CN101404642A - Soft demodulation method for 32 ary quadrature amplitude modulation - Google Patents

Abstract

The invention provides a soft demodulation method of 32QAM modulation in a communication system. The realization method comprises the following steps: finding out a hard-decision line in each block through simple complex plane blocking, and taking the distance from a received signal to the hard-decision line as the dimension of corresponding bit soft information, the symbol of which is determined by the position of the received signal relative to the hard-decision line.

Description

A kind of soft demodulating method of 32 ary quadrature amplitude

Technical field

The present invention relates to the communications field, the soft demodulating method of particularly a kind of 32 ary quadrature amplitude (hereinafter referred to as 32QAM).

Background technology

Quadrature amplitude modulation (QAM, Quadrature Amplitude Modulation) is a kind of two-dimensional modulation technology, removes to modulate the homophase carrier wave with the projection of constellation point on real axis, with the projection modulation orthogonal carrier wave on the imaginary axis.32QAM is a kind of efficient quadrature amplitude modulation mode, and each symbol can transmit 5 bits.

For the information that makes full use of received signal is carried out channel-decoding, demodulator should provide and can reflect the soft information that receives bit value probability.Generally (LLR Log-LikelihoodRatio) represents soft information, is defined as and receives bit to be 0 posterior probability and to be the logarithm of the ratio of 1 posterior probability with the log-likelihood ratio that receives bit.

$\mathrm{LLR}\left(b\right)=\ln \left(\frac{\mathrm{Pr}(b=0|r)}{\mathrm{Pr}(b=1|r)}\right)$

Wherein r is a received signal, uses complex representation.B is certain bit in the received signal.

Under the white Gaussian noise channel, suppose that the probability that all constellation point occur equates that then likelihood ratio LLR (b) can be expressed as

$\mathrm{LLR}\left(b\right)=\ln \left(\frac{\underset{s\&Element;{{\mathrm{<figure-callout\; id="0"\; label="S\; b"\; filenames="A20081023469900092.png,A20081023469900101.png"\; state="\{\{state\}\}">S</figure-callout>}}_b}^0}{\mathrm{\&Sigma;}}\exp (-\frac{1}{{\mathrm{\&sigma;}}^2}{|r-s|}^2)}{\underset{s\&Element;{{\mathrm{<figure-callout\; id="1"\; label="S\; b"\; filenames="A20081023469900092.png,A20081023469900101.png"\; state="\{\{state\}\}">S</figure-callout>}}_b}^1}{\mathrm{\&Sigma;}}\exp (-\frac{1}{{\mathrm{\&sigma;}}^2}{|r-s|}^2)}\right)$

S wherein _b ⁰Expression b is the constellation point set of 0 correspondence, S _b ¹Expression b is the constellation point set of 1 correspondence, σ ²Power spectral density for white Gaussian noise.

Directly relate to a large amount of exponent arithmetics according to this formula calculating, operand is excessive.Approximate computational methods are only got maximum that in following formula molecule denominator and the formula.Then the computational methods of LLR (b) can be reduced to

$\mathrm{LLR}\left(b\right)=\frac{1}{{\mathrm{\&sigma;}}^2}(\underset{s\&Element;{S_b}^0}{\min }\left({|r-s|}^2\right)-\underset{s\&Element;{S_b}^1}{\min }\left({|r-s|}^2\right))$

Can see, in the formula after the simplification received signal be arrived S _b ⁰And S _b ¹In recently the Euclidean distance difference of constellation point as the tolerance of the soft information of its corresponding bit.

Though approximation method has been saved exponent arithmetic, but still need to calculate and the Euclidean distance of all constellation point and get wherein minimum value.In the square constellation of standards such as the 4QAM, the 16QAM that press the Gray code mapping and 64QAM, ask for the method for LLR (b) can further simplify, received signal can be arrived the tolerance of the distance of hard decision line as soft information.Because the hard decision line by the square constellation of Gray code mapping all is the straight line that is parallel to the real axis and the imaginary axis, so only the real part by to received signal and imaginary part are carried out plus and minus calculation and can be extracted soft information.But the hard decision line for the non-standard square constellation as 32QAM is comparatively complicated, can not simply extract.

Summary of the invention

The present invention utilizes received signal to arrive the distance of hard decision line, extracts bit soft information in conjunction with the area dividing of constellation, finishes the soft demodulation to the 32QAM signal.

Constellation mapping mode of the present invention as shown in Figure 1.Corresponding 5 bit: the b of each constellation point ₀, b ₁, b ₂, b ₃, b ₄With i bit b _iThe constellation point set that is 0 correspondence is designated as S _i ⁰, be that the constellation point set of 1 correspondence is designated as S _i ¹, apart from S _i ⁰In constellation point nearer the zone and apart from S _i ¹The line of demarcation in the zone that middle constellation point is nearer is called b _iCorresponding hard decision line.b _iCorresponding soft information can obtain by the distance of calculating received signal to the hard decision line.

b ₀, b ₁, b ₂Corresponding hard decision line is comparatively complicated, but its corresponding S _i ⁰And S _i ¹Distribution about the complete symmetry of reference axis, therefore can be by asking for received signal real part absolute value and the imaginary part absolute value is folded to first quartile with received signal, b then ₀, b ₁, b ₂Therefore only the problem of asking for soft information in four quadrants just is converted to the problem of asking for soft information in first quartile, need be in first quartile coordinate plane be cut apart to get final product.

Investigate b ₀Corresponding S ₀ ⁰And S ₀ ¹Situation about distributing in first quartile as shown in Figure 2.Under the situation of taking slightly to be similar to, according to solid line among the figure first quartile is divided into 3 zones, and finds out the hard decision line (as shown in phantom in FIG.) between 0 and 1 in each zone.The leaching process of soft information is as follows:

The first step is at first judged the zone at received signal place;

Second step, with the distance of received signal hard decision line in this zone as b ₀The size of soft information;

In the 3rd step, the symbol of soft information is determined with respect to the position of hard decision line by received signal.When received signal was positioned at 0 side of hard decision line, soft information had been for just, and soft information is for negative when being positioned at 1 side of hard decision line.

b ₁And b ₂The leaching process of soft information and b ₀Similar, dividing region and hard decision line are respectively as shown in Figure 3 and Figure 4.Solid line is represented the zone boundary among the figure, and dotted line is represented the hard decision line.

b ₃Corresponding S ₃ ⁰And S ₃ ¹Lay respectively at the imaginary axis about, so the imaginary axis is the hard decision line, can be with the real part of received signal as b ₃The soft value of information.Same b ₄Corresponding S ₃ ⁰And S ₃ ¹Lay respectively at real axis about, therefore can be with the imaginary part of received signal as b ₄The soft value of information.

Be convenient soft information extraction, the present invention is similar to slightly to the definite of dividing region and hard decision line, to reduce amount of calculation, performance of BER such as Fig. 5.

In addition, the inventive method also is applicable to following constellation: (1) b ₀, b ₁, b ₂, b ₃, b ₄The difference corresponding constellation that puts in order; (2) 0 and 1 exchanges the pairing constellation of (or negate) back mapping; (3) constellation is through rotating the constellation that obtains after certain angle.

Description of drawings

Figure 132 QAM mapped constellation figure;

Fig. 2 b ₀Corresponding quadrant partitioned mode (representing) and every hard decision line (dotting) with solid line;

Fig. 3 b ₁Corresponding quadrant partitioned mode and every hard decision line;

Fig. 4 b ₂Corresponding quadrant partitioned mode and every hard decision line;

Analogous diagram under Fig. 5 white Gaussian noise channel, chnnel coding adopt code check be 1/2 (2,1,2) convolution code, the ber curve that the minimum range for utilizing received signal and standard constellation point that indicates with * among the figure asks the method for posterior probability to obtain, with o indicate for utilizing method of the present invention to obtain ber curve.

Specific implementation method

The implementation method of this patent is described below by instantiation.

Transmitting terminal is bit stream groupings, every group 5 bit.If 5 bits of certain group are b ₀, b ₁, b ₂, b ₃, b ₄, mapping mode as shown in Figure 1.The size of a reflection constellation among the figure, for the constellation of power normalization, a gets

Suppose that the signal that receiving terminal receives is r (r is plural number), make x=abs (real (r)), y=abs (imag (r)), wherein abs () expression takes absolute value, and real part is got in real () expression, and imaginary part is got in imag () expression.b ₀, b ₁, b ₂, b ₃, b ₄Corresponding soft information extracting method is as follows:

b ₀The extraction of soft information:

if(x＞y)or(y＜3a)，llr(b0)＝y-2a

else{

if(x+y)＜6a，llr(b0)＝4a-y

else，llr(b0)＝x-2a

}

b ₁The extraction of soft information:

if(x+y)＜6a，llr(b1)＝x-2a

else{

if(x＜4a)，llr(b1)＝4a-y

$\mathrm{elsellr}\left(b1\right)=(x-y)/\sqrt{2}$

}

b ₂The extraction of soft information:

if(x＜y)，llr(b2)＝y-4a

else?llr(b2)＝x-4a

b ₃The extraction of soft information:

llr(b3)＝real(r)

b ₄The extraction of soft information:

llr(b4)＝imag(r)。

Claims (2)

1, a kind of soft demodulating method that is applied to 32QAM, it is characterized in that complex plane being carried out area dividing according to different bits, determine the hard decision line in each zone, soft demodulator is judged the residing zone of received signal earlier, with the distance of received signal hard decision line in this zone size as soft information, the symbol of soft information is determined with respect to the position of hard decision line by received signal.

2, soft demodulating method according to claim 1 is characterized in that extracting b ₀, b ₁, b ₂, b ₃, b ₄The corresponding soft information llr (b of these 5 bits _i) method, suppose that received signal is r, make x=abs (real (r)), y=abs (imag (r)), the concrete grammar of demodulation is as follows:

(1), for b ₀Soft information, get the absolute value of received signal real part and imaginary part earlier, then constellation division methods shown in 2 and hard decision line with reference to the accompanying drawings, judge its zone of living in and obtain its distance to the hard decision line, with the tolerance of this distance value as soft information, implementation method is as follows

if(x＞y)or(y＜3a)，llr(b0)＝y-2a

else {

if(x+y)＜6a，llr(b0)＝4a-y

else，llr(b0)＝x-2a

}

(2), for b ₁Soft information, get the absolute value of received signal real part and imaginary part earlier, then constellation division methods shown in 3 and hard decision line with reference to the accompanying drawings, judge its zone of living in and obtain its distance to the hard decision line, with the tolerance of this distance value as soft information, implementation method is as follows

if(x+y)＜6a，llr(b1)＝x-2a

else {

if(x＜4a)，llr(b1)＝4a-y

$\mathrm{else\; llr}\left(b1\right)=(x-y)/\sqrt{2}$

}

(3), for b ₂Soft information, get the absolute value of received signal real part and imaginary part earlier, then constellation division methods shown in 4 and hard decision line with reference to the accompanying drawings, judge its zone of living in and obtain its distance to the hard decision line, with the tolerance of this distance value as soft information, implementation method is as follows

if(x＜y)，llr(b2)＝y-4a

else?llr(b2)＝x-4a

(4), with the real part of received signal and imaginary part respectively as b ₃And b ₄The soft value of information.

llr(b3)＝real(r)

llr(b4)＝real(r)。

Images