CN1787510B - Method and apparatus for demodulation of 16 grade orthogonal amplitude modulation in digital communication system - Google Patents

Abstract

This invention provides a modem method for 16-stage orthogonal amplitude in a digital communication system including the following steps: segmenting a LLR function on an interval, fitting each section with one stage function and computing the LLR value of each bit by one stage function, deriving and simplifying the LLR formula to get a simplified approximate formula. The modem includes the following parts: a normalized module, a virtual and solid separation module and one stage function fitting module, which uses an approximate formula to replace the original complicated LLR to reduce the complexity of the algorithm greatly.

Description

The demodulation method of 16 grades of quadrature amplitude modulation and device in the digital communication system

Technical field

The present invention relates to 16 grades of QAM (Quadrature Amplitude Modulation in the digital communication system, quadrature amplitude modulation) demodulation method and device more specifically are demodulation method and the devices that how 16QAM is carried out soft output in wireless communication system.

Background technology

In digital communication system, the information source data are carried out can using 16QAM to come data are modulated usually after the chnnel coding, improve spectrum utilization efficiency by this method.At receiving terminal, the just demodulation that need provide the 16QAM demodulator to realize signal accordingly, the demodulator of traditional 16QAM is general direct to carry out hard decision according to its planisphere quantification output, obtains result 0 or 1.But because receiver carries out independently hard decision to signal, make received signal produce expendable information loss, use the channel decoding of soft-decision in therefore now a lot of communication systems.Under the normal condition, the Viterbi of soft-decision (Viterbi) is than the good 2-3dB of Viterbi decoding performance of hard decision.

Turbo code concern on boundary that obtains communicating by letter because of its remarkable performance, and be used as chnnel coding and extensive use in a lot of communication systems.And its decoder is a kind of typical soft inputting and soft way of output, therefore just needs demodulator to export soft information and imports as it.Propose in the United States Patent (USP) (US 6,594,318 B1) to calculate the soft information of decoding, come operate approximately LLR value with following formula by the method for calculating LLR (Log-Likelihood Ratio, log-likelihood ratio),

$\mathrm{LLR}\left(c_0\right)=1.2649\left|Y_I\right|-0.8\left(\frac{C}{I}\right)$

LLR(c ₁)＝1.2649Y _I

$\mathrm{LLR}\left(c_2\right)=1.2649\left|Y_Q\right|-0.8\left(\frac{C}{I}\right)$

LLR(c ₃)＝1.2649Y _Q

Wherein

Be the carrier wave signal interference ratio, this value obtains by measurement pilot signals, Y _I, Y _QBe respectively the real part of the modulation signal that receives and imaginary part according to

Carry out the result after the normalization, c ₀..., c ₃Be 4 input bits corresponding to a modulation signal.The subject matter of this method is too simple to being similar to of LLR formula, can cause the soft information of demodulation output bigger with actual LLR value error ratio, and this can have a strong impact on the channel-decoding performance after the demodulation.Be to use the parameter signal interference ratio on the other hand

Participate in calculating soft information,, increased the influence of noise soft information owing to introduce the Interference Estimation value.The present invention is exactly under the situation that does not improve the receiver complexity, proposes a kind of demodulation method that can more accurately export soft information, thereby can improve the performance of back channel-decoding.

Summary of the invention

The objective of the invention is LLR, and carry out under the necessarily approximate situation,, improve the precision that the 16QAM demodulator is exported soft information, thereby improve the overall performance of Rake receiver as the soft information of restituted signal by data after the calculating demodulation.

To achieve these goals, the technical solution adopted in the present invention is: the LLR value of utilizing each bit of the real part of LLR modulation signal that function calculation receives and imaginary part, with the soft information of result of calculation as demodulator output, it is characterized in that asking the process of LLR value may further comprise the steps: with the LLR function at interval [∞,-2a], [2a, 2a] and [2a, + ∞] on carry out segmentation, wherein a is the range value of modulation signal; Each section is carried out match with function of first order, describedly carry out match and comprise with the range value a of modulation signal the real part and the imaginary part of the modulation signal that received are carried out normalized, make that the modulation signal unit amplitude of real part and imaginary part equals 1 after the normalization; And to the result after the normalization carry out real part and imaginary component from, obtain real part X _kWith imaginary part Y _k

Calculate the log-likelihood ratio function LLR of each bit after the normalized, this log-likelihood ratio function is as follows:

$\mathrm{LLR}\left(s_{k,0}\right)=\left\{\begin{array}{cc}b{X}_k+c& X_k\∈[-\∞,-2]\\ {\mathrm{dX}}_k& X_k\∈[-\mathrm{2,2}]\\ b{X}_k-c& X_k\∈[2,+\∞]\end{array}\right.,k=0,...,K-1$

LLR(s _k，1)＝e+fabs(X _k) k＝0，...，K-1

$\mathrm{LLR}\left(s_{k,2}\right)=\left\{\begin{array}{cc}b{Y}_k+c& Y_k\∈[-\∞,-2]\\ {\mathrm{dY}}_k& Y_k\∈[-\mathrm{2,2}]\\ b{Y}_k-c& Y_k\∈[2,+\∞]\end{array}\right.,k=0,...,K-1$

LLR(s _k，3)＝e+fabs(Y _k) k＝0，...，K-1

Wherein, s _{K, 0}s _{K, 1}, s _{K, 2}, s _{K, 3} Be 4 input bits corresponding to k modulation signal.Described b=7.876, c=7.488, d=4.17, e=-8 and f=4.1.

The range value of described modulation signal obtains by following steps: utilize the pilot signal that receives, calculate the intensity of pilot signal, draw the amplitude size of data-signal again according to the energy relationship of pilot signal and data-signal.

Another aspect of the present invention is to provide the demodulating equipment of 16 grades of quadrature amplitude modulation in a kind of digital communication system, utilize the log-likelihood ratio of each bit of the real part of LLR modulation signal that function calculation receives and imaginary part, with the soft information of result of calculation, it is characterized in that asking log-likelihood ratio in order to lower device as demodulator output:

The normalization module, it uses the range value a of modulation signal that the modulation signal that is received is carried out normalized, makes that the modulation unit amplitude of modulation signal real part and imaginary part equals 1 after the normalization;

Imaginary part real part separation module carries out the separation of real part imaginary part to the modulation signal after the normalization, obtains real part X _kWith imaginary part Y _k

The function of first order fitting module uses function of first order at interval [∞ ,-2], [2,2] and [2 ,+∞] match log-likelihood ratio function.

Described function of first order fitting module comprises absolute calculators, adder highest order extractor, two-way selector, multiplier, and these included elements of this function of first order fitting module connect with the logical operation relation, thereby is achieved as follows the computing of formula:

$\mathrm{LLR}\left(s_{k,0}\right)=\left\{\begin{array}{cc}b{X}_k+c& X_k\∈[-\∞,-2]\\ {\mathrm{dX}}_k& X_k\∈[-\mathrm{2,2}]\\ b{X}_k-c& X_k\∈[2,+\∞]\end{array}\right.,k=0,...,K-1$

LLR(s _k，1)＝e+fabs(X _k) k＝0，...，K-1

$\mathrm{LLR}\left(s_{k,2}\right)=\left\{\begin{array}{cc}b{Y}_k+c& Y_k\∈[-\∞,-2]\\ {\mathrm{dY}}_k& Y_k\∈[-\mathrm{2,2}]\\ b{Y}_k-c& Y_k\∈[2,+\∞]\end{array}\right.,k=0,...,K-1$

Description of drawings

LLR(s _k，3)＝e+fabs(Y _k) k＝0，...，K-1

Wherein, s _{K, 0}s _{K, 1}, s _{K, 2}, s _{K, 3} Be 4 input bits corresponding to k modulation signal.

Described b=7.876, c=7.488, d=4.17, e=-8 and f=4.1.Described normalization module is a multiplier;

Owing to use approximate formula to substitute the LLR formula of original complexity in this programme, greatly reduce the complexity of demodulation method, can in real system, use, mentioning United States Patent (USP) with the front compares, substantially do not increasing under the prerequisite of demodulation method complexity, improve the precision of LLR calculated value greatly, receiver performance improves.By the normalization operation, the coefficient that calculates the LLR formula is fixed simultaneously, exempted the complex calculation of design factor, improve the flexibility of receiver.

Fig. 1 modulation signal unit amplitude a is the planisphere of 1 o'clock 16QAM;

Fig. 2 modulation signal unit amplitude a is 1 o'clock s _{K, 0}The LLR function curve diagram;

Fig. 3 modulation signal unit amplitude a is 1 o'clock s _{K, 1}The LLR function curve diagram;

Fig. 4 modulation signal unit amplitude a is 1 o'clock $\mathrm{<figure-callout\; id="1"\; label="Log"\; filenames="H04166014X20050121E000022.png,H04166014X20050121E000032.png"\; state="\{\{state\}\}">Log</figure-callout>}\frac{1+e^{4a{X}_k-8a^2}}{1+e^{-4a{X}_k-8a^2}}$ Function curve diagram;

Embodiment

Fig. 5 flow chart of the present invention;

The functional block diagram of Figure 61 6QAM demodulation part;

The hardware block diagram of Figure 71 6QAM demodulation part;

The demodulation method performance comparison diagram of Fig. 8 the present invention and prior art.

Below in conjunction with the drawings and specific embodiments technical scheme of the present invention is described in further detail, according to these structure charts, the technical staff in same field can be easy to realize the present invention.

In wireless communication system, often use the method for 16QAM modulation to improve the availability of frequency spectrum, its planisphere is referring to Fig. 1.4 bit signal s in the former data _{K, 0}s _{K, 1}, s _{K, 2}, s _{K, 3}Be divided into two groups, real part and imaginary part among the difference constellation figure, the mapping mode among Fig. 1 is s _{K, 0}s _{K, 1}Corresponding real part, s _{K, 2}, s _{K, 3}Corresponding imaginary part, its corresponded manner can be formulated as:

X _k＝(2s _k，0-1)*2a+(2s _k，1-1)*a

Y _k＝(2s _k，2-1)*2a+(2s _k，3-1)*a

X wherein _kAnd Y _kReal part and the imaginary part of representing the modulation signal of reception respectively, s _{K, 0}s _{K, 1}, s _{K, 2}, s _{K, 3} Be 4 input bits corresponding to k modulation signal, a is the amplitude units in the modulation.Its mapping mode is referring to Fig. 2.

Turbo code concern on boundary that obtains communicating by letter because of its remarkable performance, and be used as chnnel coding and extensive use in a lot of communication systems.And its decoder is a kind of typical soft inputting and soft way of output, therefore just needs demodulator to export soft information and imports as it.Because general Turbo coder all is to adopt binary mode now, so this just requires the 16QAM demodulator to export the soft information of each bit.Here just relate to the soft information that how to obtain each bit, optimum method is exactly to calculate the LLR value of each bit.

$\mathrm{LLR}\left(s_{k,i}\right)=\mathrm{Log}\frac{\mathrm{\&Sigma;p}(s_{k,i}=1|X_k,Y_k)}{\mathrm{\&Sigma;p}(s_{k,i}=0|X_k,Y_k)},i=0,...,3,k=0,...,K-1$

S wherein _{K, i}Be i the bit signal that k 16QAM receives the complex signal correspondence, X _kBe the k complex signal real part that receives, Y _kIt is the k complex signal imaginary part that receives.Because the signal of respectively corresponding two bits of real part and imaginary part, separate, following formula just can become:

$\mathrm{LLR}\left(s_{k,i}\right)=\mathrm{Log}\frac{\mathrm{\&Sigma;p}(s_{k,i}=1|X_k)}{\mathrm{\&Sigma;p}(s_{k,i}=0|X_k)},i=0,1,k=0,...,K-1$

$\mathrm{LLR}\left(s_{k,i}\right)=\mathrm{Log}\frac{\mathrm{\&Sigma;p}(s_{k,i}=1|Y_k)}{\mathrm{\&Sigma;p}(s_{k,i}=0|Y_k)},i=2,3,k=0,...,K-1$

Since the signal of respectively corresponding two bits of real part and imaginary part, and corresponded manner is identical, and can only discuss the LLR formula of real part here.Because the supposition noise is an additive Gaussian noise, brings probability-distribution function into following formula and just can obtain:

$\mathrm{LLR}\left(s_{k,0}\right)=\mathrm{Log}\frac{e^{-{(X_k-3a)}^2}+e^{-{(X_k-a)}^2}}{e^{-{(X_k+3a)}^2+}{e}^{-{(X_k+a)}^2}},k=0,...,K-1$

$\mathrm{LLR}\left(s_{k,1}\right)=\mathrm{Log}\frac{e^{-{(X_k-a)}^2}+e^{-{(X_k+a)}^2}}{e^{-{(X_k-3a)}^2+}{e}^{-{(X_k+3a)}^2}},k=0,...,K-1$

Fig. 2 and Fig. 3 are respectively s _{K, 0}And s _{K, 1}The LLR function curve.Because have exponent arithmetic in the formula, complexity is very big, can not use in the reality.Through the derivation of formula, top formula can be simplified as follows:

$\mathrm{LLR}\left(s_{k,0}\right)=4a{X}_k+\mathrm{<figure-callout\; id="1"\; label="Log"\; filenames="H04166014X20050121E000022.png,H04166014X20050121E000032.png"\; state="\{\{state\}\}">Log</figure-callout>}\frac{1+e^{4a{X}_k-8a^2}}{1+e^{-4a{X}_k-8a^2}},k=0,...,K-1$

$\mathrm{LLR}\left(s_{k,1}\right)=-8a^2+\mathrm{Log}\frac{e^{8a{X}_k}+e^{-4a{\mathrm{<figure-callout\; id="1"\; label="X\; k"\; filenames="H04166014X20050121E000022.png,H04166014X20050121E000032.png"\; state="\{\{state\}\}">X</figure-callout>}}_k}}{1+e^{4a{X}_k}},k=0,...,K-1$

Above two formula still have exponent arithmetic, but we can be similar to.For s _{K, 0}The LLR function, one is complicated exponent arithmetic thereafter, its function curve is referring to Fig. 4.This function changes very for a short time in that [2a, 2a] is interval basic as can be seen, and all is similar to clothes+from function of first order in rest position.In order to simplify computing, we are to s _{K, 0}The LLR function carry out segmentation, [∞ ,-2a], [2a, 2a] and [2a ,+∞], to each the section carry out match with function of first order.And as can be seen from Figure 3 to s _{K, 1}The LLR function also can carry out good match with function of first order.

The range value a of modulation signal obtains by following steps: utilize the pilot signal that receives to calculate the intensity of pilot signal, draw the amplitude size of data-signal again according to the energy relationship of pilot signal and data-signal.

In order to reduce the problem that to calculate fitting function in the actual realization to different modulation signal unit amplitude a values respectively.We carry out normalization with a to the real part and the imaginary part that receive modulation signal, make that the modulation signal unit amplitude of real part and imaginary part equals 1 after the normalization, so just do not need all to have calculated the coefficient of fitting function at every turn.Through the over-fitting computing, the formula that can obtain the LLR function is as follows:

$\mathrm{LLR}\left(s_{k,0}\right)=\left\{\begin{array}{cc}7.876X_k+7.488& X_k\&Element;[-\&infin;,-2]\\ {4.17X}_k& X_k\&Element;[-\mathrm{2,2}]\\ 7.876X_k-7.488& X_k\&Element;[2,+\&infin;]\end{array}\right.,k=0,...,K-1$

LLR(s _k，1)＝-8+4.1abs(X _k) k＝0，...，K-1

Fig. 5 is the demodulation flow chart after obtaining modulation signal unit amplitude a, and basic process is as follows:

1. the modulation signal that receives is carried out normalization with a.

2. the result after the normalization is carried out the separation of real part imaginary part, obtain X _k, Y _k

3. calculate s _{K, 1}And s _{K, 3}The LLR value:

LLR(s _k，1)＝-8+4.1abs(X _k)

LLR(s _k，3)＝-8+4.1abs(Y _k)

4. at first judge real part X _kWhether greater than-2, if less than, LLR (s _{K, 0})=7.876X _k+ 7.488, change step 6 over to, otherwise change step 5 over to.

5. judge X _kWhether less than 2, if less than, LLR (s _{K, 0})=4.17X _k, otherwise LLR (s _{K, 0})=7.876X _k-7.488.

6. at first judge imaginary part Y _kWhether greater than-2, if less than, LLR (s _{K, 2})=7.876Y _k+ 7.488, change step 8 over to, otherwise change step 7 over to.

7. judge Y _kWhether less than 2, if less than, LLR (s _{K, 2})=4.17Y _k, otherwise LLR (s _{K, 2})=7.876Y _k-7.488.

8.LLR calculating, value finishes.

Fig. 6 is the module frame chart of 16QAM demodulation part, the modulation complex signal Z that input receives _kAnd 1/a, by modulation amplitude normalization module 601, imaginary part real part separation module 603 and function of first order fitting module 602,604 just can obtain soft information output LLR (S _{K, 0}), LLR (S _{K, 1}), LLR (S _{K, 2}) and LLR (S _{K, 3}).Fig. 7 is the hardware capability block diagram of 16QAM demodulation part, and in the present embodiment, above-mentioned normalization module 601 is a multiplier 701, by multiplier 701, can realize the normalization operation to the modulation complex signal.By imaginary part real part separator 714, realize lock out operation to modulation complex signal real part and imaginary part.Because the processing procedure of real part and imaginary part is similar, only describe here by real part X _kObtain LLR (S _{K, 0}) and LLR (S _{K, 1}) process.

X _kAt first be sent in the absolute calculators 704 and obtain | X _k|, then it with-2 send in the adder 705 and obtain | X _k|-2, be sent to again in the highest order extractor 706, obtain | X _k|-2 positive and negative, represented in 0 o'clock on the occasion of, 1 is negative value, as the control input of two-way selector 708.

X _kBe sent in the multiplier 702 and obtain 4.17X _k, as one tunnel input of two-way selector 708.

X _kBe sent in the multiplier 703 and obtain 7.876X _k, as one tunnel input of adder 707.

X _kBe sent in the highest order extractor 710 and obtain X _kPositive and negative, represented in 0 o'clock on the occasion of, 1 is negative value.Positive negative value is sent to two-way selector 709 and selects-7.488 or 7.488, is sent to again in the adder 707, and 7.876X _kAddition is as one tunnel input of two-way selector 708.

Thereby just realized following formula:

$\mathrm{LLR}\left(s_{k,0}\right)=\left\{\begin{array}{cc}7.876X_k+7.488& X_k\&Element;[-\&infin;,-2]\\ {4.17X}_k& X_k\&Element;[-\mathrm{2,2}]\\ 7.876X_k-7.488& X_k\&Element;[2,+\&infin;]\end{array}\right.,k=0,...,K-1$

X _kAt first be sent in the absolute calculators 713 and obtain | X _k|, then it with 4.1 send in the multiplier 711 and obtain 4.1X _k, be sent to again in the adder 712, obtain 4.1X _kThereby-8 have realized following formula:

LLR(s _k，1)＝-8+4.1abs(X _k) k＝0，...，K-1

Fig. 8 is the demodulation method performance comparison diagram of the present invention and prior art, and as chnnel coding, data length is 630 to use 16QAM as the Turbo code of modulation system and three times of code rates in the emulation.Article two, performance curve is represented method and the method for the present invention that prior art adopts respectively, and as can be seen from the figure, new method can realize that the performance of 0.1dB improves.

The use approximate formula has substituted the LLR formula of original complexity in this programme, can reduce the complexity of method greatly, can use in real system.By the normalization operation, the coefficient that calculates the LLR formula is fixed simultaneously, exempted the complex calculation of design factor, improve the flexibility of receiver.

Claims (5)

1. the demodulation method of 16 grades of quadrature amplitude modulation in the digital communication system, utilize the log-likelihood ratio of each bit of the real part of log-likelihood ratio modulation signal that function calculation receives and imaginary part, with the soft information of result of calculation, it is characterized in that asking the process of log-likelihood ratio may further comprise the steps as demodulator output:

The log-likelihood ratio function at interval [∞ ,-2a], is carried out segmentation on [2a, 2a] and [2a ,+∞], and wherein a is the range value of modulation signal;

Each section is carried out match with function of first order, describedly carry out match and comprise with the range value a of modulation signal the real part and the imaginary part of the modulation signal that received are carried out normalized, make that the modulation signal unit amplitude of real part and imaginary part equals 1 after the normalization; And to the result after the normalization carry out real part and imaginary component from, obtain real part X _kWith imaginary part Y _k

Calculate the log-likelihood ratio function LLR of each bit after the normalized, this log-likelihood ratio function is as follows:

$\mathrm{LLR}\left(s_{k,0}\right)=\left\{\begin{array}{cc}{\mathrm{bX}}_k+c& X_k\&Element;[-\&infin;,-2]\\ d{X}_k& X_k\&Element;[-\mathrm{2,2}]\\ {\mathrm{bX}}_k-c& X_k\&Element;[2,+\&infin;]\end{array}\right.k=0,...,K-1$

LLR(s _k，1)＝e+fabs(X _k) k＝0，...，K-1

$\mathrm{LLR}\left(s_{k,2}\right)=\left\{\begin{array}{cc}{\mathrm{bY}}_k+c& Y_k\&Element;[-\&infin;,-2]\\ d{Y}_k& Y_k\&Element;[-\mathrm{2,2}]\\ {\mathrm{bY}}_k-c& Y_k\&Element;[2,+\&infin;]\end{array}\right.k=0,...,K-1$

LLR(s _k，3)＝e+fabs(Y _k) k＝0，...，K-1

Wherein, s _{K, 0}s _{K, 1}, s _{K, 2}, s _{K, 3}Be 4 input bits corresponding to k modulation signal, wherein, described b=7.876, c=7.488, d=4.17, e=-8 and f=4.1.

2. demodulation method according to claim 1, its feature are that further the range value of modulation signal obtains by following steps:

The pilot signal that utilization receives is calculated the intensity of pilot signal, draws the amplitude size of data-signal again according to the energy relationship of pilot signal and data-signal.

3. demodulation method according to claim 1 is characterized in that calculating s _{K, 0}s _{K, 1}, s _{K, 2}, s _{K, 3}The step of the LLR value of each bit comprises:

(1) calculates s _{K, 1}And s _{K, 3}The LLR value:

LLR(s _k，1)＝-8+4.1abs(X _k)

LLR(s _k，3)＝-8+4.1abs(Y _k)

(2) at first judge real part X _kWhether greater than-2, if less than, LLR (s _{K, 0})=7.876X _k+ 7.488, change step (4) over to, otherwise change step (3) over to;

(3) judge X _kWhether less than 2, if less than, LLR (s _{K, 0})=4.17X _k, otherwise LLR (s _{K, 0})=7.876X _k-7.488;

(4) at first judge imaginary part Y _kWhether greater than-2, if less than, LLR (s _{K, 2})=7.876Y _k+ 7.488, change step (6) over to, otherwise change step (5) over to;

(5) judge Y _kWhether less than 2, if less than, LLR (s _{K, 2})=4.17Y _k, otherwise LLR (s _{K, 2})=7.876Y _k-7.488;

(6) log-likelihood ratio calculates and finishes.

4. the demodulating equipment of 16 grades of quadrature amplitude modulation in the digital communication system, utilize the log-likelihood ratio of each bit of the real part of log-likelihood ratio modulation signal that function calculation receives and imaginary part, with the soft information of result of calculation, it is characterized in that asking log-likelihood ratio with following described device as demodulator output:

The normalization module, it uses the range value a of modulation signal that the modulation signal that is received is carried out normalized, makes that the modulation unit amplitude of modulation signal real part and imaginary part equals 1 after the normalization, and obtains real part X _kWith imaginary part Y _k

Imaginary part real part separation module, with the modulation signal after the normalization carry out real part and imaginary component from;

The function of first order fitting module uses function of first order at interval [∞ ,-2], [2,2] and [2 ,+∞] match log-likelihood ratio function; Described function of first order fitting module comprises absolute calculators, adder, highest order extractor, two-way selector and multiplier; These included elements of this function of first order fitting module connect with the logical operation relation, thereby are achieved as follows the computing of formula:

$\mathrm{LLR}\left(s_{k,0}\right)=\left\{\begin{array}{cc}{\mathrm{bX}}_k+c& X_k\&Element;[-\&infin;,-2]\\ d{X}_k& X_k\&Element;[-\mathrm{2,2}]\\ {\mathrm{bX}}_k-c& X_k\&Element;[2,+\&infin;]\end{array}\right.k=0,...,K-1$

LLR(s _k，1)＝e+fabs(X _k) k＝0，...，K-1

$\mathrm{LLR}\left(s_{k,2}\right)=\left\{\begin{array}{cc}{\mathrm{bY}}_k+c& Y_k\&Element;[-\&infin;,-2]\\ d{Y}_k& Y_k\&Element;[-\mathrm{2,2}]\\ {\mathrm{bY}}_k-c& Y_k\&Element;[2,+\&infin;]\end{array}\right.k=0,...,K-1$

LLR(s _k，3)＝e+fabs(Y _k) k＝0，...，K-1

Wherein, s _{K, 0}s _{K, 1}, s _{K, 2}, s _{K, 3}Be 4 input bits corresponding to k modulation signal, wherein, described b=7.876, c=7.488, d=4.17, e=-8 and f=4.1.

5. demodulating equipment according to claim 5, wherein said normalization module is a multiplier.

Images