KR20070079923A - Low complexity and power-consumption turbo-decoder using variable scaling factor - Google Patents

Abstract

A turbo decoder with low complexity and low power consumption using a variable scaling factor is provided to reduce the power consumption of the turbo decoder by reducing the average number of repeated decoding. A turbo decoder with low complexity and low power consumption using a variable scaling factor includes a first component decoder(210), a second component decoder(230), an early stoppage judging unit(250), and a variable scaling factor computing unit(260). The first and second component decoders(210,230) repeatedly perform decoding regarding the bits decoded by a turbo code according to the variable scaling factor. The early stoppage judging unit(250) checks the value of SDR(Sign Difference Ratio), and variably controls the number of repeated decoding according to the value of the SDR. The variable scaling factor computing unit(260) computes a variable scaling factor in each area of decoding convergence according to the value of SDR.

Description

Low complexity and power-consumption Turbo-decoder using variable scaling factor

1A and 1B are configuration diagrams of a conventional turbo decoder.

2 is a block diagram of a low complexity and low power turbo decoder using a variable gain coefficient according to the present invention.

3 is a diagram illustrating an EXIT chart of an EMLMAP algorithm according to a gain factor of 0.6 to 0.9.

4 is a diagram schematically illustrating decoding performance of an EMLMAP algorithm according to the gain coefficient of FIG. 2.

5 is a graph comparing the BER performance of the turbo decoder of the present invention and the conventional turbo decoder when N = 5114 and R = 1/3.

6 is a graph comparing and analyzing the average number of times of repeated decoding of the turbo decoder of the present invention and the conventional turbo decoder when N = 5114 and R = 1/3.

Explanation of symbols on the main parts of the drawings

210: first configuration decoder

220: interleaver

230: second configuration decoder

240: deinterleaver

250: early stop determination unit

260: variable gain coefficient calculation unit

The present invention relates to a low complexity and low power turbo decoder using a variable gain factor, and more particularly, to apply a variable scaling factor having an optimal performance to each decoded convergence region using a SDR (Sign Difference Ratio). By limiting the number of iterative decoding, the present invention relates to a turbo decoder that can improve performance deterioration of a low complexity turbo decoder and reduce power consumption by reducing the average number of iterative decoding.

As mobile communication services provide full-scale multimedia services such as video and wireless Internet, low bit error rate (BER) performance as well as high-speed transmission are required. To this end, researches on error correction techniques and performance improvements have been actively conducted, and turbo codes are currently used as error correction codes for next generation mobile communication systems such as HSDPA (High Speed Downlink Packet Access) and WiBro. Was adopted.

1A and 1B illustrate a conventional turbo decoder for decoding such a turbo code. Referring to FIG. 1A, a conventional turbo decoder connects two component decoders 110 and 130 in series, and each component. Iterative decoding is performed while exchanging additional information generated by the decoders 110 and 130 with each other. At this time, the decoding algorithms used by the component decoders 110 and 130 include a LMAP algorithm and a MLMAP algorithm.

The LMAP algorithm implements the Maximum A Posteriori (MAP) algorithm, which is an optimal algorithm for decoding information words on trellis, in a log domain. Simple implementation.

In particular, the MLMAP algorithm is to reduce the complexity and decoding delay of the LMAP algorithm which performs probabilistic iterative decoding, and is easier to implement than the LMAP algorithm. However, when the receiver can know the correct signal-to-noise ratio, there is a disadvantage in that there is a performance degradation compared to decoding by the LMAP algorithm.

)를 곱하여 반복 복호를 수행하는 EMLMAP(Enhanced Max-Log-MAP) 알고리즘이 제안되었다.In order to improve performance degradation of such an MLMAP algorithm, as shown in FIG. 1B, a fixed scaling factor is applied to an output of each of the component decoders 110 and 130.

An EMLMAP algorithm has been proposed, which performs iterative decoding by multiplying ().

However, in the case of the EMLMAP algorithm, the decoding performance of the coding block is relatively close to that of the LMAP algorithm when the coding block is relatively small. However, when the coding block is large, the performance deterioration occurs compared to the LMAP algorithm. .

In addition, in the turbo decoder using the EMLMAP algorithm, as the number of iterative decoding increases, the bit error rate and frame error rate decrease, but the degree of improvement gradually decreases. Therefore, iterative decoding after reaching the performance limitations of the turbo code does not have much significance, only the additional operation and the decoding delay.

Therefore, since it is important to determine the power consumption of the turbo decoder and the time delay caused by the decoding process, at which point it is necessary to stop repetitive decoding while maintaining the decoding performance. There is a need for a means which can limit the number of times of repeated decoding to a certain threshold.

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems, and an object of the present invention is to apply a variable scaling factor having an optimal performance for each decoding convergence region by using a signal difference ratio (SDR) and to iterate. By limiting the number of decoding, the turbo decoder can reduce the power consumption by improving the performance deterioration of the low complexity turbo decoder and reducing the average number of iterations.

In order to achieve the above object, a low complexity and low power turbo decoder according to the present invention includes a first and second component decoders for performing repetitive decoding on bits encoded by a turbo code according to a variable gain coefficient, and a signal difference (SDR). And a variable gain coefficient calculation unit that calculates a variable gain coefficient in each decoding convergence region according to the SDR value by checking a ratio value and variably limits the number of iteration decoding according to the SDR value. It features.

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

2 is a block diagram of a low complexity and low power turbo decoder using a variable gain coefficient according to the present invention.

)를 곱하기 위한 제1, 2 곱셈기(M1, M2)와, SDR값을 체크하여 그 SDR값에 따라 반복 복호 횟수를 가변적으로 제한하는 조기 정지 판단부(250)와, 상기 SDR값에 따라 각 복호수렴 영역에서 최적의 성능을 갖는 가변 이득계수(

)를 계산하는 가변 이득계수 계산부(260)를 포함한다. As shown in FIG. 2, a turbo decoder according to the present invention includes first and second component decoders 210 and 230 that perform iterative decoding on bits encoded by a turbo code according to a variable gain factor to be described later. And an interleaver 220 for matching the order of the bit strings between the first and second component decoders 210 and 230, and the information rearranged by the interleaver 220 in the same order as the original information bits. A deinterleaver 240 rearranging the data and an input value and an output value of the first and second component decoders 210 and 230 to generate new additional information to be used as an input value of the next component decoder. Variable gain coefficients for new additional information to be used as input values for the first and second adders A1 and A2 and the component decoder of the next stage.

The first and second multipliers (M1, M2) for multiplying (), an early stop determination unit (250) for checking the SDR value and variably limiting the number of repeated decodings according to the SDR value, and for each decoding according to the SDR value. Variable Gain Coefficient with Optimal Performance in Convergence Domain

It includes a variable gain coefficient calculation unit 260 for calculating ().

,

를 생성하여 출력하며, 이 때, 상기 각 구성 복호기(210, 230)의 입력에는 곱셈기(M1, M2)를 통해 가변 이득계수(

)가 곱해져 반복 복호가 수행된다.The first and second component decoders 210 and 230 perform an iterative decoding while exchanging additional information with each other through an EMLMAP algorithm, thereby performing a log-likelihood ratio (LLR).

,

Generates and outputs a variable gain coefficient through the multipliers M1 and M2 to the inputs of the component decoders 210 and 230.

) Is multiplied to perform repeated decoding.

)과 제 2 구성 복호기(230)의 출력(

)의 부호변화의 개수, 즉, SDR값을 측정하여, 상기 SDR값이 '0'이면 반복 복호가 바로 중지되도록 제어하며, 반대로 상기 SDR값이 '0'이 아니면 반복 복호가 계속 수행되도록 제어한다.The early stop determination unit 250 outputs the first component decoder 210 (

) And the output of the second configuration decoder 230 (

The number of sign changes, i.e., the SDR value, is measured to control the repeated decoding to stop immediately if the SDR value is '0', and to control the repeated decoding to continue if the SDR value is not '0'. .

)를 적용하고 반복 복호 횟수를 제한함으로써, 저복잡도 터보 복호기의 성능 열화를 개선시킴과 동시에 평균 반복 복호 횟수를 줄여 전력소모량을 줄일 수 있는 것에 가장 큰 특징이 있는 바, 이하의 설명에서 본 발명에 따른 가변 이득계수(

) 적용 방법에 대하여 보다 상세히 설명하면 다음과 같다.According to the present invention, a variable gain coefficient having an optimal performance for each decoding convergence region using SDR (

By applying the number of times and limiting the number of repeated decoding, the most characteristic is that it can improve the performance degradation of the low-complexity turbo decoder and at the same time reduce the power consumption by reducing the average number of repeated decoding, as described below in the present invention Variable gain coefficient

The application method is described in more detail as follows.

)에 따른 EMLMAP 알고리즘의 복호수렴 성능을 분석해보면, 도 3에 도시된 바와 같이 각 복호수렴 영역마다 최적의 성능을 나타내는 이득계수(

)가 다른 것을 알 수 있다.First, gain coefficients of 0.6 to 0.9 are obtained using an EXIT (Extrinsic Information Transfer) chart, which is used as a tool for analyzing the convergence behavior of the iterative decoding of a turbo decoder.

Analyzing the decoding convergence performance of the EMLMAP algorithm according to), as shown in FIG.

You can see that is different.

Here, the EXIT chart outputs mutual information I _A = H (X; L _A ) between the Priori LLR (L _A ) and the information block X inputted to the turbo decoder, and is output from the turbo decoder. A graph showing the relationship between mutual information I _E = H (X; L _E ) between an extrinsic LLR (L _E ) and the information block (X), and having a superior curve convergence on the EXIT chart. It means a turbo decoder having a performance.

)가 0.6인 터보 복호기, I_A = 0.1 ~ 0.3 영역에서는 이득계수(

)가 0.7인 터보 복호기, I_A = 0.3 ~ 0.55 영역에서는 이득계수(

)가 0.8인 터보 복호기, I_A = 0.55 ~ 1 영역에서는 이득계수(

)가 0.9인 터보 복호기가 가장 우수한 성능을 가지는 것을 알 수 있다. That is, about I _A = Gain factor in the range 0 ~ 0.1

Turbo Decoder with)), I _A = Gain factor in the range 0.1 to 0.3

Turbo Decoder with), I _A = 0.3 ~ 0.55, gain coefficient (

Turbo Decoder with 0.8), I _A = Gain factor in the range 0.55 to 1

It can be seen that the turbo decoder having a value of 0.9 has the best performance.

4 shows a decoding performance of the EMLMAP algorithm according to the gain coefficient.

)가 각각 존재하며, 터보 복호기가 I_A 값을 평가할 수 있다면

로 표현되는 이득계수(

)를 가변적으로 적용하여 EMLMAP 알고리즘의 성능 최적화를 이룰 수 있다. As shown in FIG. 4, a gain coefficient having an optimal performance for each decoding convergence region (

), And the turbo decoder is I _A If you can evaluate the value

Gain factor expressed as

) Can be applied variably to achieve performance optimization of EMLMAP algorithm.

However, in practical situations, it is impossible to obtain mutual information I _A between the Priori LLR (L _A ) and the information block (X) to which the turbo decoder is input. In order to variably apply gain coefficients, information that can evaluate the degree of decoding convergence other than I _A is needed.

)를 계산하는데, 이에 대하여 더 자세히 설명하면 다음과 같다.To this end, the early stop determination unit 250 of the present invention measures the SDR value to evaluate the decoding convergence degree of the turbo decoder, while the variable gain coefficient calculation unit 260 is optimal for each decoding convergence region based on the SDR value. Variable gain factor with

), Which is explained in more detail as follows.

)는 다음의 수학식 1과 같은 조건식에 의해 계산될 수 있으며, 이와 같은 계산은 가변 이득계수 계산부(260)에서 수행된다.First, if the gain coefficient values are 0.6, 0.7, 0.8, and 0.9 using computer simulations, the average ratios of the size (N) of the coding block of the SDR values corresponding to the boundary points are respectively 0.109, Values of 0.0547 and 0.0183 are calculated, and thus, a variable gain coefficient having optimal performance for each decoding convergence region (

) May be calculated by a conditional expression as shown in Equation 1 below, and the calculation is performed by the variable gain coefficient calculator 260.

는 가변 이득계수를 나타내며, SDR은 부호변화의 개수를 나타낸다.In Equation 1, N represents the size of the coding block,

Denotes a variable gain coefficient and SDR denotes the number of code changes.

에 적용하여 해당 영역에서 최적의 성능을 갖는 가변 이득계수(

)를 구하며, 그 가변 이득계수(

)를 상기 각 구성 복호기(210, 230)의 입력에 곱하여 반복 복호를 수행함으로써, 이에 따라 반복 복호 방식에서 복호 시간을 단축하고 복호에 따른 전력 소모를 최소화하여 저복잡도의 특성을 갖는 EMLMAP 알고리즘의 성능을 최적화할 수 있게 된다.That is, when the SDR value is measured through the early stop determination unit 250, the variable gain coefficient calculation unit 260 converts the measured SDR value as expressed by Equation 1 above.

Variable gain coefficient with optimal performance

) And its variable gain factor (

) Is repeated by multiplying the inputs of the respective decoders 210 and 230 to perform repeated decoding, thereby reducing the decoding time and minimizing power consumption according to the decoding in the iterative decoding scheme. Can be optimized.

(Computer simulation)

The results of comparing and analyzing the efficiency of the turbo decoder of the present invention and the conventional turbo decoder using computer simulations are shown in FIGS. 5 and 6.

In computer simulation, each turbo decoder was applied to WCDMA system, and BER performance and average iterative decoding were analyzed to evaluate the low power efficiency of each turbo decoder. Binary Phase Shift Keying (BPSK) modulation was used, and the maximum number of iterations of the turbo decoder was set to eight.

As shown in FIG. 5, when N = 5114 and R = 1/3, comparing the curves of the BER performance of the turbo decoder of the present invention and the conventional turbo decoder, the turbo decoder of the present invention is an early stop using SDR. Algorithm and improved EMLMAP algorithm using variable gain coefficient are simultaneously applied to show BER performance like rhombus curve, whereas conventional turbo decoder using EMLMAP algorithm using fixed gain coefficient shows BER performance like triangle curve. It can be seen that the BER performance of the turbo decoder of the present invention shows better performance.

That is, it can be seen that the turbo decoder of the present invention has improved performance deterioration when the size (N) of the coding block, which is a disadvantage of the existing EMLMAP algorithm, is improved compared to the conventional turbo decoder.

In addition, as shown in Figure 6, in the case of the average number of iteration decoding, the average number of iteration decoding (diamond curve) of the turbo decoder of the present invention is about 0.3 times less than the average number of iterations (triangle curve) of the conventional turbo decoder It can be seen that the number of repeated decoding is shown. This results in a similar result to the average number of iterations in the GENIE mode, which is a downward limit of the average number of iterations of a conventional turbo decoder.

That is, the turbo decoder of the present invention exhibits the same performance as the conventional turbo decoder with only an average number of repeated decoding times about 0.3 times less than that of the conventional turbo decoder. The power consumption can be reduced by that much.

So far, the present invention has been described with reference to the preferred embodiments, and those skilled in the art to which the present invention belongs may be embodied in a modified form without departing from the essential characteristics of the present invention. You will understand. Therefore, the disclosed embodiments should be considered in descriptive sense only and not for purposes of limitation. The scope of the present invention is shown in the claims rather than the foregoing description, and all differences within the scope will be construed as being included in the present invention.

As described above, according to the present invention, the EMLMAP algorithm is evaluated in comparison with a turbo decoder that applies a fixed gain coefficient by evaluating the degree of decoding convergence of the turbo decoder using an SDR value and applying an optimal variable gain coefficient accordingly. It is possible to optimize the performance of, thereby improving the performance degradation for the case where the size of the coding block, which is a disadvantage of the conventional EMLMAP algorithm is large.

Further, according to the present invention, since the average number of repeated decoding of the turbo decoder can be reduced, the power consumption of the turbo decoder can be reduced accordingly.

Claims (7)

First and second component decoders for performing iterative decoding on bits encoded by a turbo code according to a variable gain coefficient; An early stop determination unit which checks a signal difference ratio (SDR) value and variably limits the number of iteration decoding according to the SDR value; And And a variable gain coefficient calculation unit for calculating a variable gain coefficient in each decoding convergence region according to the SDR value. The method of claim 1, wherein the first and second component decoders, A low complexity and low power turbo decoder using a variable gain coefficient, characterized in that iterative decoding is performed while exchanging additional information with each other through an EMLMAP (Enhanced Max-Log-MAP) algorithm. The method of claim 1, wherein the early stop determination unit, Measuring the SDR value from the outputs of the first and second component decoders so that repetitive decoding is stopped when the SDR value is '0' and repeating decoding is continued when the SDR value is not '0' Low Complexity and Low Power Turbo Decoder Using Variable Gain Coefficient. The method of claim 1, wherein the variable gain coefficient (

),

Where N represents the size of the coding block,

Denotes the variable gain coefficient, and SDR denotes the number of code changes) Low complexity and low power turbo decoder using a variable gain coefficient, characterized in that calculated by. The method of claim 1, An interleaver for matching the order of the bit strings between the first and second component decoders, and a deinterleaver for rearranging the information rearranged by the interleaver in the same order as the original information bits. Low complexity and low power turbo decoder using a variable gain coefficient, characterized in that. The method of claim 2, And a first and second adders for generating new additional information by adding input values and output values of the first and second component decoders. The method of claim 6, And a first and second multipliers for multiplying the new additional information by the variable gain coefficient.

Images