude R: Coding rate E _b / N ₀ : Ratio of power required to transmit a signal of ₁ bps and thermal noise power per 1 Hz.

In addition, the noise variance calculation unit 201 obtains the quantization noise variance σ _q ² in the quantization noise variance calculation unit 402 using the equations (7) and (8) from the dynamic range information and the bit information.

ここで、Ｅ_ｍａｘ：ダイナミックレンジ
ｂ：ビット数。

Here, E _max : Dynamic range b: Number of bits.

Next, the noise variance calculation unit 201 uses the equation (9) from the Gaussian noise variance σ _g ² and the quantization noise variance σ _q ² to add the Gaussian noise variance σ _g ² and the quantization noise variance using the adder 403. A total dispersion value σ ² with σ _q ² is obtained.

Then, the noise variance calculation unit 201 outputs the information of the obtained total variance value σ ² to the posterior probability calculation unit 306 of the soft decision decoding unit 203 and the soft decision decoding unit 205.

Next, the posterior probability calculation unit 306 obtains the posterior probability Λ (d _k ) using the equation (10) from the total variance σ ² , the soft decision decoding result, and the equation (5). At this time, the posterior probability calculation unit 306 uses the total variance value input from the noise variance calculation unit 201 for each soft decision decoding of the soft decision decoding unit 203 or the soft decision decoding unit 205, and the variance and external signal in the signal before the soft decision decoding The variance of the information likelihood is estimated, and the total variance σ ² is substituted for each of σ ² and σ _z ^{2 in} the equation (10).

Then, the posterior probability calculation unit 306 outputs information on the posterior probability Λ (d _k ) obtained from the equation (10) to the deinterleaver 207.

The posterior probability Λ (d _k ) obtained in this way is hard judged by the hard decision unit 208. That is, the hard decision unit 208 determines “1” if the posterior probability posterior probability Λ (d _k ) is “0” or more, and determines that “posterior probability posterior probability Λ (d _k ) is less than“ 0 ”. 0 ”.

  FIG. 5 shows a conventional BER characteristic # 501 in which quantization noise is not considered in soft decision decoding, and a BER characteristic # 502 in the present embodiment in which quantization noise is considered in soft decision decoding. It is a figure which shows what compared. As shown in FIG. 5, when quantization noise is considered in soft decision decoding, the bit error rate characteristic can be improved.

  As described above, according to the present embodiment, the likelihood can be obtained with high accuracy by performing the soft decision decoding in the turbo decoding in consideration of the quantization noise, so that the deterioration of the error rate characteristic can be prevented. it can.

  In the above embodiment, the OFDM signal is transmitted / received. However, the present invention is not limited to this, and the present invention is applicable to transmission / reception of any signal other than OFDM. In the above embodiment, the variance of the Gaussian noise is obtained from the SNR. However, the present invention is not limited to this. The variance of the Gaussian noise is obtained using a predetermined coefficient instead of the SNR. May be. In the above embodiment, the soft decision decoding is performed using the total variance value obtained by adding the variance of the Gaussian noise and the variance of the quantization noise. However, the present invention is not limited to this, and the variance of the Gaussian noise and the quantization noise are used. Alternatively, soft decision decoding may be performed by obtaining a variance value in consideration of both Gaussian noise and quantization noise obtained by a method other than adding the variances of the two.

  The decoding device and the decoding method according to the present invention have an effect of preventing deterioration of error rate characteristics by obtaining likelihood with high accuracy, and are useful for decoding received data.

The figure which shows the communication system which concerns on embodiment of this invention The block diagram which shows the structure of the decoding part which concerns on embodiment of this invention The block diagram which shows the structure of the soft decision decoding part which concerns on embodiment of this invention The block diagram which shows the structure of the noise dispersion calculating part which concerns on embodiment of this invention Diagram showing BER characteristics Diagram showing BER characteristics

Explanation of symbols

109 RF unit 110 A / D conversion unit 111 GI removal unit 112 FFT unit 113 demodulation unit 114 rate dematching unit 115 SNR measurement unit 116 decoding unit 121 receiver 201 noise variance calculation unit 203, 205 soft decision decoding unit 208 hard decision unit

Claims (5)

A noise variance calculation means for obtaining a variance of Gaussian noise and a variance of quantization noise in received data comprising systematic bit data and parity bit data;
Soft decision decoding means for repeatedly performing soft decision decoding on the received data based on the variance of the Gaussian noise and the variance of the quantization noise obtained by the noise variance calculation means;
Hard decision means for performing a hard decision on a soft decision decoding result after being subjected to soft decision decoding repeatedly a predetermined number of times by the soft decision decoding means;
A decoding apparatus comprising:   The soft decision decoding means estimates a total variance value of the variance of the Gaussian noise and the variance of the quantization noise as a variance in a signal before soft decision decoding for each soft decision decoding, and soft decoding for each soft decision decoding The decoding apparatus according to claim 1, wherein soft decision decoding is repeatedly performed by estimating a variance of external information likelihood indicating reliability in output.   A receiving device comprising the decoding device according to claim 1. Determining Gaussian noise variance and quantization noise variance in received data comprising systematic bit data and parity bit data;
Repeatedly performing soft decision decoding on the received data consisting of the systematic bit data and parity bit data based on the obtained variance of the Gaussian noise and the variance of the quantization noise;
Performing a hard decision on the soft decision decoding result after the soft decision decoding is repeated a predetermined number of times;
The decoding method characterized by comprising.   An external value indicating the reliability of the soft decoding output for each soft decision decoding and estimating the total variance value of the variance of the Gaussian noise and the variance of the quantization noise as the variance in the signal before soft decision decoding for each soft decision decoding 5. The decoding method according to claim 4, wherein soft decision decoding is repeatedly performed by estimating the variance of information likelihood.

Images