CN101232349B

CN101232349B - Method for Quickly Generating QAM Bit Confidence Soft Decision Metrics

Info

Publication number: CN101232349B
Application number: CN200710173300XA
Authority: CN
Inventors: 赵晋; 张建秋
Original assignee: SHANGHAI FUDAN MICRONANO ELECTRONICS CO Ltd; Fudan University
Current assignee: SHANGHAI FUDAN MICRONANO ELECTRONICS CO Ltd; Fudan University
Priority date: 2007-12-27
Filing date: 2007-12-27
Publication date: 2011-08-03
Anticipated expiration: 2027-12-27
Also published as: CN101232349A

Abstract

The invention belongs to the technical field of wireless communication demodulation, which is a method for the rapid generation of a QAM bit confidence level soft decision metric. The channel decoders in a modern communication system all adopt the soft input technical, so as to ensure the high channel decoding gain. The traditional soft decision metric is mostly generated based on the minimum distance criteria. The invention proposes the method for the rapid generation of the QAM soft decision metric on the basis of the information of the bit confidence level, which introduces the bit confidence level to the calculation of the QAM soft decision metric to reduce the complexity of the calculation. The method is applied in the China digital television ground transmission standard DTMB system for simulation, the results show that the soft decision metric generation algorithm significantly reduces the calculation amount and the performances are not damaged almost, so the method has good performance and realizability.

Description

Method for Quickly Generating QAM Bit Confidence Soft Decision Metrics

技术领域technical field

本发明属于无线通信解调制技术领域，具体涉及一种快速生成QAM比特置信度软判决度量的方法。The invention belongs to the technical field of wireless communication demodulation, and in particular relates to a method for rapidly generating QAM bit confidence degree soft decision measure.

背景技术Background technique

随着现代通信技术的发展，在移动及个人通信系统中对数据传输速率的要求越来越高。一种有效的提高频谱利用率的方法是在调制时采用多元调制方式(如64QAM)，但在发射功率一定的情况下，采用多元调制会减小调制星座点之间的欧式距离，增加系统的误码率，因此在采用多元调制的系统中，均采用了性能优异的信道编码方式(如Turbo码或LDPC码等)来弥补误码率的损失。为了使这些信道解码器更好的工作，在解映射时，均采用软判决技术来代替传统的硬判决技术，以保证信道解码器获得更多的信道信息。研究结果表明^[5]，在加性高斯噪声(AWGN)信道中，软判决比硬判决解码要多2dB的软判决增益，而在衰落信道中的软判决增益则超过3dB。With the development of modern communication technology, the requirements for data transmission rate in mobile and personal communication systems are getting higher and higher. An effective way to improve spectrum utilization is to use multi-element modulation (such as 64QAM) during modulation. However, when the transmit power is constant, the use of multi-element modulation will reduce the Euclidean distance between modulation constellation points and increase the system efficiency. Therefore, in systems using multi-element modulation, channel coding methods with excellent performance (such as Turbo codes or LDPC codes, etc.) are used to compensate for the loss of bit error rates. In order to make these channel decoders work better, soft decision technology is used to replace traditional hard decision technology during demapping, so as to ensure that the channel decoder obtains more channel information. The research results show that ^[5] , in the additive Gaussian noise (AWGN) channel, the soft decision gain is 2dB more than the hard decision decoding, and the soft decision gain in the fading channel is more than 3dB.

传统的软判决度量生成算法多基于最小欧式距离^[6]，是一种基于ML(最大似然)准则的解调方式。在计算欧式距离时，需要对复数的模进行平方运算，对硬件的开销较大。Traditional soft decision metric generation algorithms are mostly based on minimum Euclidean distance ^[6] , which is a demodulation method based on ML (maximum likelihood) criterion. When calculating the Euclidean distance, it is necessary to perform a square operation on the modulus of the complex number, which has a large overhead on hardware.

文献[3]中提到了比特置信度的概念，但其只是简单的把比特置信度与信道状态信息(CSI)的乘积作为软判决度量，并没有在数学上给出严格的推导。The concept of bit confidence is mentioned in [3], but it simply takes the product of bit confidence and channel state information (CSI) as a soft decision measure, and does not give a strict mathematical derivation.

本发明提出一种针对QAM调制方式，快速生成比特置信度软判决度量的方法，通过将比特置信度引入到软判决度量的计算中来减少计算的复杂性，相比于传统的基于最小欧式距离的软判决度量计算方法，本方法在几乎不损失任何性能的前提下，减小了计算量，降低了实现复杂度。The present invention proposes a method for quickly generating bit confidence soft decision metrics for QAM modulation, and reduces the complexity of calculation by introducing bit confidence into the calculation of soft decision metrics. Compared with the traditional method based on the minimum Euclidean distance The calculation method of the soft decision metric, this method reduces the amount of calculation and the implementation complexity under the premise of almost no loss of any performance.

发明内容Contents of the invention

本发明的目的在于提出一种计算量小、实现复杂度低的生成QAM比特置信度软判决度量的方法。The object of the present invention is to propose a method for generating QAM bit confidence soft decision metrics with small calculation amount and low implementation complexity.

在存在衰落的信道环境，假设接收机理想同步，接收到的单个符号信息可以表示为：In a fading channel environment, assuming that the receiver is ideally synchronized, the received single symbol information can be expressed as:

Y＝HX+W (1)Y＝HX+W (1)

其中H为信道在符号X处的复数增益，W为加性复高斯白噪声，X和Y分别为发送和接收到的符号。where H is the complex gain of the channel at symbol X, W is additive complex white Gaussian noise, and X and Y are the transmitted and received symbols, respectively.

符号X所对应的比特信息中第i位b_i的对数似然比(LLR)信息的表达式为：The expression of the logarithmic likelihood ratio (LLR) information of the i-th bit b _i in the bit information corresponding to the symbol X is:

$LLR LLR (({b b}_{i i})) = = ln ln \frac{p p (({b b}_{i i} = = 00 | | Y Y))}{p p (({b b}_{i i} = = 11 | | Y Y))} - - - - - - ((22))$

其中p(*)表示条件概率。where p(*) represents the conditional probability.

当X满足均匀分布时，LLR可以表示为：When X satisfies a uniform distribution, LLR can be expressed as:

$LLR LLR (({b b}_{i i})) = = ln ln \frac{p p (({Y Y | | b b}_{i i} = = 00))}{p p (({Y Y | | b b}_{i i} = = 11))} - - - - - - ((33))$

将式(1)代入式(3)，符号比特b_i的LLR信息满足Substituting equation (1) into equation (3), the LLR information of sign bit b _i satisfies

$LLR LLR (({b b}_{i i})) = = ln ln \frac{\underset{{s the s}^{00} &Element; &Element; {{s the s : : {b b}_{i i} = = 00}}}{Σ Σ} exp exp ((- - \frac{{| | Y Y - - H h {s the s}^{00} | |}^{22}}{{σ σ}_{n no}^{22}}))}{\underset{{s the s}^{11} &Element; &Element; {{s the s : : {b b}_{i i} = = 11}}}{Σ Σ} exp exp ((- - \frac{{| | Y Y - - H h {s the s}^{11} | |}^{22}}{{σ σ}_{n no}^{22}}))} - - - - - - ((44))$

其中s⁰是满足b_i＝0的所有星座点，s¹是满足b_i＝1的所有星座点，σ_n ²是高斯白噪声的方差。Where s ⁰ is all constellation points satisfying b _i =0, s ¹ is all constellation points satisfying b _i =1, and σ _n ² is the variance of Gaussian white noise.

由于指数运算随着自变量的增大而迅速增加，因此在指数求和运算中，自变量最大的一项将对最后的结果占主导作用，式(4)可以近似表示为^[2] Since the exponential operation increases rapidly with the increase of the independent variable, in the exponential summation operation, the item with the largest independent variable will play a dominant role in the final result, and the formula (4) can be approximately expressed as ^[2]

$LLR LLR (({b b}_{i i})) \approx \approx ln ln \frac{\underset{{s the s}^{00} &Element; &Element; {{s the s : : {b b}_{i i} = = 00}}}{max max} [[exp exp ((- - \frac{{| | Y Y - - H h {s the s}^{00} | |}^{22}}{{σ σ}_{n no}^{22}}))]]}{\underset{{s the s}^{11} &Element; &Element; {{s the s : : {b b}_{i i} = = 11}}}{max max} [[exp exp ((- - \frac{{| | Y Y - - H h {s the s}^{11} | |}^{22}}{{σ σ}_{n no}^{22}}))]]}$

$= = ln ln \frac{\underset{{s the s}^{00} &Element; &Element; {{s the s : : {b b}_{i i} = = 00}}}{max max} [[exp exp ((- - \frac{{| | Y Y / / H h - - {s the s}^{00} | |}^{22} {| | H h | |}^{22}}{{σ σ}_{n no}^{22}}))]]}{\underset{{s the s}^{11} &Element; &Element; {{s the s : : {b b}_{i i} = = 11}}}{max max} [[exp exp ((- - \frac{{| | Y Y / / H h - - {s the s}^{11} | |}^{22} {| | H h | |}^{22}}{{σ σ}_{n no}^{22}}))]]} - - - - - - ((55))$

$= = ln ln \frac{\underset{{s the s}^{00} &Element; &Element; {{s the s : : {b b}_{i i} = = 00}}}{max max} [[exp exp ((- - \frac{{| | Z Z - - {s the s}^{00} | |}^{22} {| | H h | |}^{22}}{{σ σ}_{n no}^{22}}))]]}{\underset{{s the s}^{11} &Element; &Element; {{s the s : : {b b}_{i i} = = 11}}}{max max} [[exp exp ((- - \frac{{| | Z Z - - {s the s}^{11} | |}^{22} {| | H h | |}^{22}}{{σ σ}_{n no}^{22}}))]]}$

其中Z＝Y/H为均衡后的符号。Where Z=Y/H is the equalized symbol.

将分子中满足最大项的s⁰记为s_m ⁰，分子中满足最大项的s¹记为s_m ¹。式(5)可以重写为：The s ⁰ that satisfies the maximum term in the numerator is recorded as s _m ⁰ , and the s ¹ that satisfies the maximum term in the numerator is recorded as s _m ¹ . Equation (5) can be rewritten as:

$LLR LLR (({b b}_{i i})) = = (({| | Z Z - - {s the s}_{}^{m m} | |}^{22} - - {| | Z Z - - {s the s}_{m m}^{00} | |}^{22})) \cdot &Center Dot; \frac{{| | H h | |}^{22}}{{σ σ}_{n no}^{22}} - - - - - - ((66))$

式(6)为衰落信道中，基于最小欧式距离的软判决度量^[6]。Equation (6) is the soft decision metric based on the minimum Euclidean distance in a fading channel ^[6] .

在得到信道增益H、信道噪声方差σ_n ²以及均衡化的符号Z之后，可以利用式(6)计算LLR信息。但是在式(6)中，每个比特对应的s_m ¹、s_m ⁰各不相同，需要分别进行计算，且存在复数绝对值平方的运算，这在硬件实现时会造成比较大的开销。考虑QAM调制方式的特点，将比特置信度引入软判决度量的计算中，可以对式(6)的计算进行简化。After obtaining the channel gain H, the channel noise variance σ _n ² and the equalized symbol Z, the LLR information can be calculated using formula (6). However, in Equation (6), the s _m ¹ and s _m ⁰ corresponding to each bit are different and need to be calculated separately, and there is an operation of the square of the absolute value of the complex number, which will cause relatively large overhead in hardware implementation. Considering the characteristics of the QAM modulation mode, the calculation of the formula (6) can be simplified by introducing the bit confidence into the calculation of the soft decision metric.

本发明将比特置信度引入软判决度量的计算中，给出一种计算量小，实现复杂度低的生成QAM比特置信度软判决度量的方法。具体步骤如下：The invention introduces the bit confidence degree into the calculation of the soft decision metric, and provides a method for generating the QAM bit confidence soft decision metric with small calculation amount and low complexity. Specific steps are as follows:

1、计算比特置信度1. Calculate bit confidence

在采用格雷码编码的QAM调制方式中(星座图如图1所示)，定义比特b_i的比特置信度信息为均衡后的符号到判决门限的欧式距离，并带有正负号，记为m_i。m_i的符号决定了判决的结果：当m_i大于0时，对应的比特位判决为1；小于0时，对应的比特位判决为0；m_i的绝对值代表了判决的置信程度。例如，虽然当m_i＝0.1和m_i＝10时，我们都将比特i判为1，但显然当m_i＝10时我们对判决的结果更有信心。由于采用格雷码编码方式，符号中的每一比特的判决均只与符号的实部或虚部有关。各种QAM调制方式下比特置信度的表达式如式(7)(8)(9)所示。In the QAM modulation method using Gray code coding (constellation diagram shown in Figure 1), the bit confidence information of bit _bi is defined as the Euclidean distance from the equalized symbol to the decision threshold, with a sign, denoted as m _i . The sign of _mi determines the result of the judgment: when _mi is greater than 0, the corresponding bit is judged as 1; when it is less than 0, the corresponding bit is judged as 0; the absolute value of _mi represents the confidence level of the judgment. For example, although we judge bit i as 1 when _mi =0.1 and _mi =10, obviously we are more confident in the judgment result when _mi =10. Due to the Gray code encoding method, the judgment of each bit in the symbol is only related to the real or imaginary part of the symbol. The expressions of bit confidence in various QAM modulation modes are shown in formulas (7)(8)(9).

4-QAM调制方式下各个比特的置信度度量为：The confidence measure of each bit in the 4-QAM modulation mode is:

m₁＝Im(Z) m₀＝Re(Z) (7)m ₁ =Im(Z) m ₀ =Re(Z) (7)

16-QAM调制方式下各个比特的置信度度量为：The confidence measure of each bit in the 16-QAM modulation mode is:

m₃＝Im(Z) m₂＝Re(Z) (8)m ₃ =Im(Z) m ₂ =Re(Z) (8)

m₁＝4-|Im(Z)| m₀＝4-|Re(Z)|m ₁ =4-|Im(Z)| m ₀ =4-|Re(Z)|

64-QAM调制方式下各个比特的置信度度量为：The confidence measure of each bit in the 64-QAM modulation mode is:

m₅＝Im(Z)； m₂＝Re(Z)；m ₅ =Im(Z); m ₂ =Re(Z);

m₄＝4-|Im(Z)|； m₁＝4-|Re(Z)|； (9)m ₄ =4-|Im(Z)|; m ₁ =4-|Re(Z)|; (9)

${m m}_{33} = = \{\begin{matrix} | | Im Im ((Z Z)) | | - - 22 & | | Im Im ((Z Z)) | | \leq \leq 44 \\ - - | | Im Im ((Z Z)) | | + + 66 & | | Im Im ((Z Z)) | | > > 44 \end{matrix}$ ${m m}_{00} = = \{\begin{matrix} | | Re Re ((Z Z)) | | - - 22 & | | Re Re ((Z Z)) | | \leq \leq 44 \\ - - | | Re Re ((Z Z)) | | + + 66 & | | Re Re ((Z Z)) | | > > 44 \end{matrix}$

其中Z＝Y/H为均衡后的符号，Im(Z)表示符号Z的虚部，Re(Z)表示符号Z的实部；Wherein Z=Y/H is the symbol after equalization, and Im(Z) represents the imaginary part of symbol Z, and Re(Z) represents the real part of symbol Z;

2、生成软判决度量2. Generating Soft Decision Metrics

假设比特b_i的判决只与符号虚部有关，式(6)可以简化为：Assuming that the decision of bit _bi is only related to the imaginary part of the symbol, formula (6) can be simplified as:

$LLR LLR (({b b}_{i i})) = = \frac{[[22 * * Im Im ((Z Z)) - - Im Im (({s the s}_{m m}^{00})) - - Im Im (({s the s}_{m m}^{11}))]] [[Im Im (({s the s}_{m m}^{00})) - - Im Im (({s the s}_{m m}^{11}))]] {| | H h | |}^{22}}{{σ σ}_{n no}^{22}} - - - - - - ((1010))$

$= = \frac{{{22 * * [[Im Im ((Z Z)) - - T T {h h}_{i i}]] - - [[Im Im (({s the s}_{m m}^{00})) - - T T {h h}_{i i}]] - - [[Im Im (({s the s}_{m m}^{11})) - - T T {h h}_{i i}]]}} {{[[Im Im (({s the s}_{m m}^{00})) - - T T {h h}_{i i}]] - - [[Im Im (({s the s}_{m m}^{11})) - - T T {h h}_{i i}]]}} {| | H h | |}^{22}}{{σ σ}_{n no}^{22}}$

其中Th_i为判决门限。在式(10)中，Im(Z)-Th_i可能为m_i或-m_i。分析星座图可以得到，当Im(Z)-Th_i＝m_i时， $Im (s_{m}^{0}) - T h_{i} = - | Im (s_{m}^{0}) - T h_{i} |$ $Im (s_{m}^{1}) - T h_{i} = | Im (s_{m}^{1}) - T h_{i} |;$ 当Im(Z)-Th_i＝-m_i时， $Im (s_{m}^{0}) - T h_{i} = | Im (s_{m}^{0}) - T h_{i} |,$ $Im (s_{m}^{1}) - T h_{i} = - | Im (s_{m}^{1}) - T h_{i} | .$ 因此，式(10)可以重写为：Where Th _i is the decision threshold. In formula (10), Im(Z)-Th _i may be m _i or -m _i . By analyzing the constellation diagram, it can be obtained that when Im(Z)-Th _i =m _i , $Im ({the s}_{m}^{0}) - T h_{i} = - | Im ({the s}_{m}^{0}) - T h_{i} |$ $Im ({the s}_{m}^{1}) - T h_{i} = | Im ({the s}_{m}^{1}) - T h_{i} |;$ When Im(Z)-Th _i =-m _i , $Im ({the s}_{m}^{0}) - T h_{i} = | Im ({the s}_{m}^{0}) - T h_{i} |,$ $Im ({the s}_{m}^{1}) - T h_{i} = - | Im ({the s}_{m}^{1}) - T h_{i} | .$ Therefore, equation (10) can be rewritten as:

$LLR LLR (({b b}_{i i})) = = \frac{- - ((22 * * {m m}_{i i} + + | | Im Im (({s the s}_{m m}^{00})) - - T T {h h}_{i i} | | - - | | Im Im (({s the s}_{m m}^{11})) - - T T {h h}_{i i} | |)) ((| | Im Im (({s the s}_{m m}^{00})) - - T T {h h}_{i i})) | | + + | | Im Im (({s the s}_{m m}^{11})) - - T T {h h}_{i i} | | {| | H h | |}^{22}}{{σ σ}_{n no}^{22}} - - - - - - ((1111))$

在某种特定的QAM调制方式下，当m_i＞0时，|Im(s_m ⁰)-Th_i|为一个确定的值，同样地，当m_i＜0时，|Im(s_m ¹)-Th_i|也为一个确定的值，且两个值相同，这里设为p。在不同的QAM调制方式下，p的取值有所不同。将p代入式(11)中，当m_i＜0时，式(11)可重写为：In a specific QAM modulation mode, when m _i >0, |Im(s _m ⁰ )-Th _i | is a definite value, similarly, when m _i <0, |Im(s _m ¹ )-Th _i | is also a definite value, and the two values are the same, here it is set to p. Under different QAM modulation modes, the value of p is different. Substituting p into formula (11), when m _i <0, formula (11) can be rewritten as:

$LLR LLR (({b b}_{i i})) = = \frac{- - ((22 * * {m m}_{i i} + + | | Im Im (({s the s}_{m m}^{00})) - - T T {h h}_{i i} | | - - p p)) ((| | Im Im (({s the s}_{m m}^{00})) - - T T {h h}_{i i} | | + + p p)) {| | H h | |}^{22}}{{σ σ}_{n no}^{22}}$

$= = \frac{- - sign sign (({m m}_{i i})) ((22 * * | | {m m}_{i i} | | - - | | Im Im (({s the s}_{m m}^{00})) - - T T {h h}_{i i} | | + + p p)) ((| | Im Im (({s the s}_{m m}^{00})) - - T T {h h}_{i i} | | + + p p)) {| | H h | |}^{22}}{{σ σ}_{n no}^{22}} - - - - - - ((1212))$

当m_i＞0时，式(11)可重写为：When m _i >0, formula (11) can be rewritten as:

$LLR LLR (({b b}_{i i})) = = \frac{- - ((22 {m m}_{i i} + + | | Im Im (({s the s}_{m m}^{11})) - - T T {h h}_{i i} | | + + p p)) ((| | Im Im (({s the s}_{m m}^{11})) - - T T {h h}_{i i} | | + + p p)) {| | H h | |}^{22}}{{σ σ}_{n no}^{22}}$

$= = \frac{- - sign sign (({m m}_{i i})) ((22 * * | | {m m}_{i i} | | - - | | Im Im (({s the s}_{m m}^{11})) - - T T {h h}_{i i} | | + + p p)) ((| | Im Im (({s the s}_{m m}^{11})) - - T T {h h}_{i i} | | + + p p)) {| | H h | |}^{22}}{{σ σ}_{n no}^{22}} - - - - - - ((1313))$

通过分析星座图，可以根据m_i直接计算出|Im(s_m ⁰)-Th_i|或|Im(s_m ¹)-Th_i|，在不同的调制方式下，它们的表达式有所不同，这里统一用q来表示，将q代入式(12)和(13)中，得到比特b_i的LLR信息的通式为：By analyzing the constellation diagram, |Im(s _m ⁰ )-Th _i | or |Im(s _m ¹ )-Th _i | can be directly calculated according to m _i , and their expressions are different under different modulation methods , which is uniformly represented by q here, and substituting q into formulas (12) and (13), the general formula for obtaining the LLR information of bit _bi is:

$LLR LLR (({b b}_{i i})) = = \frac{- - sign sign (({m m}_{i i})) [[22 * * | | {m m}_{i i} | | - - q q + + p p]] [[q q + + p p]] {| | H h | |}^{22}}{{σ σ}_{n no}^{22}} - - - - - - ((1414))$

3、简化表达形式3. Simplified expression

在4-QAM、16-QAM以及64-QAM调制方式下，p的取值分别为1、2和4.5，q的表达式分别为floor(|m_i|/2)*2+p、floor(|m_i|/4)*4+p和p。In 4-QAM, 16-QAM and 64-QAM modulation modes, the values of p are 1, 2 and 4.5 respectively, and the expressions of q are floor(|m _i |/2)*2+p, floor( |m _i |/4)*4+p and p.

由于发送符号X满足均匀分布，在64-QAM调制方式下，从式(9)中分析可知，floor(|m_i|/2)等于0的概率为58.3％，等于1的概率为25％，等于2的概率为8.3％，等于3的概率为8.3％；在16-QAM调制方式下，从式(8)中分析可知，floor(|m_i|/4)等于0的概率为75％，等于1的概率为25％。为了计算的方便，将上述两项均取为0，即在各种QAM调试方式下，q均等于p。Since the transmitted symbol X satisfies a uniform distribution, in the 64-QAM modulation mode, it can be seen from the analysis of formula (9) that the probability that floor(| _mi |/2) is equal to 0 is 58.3%, and the probability that it is equal to 1 is 25%. The probability of being equal to 2 is 8.3%, and the probability of being equal to 3 is 8.3%. In the 16-QAM modulation mode, it can be seen from the analysis of formula (8) that the probability of floor(| _mi |/4) being 0 is 75%, The probability of equaling 1 is 25%. For the convenience of calculation, the above two items are taken as 0, that is, in various QAM debugging modes, q is equal to p.

简化后，各种QAM调制方式下LLR的通式可以表示为：After simplification, the general formula of LLR under various QAM modulation modes can be expressed as:

$LLR LLR (({b b}_{i i})) = = - - 44 p p {m m}_{i i} \cdot &Center Dot; \frac{{| | H h | |}^{22}}{{σ σ}_{n no}^{22}} - - - - - - ((1515))$

其中，在4-QAM、16-QAM以及64-QAM调制方式下，p的取值分别为1、2和4.5。Wherein, in the 4-QAM, 16-QAM and 64-QAM modulation modes, the values of p are 1, 2 and 4.5 respectively.

对比式(15)和式(6)，从中可以看出，本发明提出方法的运算量要明显低于传统基于最小欧式距离方法的运算量。Comparing Equation (15) and Equation (6), it can be seen that the calculation amount of the method proposed by the present invention is obviously lower than that of the traditional method based on the minimum Euclidean distance.

在具体计算QAM比特置信度软判决度量时，第一步先根据式(7)(8)(9)计算比特置信度信息，第二步根据式(15)生成软判决度量。When calculating the QAM bit confidence soft decision metric, the first step is to calculate the bit confidence information according to formula (7), (8) and (9), and the second step is to generate the soft decision metric according to formula (15).

技术效果technical effect

本发明为一种快速生成QAM比特置信度软判决度量的方法。将比特置信度引入LLR的计算中，来生成软判决度量。本文提出的软判决度量生成算法可以应用于DTMB系统^[1]的外接收机中。The invention is a method for rapidly generating QAM bit confidence soft decision measure. The bit confidence is introduced into the calculation of LLR to generate a soft decision metric. The soft decision metric generation algorithm proposed in this paper can be applied to the outer receiver of DTMB system ^[1] .

附图说明Description of drawings

图1为QAM星座图，其中(a)为4-QAM星座图，(b)为16-QAM星座图，(c)为64-QAM星座图。FIG. 1 is a QAM constellation diagram, wherein (a) is a 4-QAM constellation diagram, (b) is a 16-QAM constellation diagram, and (c) is a 64-QAM constellation diagram.

图2为DTMB系统接收机框图。Figure 2 is a block diagram of the DTMB system receiver.

图3为加性高斯信道环境下，两种软判决度量对LDPC解码器性能影响的比较，其中(a)为16-QAM的情况，(b)为64-QAM的情况。Figure 3 is a comparison of the impact of two soft decision metrics on the performance of an LDPC decoder under an additive Gaussian channel environment, where (a) is the case of 16-QAM, and (b) is the case of 64-QAM.

图4为多径衰落信道环境下，两种软判决度量对LDPC解码器性能影响的比较，其中(a)为16-QAM的情况，(b)为64-QAM的情况。Figure 4 is a comparison of the impact of two soft decision metrics on the performance of the LDPC decoder in a multipath fading channel environment, where (a) is the case of 16-QAM, and (b) is the case of 64-QAM.

具体实施方式Detailed ways

下面通过仿真来进一步描述本发明，仿真的具体步骤如下：Further describe the present invention by emulation below, the concrete steps of emulation are as follows:

1.发射机发送的数据符号经过多径衰落信道后，与高斯白噪声叠加，进入接收机。1. After passing through the multipath fading channel, the data symbols sent by the transmitter are superimposed with Gaussian white noise and enter the receiver.

2.假设接收机理想同步与信道均衡，根据式(7)(8)(9)计算比特置信度信息。2. Assuming ideal synchronization and channel equalization of the receiver, the bit confidence information is calculated according to equations (7)(8)(9).

3.利用步骤2得到的比特置信度信息，根据式(15)计算软判决度量。3. Using the bit confidence information obtained in step 2, calculate the soft decision metric according to formula (15).

4.将生成的软判决度量，输入信道解码器，进行信道解码。4. Input the generated soft decision metrics into a channel decoder for channel decoding.

5.DTMB系统接收机框图，以及实现本文算法的模块如图2所示。5. The block diagram of the DTMB system receiver and the modules implementing the algorithm in this paper are shown in Figure 2.

仿真结果：Simulation results:

1.仿真条件：1. Simulation conditions:

在仿真中，使用了16-QAM和64-QAM调制方式，采用的LDPC码为(7493，6096)码，码率为0.8。在仿真中使用的多经衰落信道模型如表1所示，且为准静态信道，即在一个OFDM符号的时间内，信道不发生变化。在仿真中，均假设理想同步与信道均衡，LDPC解码算法采用了文献[4]中的算法，最大迭代次数设定为50次。In the simulation, 16-QAM and 64-QAM modulation methods are used, the LDPC code adopted is (7493, 6096) code, and the code rate is 0.8. The multi-pass fading channel model used in the simulation is shown in Table 1, and it is a quasi-static channel, that is, the channel does not change within the time of one OFDM symbol. In the simulation, ideal synchronization and channel equalization are assumed. The LDPC decoding algorithm adopts the algorithm in [4], and the maximum number of iterations is set to 50.

2.实验结果：2. Experimental results:

为了评估本发明提出算法的性能，对于经过信道均衡后的符号，分别采用本文基于比特置信度的方法和基于最小欧式距离的方法来生成软判决度量，将两种方法得到的LLR信息分别输入LDPC解码器中。In order to evaluate the performance of the algorithm proposed by the present invention, for the symbols after channel equalization, the method based on bit confidence and the method based on minimum Euclidean distance are respectively used to generate soft decision metrics, and the LLR information obtained by the two methods are respectively input into LDPC in the decoder.

在加性高斯信道和多径衰落信道环境下，两种软判决度量对LDPC解码器性能影响的数值仿真结果如图3、图4所示。In the additive Gaussian channel and multipath fading channel environment, the numerical simulation results of the impact of two soft decision metrics on the performance of the LDPC decoder are shown in Figure 3 and Figure 4.

从图3、图4所示的误码率曲线可以看出，虽然本文给出的算法对基于最小欧式距离的方法进行了简化，减小了运算量，但无论是在加性高斯信道环境下，还是在多径衰落信道环境下，信道解码器的性能却几乎任何损失，是一种“性价比”非常高的软判决度量生成算法。It can be seen from the bit error rate curves shown in Figure 3 and Figure 4 that although the algorithm given in this paper simplifies the method based on the minimum Euclidean distance and reduces the amount of calculation, no matter in the additive Gaussian channel environment , or in the multipath fading channel environment, the performance of the channel decoder is almost any loss, and it is a very high "cost-effective" soft decision metric generation algorithm.

表1 ETSI多径信道模型Table 1 ETSI multipath channel model

路径path 00 11 22 33 44 55 66 77 8 8 99 1010 幅度(dB)Amplitude(dB) 00 -7.8-7.8 -24.8-24.8 -15-15 -10.4-10.4 -11.7-11.7 -24.2-24.2 -16.5-16.5 -25.8-25.8 -14.7-14.7 -7.9-7.9 延时(us)Delay (us) 00 0.520.52 1.001.00 5.425.42 2.752.75 0.600.60 1.021.02 0.140.14 0.150.15 3.323.32 1.931.93 相位(°)Phase (°) 00 336336 278.2278.2 195.9195.9 127127 215.3215.3 311.1311.1 226.4226.4 62.762.7 330.9330.9 8.88.8 路径path 1111 1212 1313 1414 1515 1616 1717 1818 1919 2020 幅度(dB)Amplitude(dB) -10.6-10.6 -9.1-9.1 -11.6-11.6 -12.9-12.9 -15.3-15.3 -16.5-16.5 -12.4-12.4 -18.7-18.7 -12.1-12.1 -11.7-11.7 延时(us)Delay (us) 0.430.43 3.223.22 0.850.85 0.070.07 0.20.2 0.190.19 0.920.92 1.381.38 0.640.64 1.371.37 相位(°)Phase (°) 339.7339.7 174.9174.9 3636 122122 6363 198.4198.4 210210 162.4162.4 191191 22.622.6

参考文献references

[1]GB20600-2006，《地面数字电视广播传输系统帧结构、信道编码和调制技术规范》，DigitalTerrestrial Multimedia Broadcasting，2006年第8号(总第95号)中国国家标准发布公告.[1] GB20600-2006, "Technical Specifications for Frame Structure, Channel Coding and Modulation of Terrestrial Digital Television Broadcasting Transmission System", Digital Terrestrial Multimedia Broadcasting, 2006 No. 8 (Total No. 95) China National Standard Announcement.

[2]Michael Mao Wang，Weimin Xiao and Tyler Brown.Soft Decision Metric Generation forQAM With Channel Estimation Error[J].IEEE Trans.on Communications，2002，50(7)：1058-1061.[2]Michael Mao Wang, Weimin Xiao and Tyler Brown.Soft Decision Metric Generation forQAM With Channel Estimation Error[J].IEEE Trans.on Communications,2002,50(7):1058-1061.

[3]王勇，艾渤，葛建华.数字电视地面广播COFDM传输中的软判决技术[J].通信学报，2003，24(9)：73-79.[3] Wang Yong, Ai Bo, Ge Jianhua. Soft decision technology in COFDM transmission of digital TV terrestrial broadcasting [J]. Journal of Communications, 2003, 24(9): 73-79.

[4]Xiao-Yu Hu，Eleftheriou，E.Amold and Dholakia，A.Elftheriou.Efficient Implementationsof the Sum-Prodect Algorithm for Decoding LDPC Codes[J].IEEE GlobalTelecommunications Conference，2001，2(11)：1036-2036[4] Xiao-Yu Hu, Eleftheriou, E.Amold and Dholakia, A.Elftheriou. Efficient Implementations of the Sum-Prodect Algorithm for Decoding LDPC Codes[J]. IEEE Global Telecommunications Conference, 2001, 2(11): 1036-2036

[5]Jean W G，Chang K H，Gho Y S.An Equalization Technique for Orthogonal FrequencyDivision Multiplexing Systems in Time Variant Multi-path Channels[J].IEEE.Trans.OnCommunications.1999，47(1)：27-32.[5] Jean W G, Chang K H, Gho Y S.An Equalization Technique for Orthogonal Frequency Division Multiplexing Systems in Time Variant Multi-path Channels[J].IEEE.Trans.OnCommunications.1999,47(1):27-32 .

[6]Moe Rahnema，Yezdi Antia，Hughers Network Systems.Optimum Soft Decision Decodingwith Channel State Information in the Presence of Fading[J].IEEE CommunicationsMagazine，1997，35(7)：110-111.[6] Moe Rahnema, Yezdi Antia, Hughers Network Systems. Optimum Soft Decision Decoding with Channel State Information in the Presence of Fading [J]. IEEE Communications Magazine, 1997, 35(7): 110-111.

Claims

1. A method for quickly generating QAM bit confidence soft-decision metrics, assuming that in a fading channel environment, the received single signal symbol information is expressed as:

Y＝HX+W (1)

where H is the complex gain of the channel at symbol X, W is additive complex Gaussian white noise, X and Y are the symbols sent and received respectively; denote LLR(bi ₎ as the ith bit information corresponding to symbol X The log-likelihood ratio of bit b _i , the specific steps are as follows:

(1) Set and calculate bit confidence

Define the bit confidence information of bit b _i as the Euclidean distance from the equalized symbol to the decision threshold, with a sign, denoted as m _i , then:

The confidence measure of each bit in the 4-QAM modulation mode is:

m ₁ =Im(Z) m ₀ =Re(Z) (7)

The confidence measure of each bit in the 16-QAM modulation mode is:

m ₃ =Im(Z) m ₂ =Re(Z) (8)

m ₁ =4-|Im(Z)| m ₀ =4-|Re(Z)|

The confidence measure of each bit in the 64-QAM modulation mode is:

m ₅ =Im(Z); m ₂ =Re(Z);

m ₄ =4-|Im(Z)|; m ₁ =4-|Re(Z)|; (9)

Wherein Z=Y/H is the symbol after equalization, and Im(Z) represents the imaginary part of symbol Z, and Re(Z) represents the real part of symbol Z;

(2) Metrics for Generating Soft Decisions

in,

q is directly calculated according to _mi

or

is the variance of Gaussian white noise, s ⁰ _m is s ⁰ satisfying the maximum term, and s ⁰ is all constellation points satisfying b _i =0, s ¹ _m is s ¹ satisfying the maximum term, and s ¹ is satisfying b For all constellation points where _i = 1, Th _i is the decision threshold.

2. method according to claim 1 is characterized in that under 4-QAM, 16-QAM and 64-QAM modulation mode, the value of p is respectively 1,2 and 4.5, and the expression of q is respectively floor( |m _i |/2)*2+p, floor(|m _i |/4)*4+p, and p.

3. The method according to claim 1, characterized in that q equals p in the step (2), then:

.

4. The method according to claim 3, characterized in that under 4-QAM, 16-QAM and 64-QAM modulation modes, the values of p are 1, 2 and 4.5 respectively. the