CN101437012A - Soft demodulation method with low complexity for Gray quadrature amplitude modulation - Google Patents

Abstract

The invention discloses a soft demodulation method for low complexity under Gray quadrature amplitude modulation. The method directly obtains a constellation point needed by calculating the bit likelihood ratio through a hard decision symbol by utilizing the discipline between the hard decision symbol and all constellation points needed by calculating the bit likelihood ratio according to a mirror mapping principle of Gray codes, which avoids the determination of the constellation point by a least Euclidean distance search method in the prior method, so the method avoids a complex search process and greatly reduces calculation complexity; and the cost is reduced in the hardware realization, thus the soft demodulation method under Gray M-QAM can be widely applied in high-speed mobile communication systems.

Description

A kind of soft demodulating method of low complexity for Gray quadrature amplitude modulation

Technical field

The invention belongs to wireless communication field, it is particularly related to the soft demodulating method under the Gray M-QAM modulation.

Background technology

In following radio communication, frequency spectrum resource is more and more valuable, improves the main target that the availability of frequency spectrum becomes wireless communication system.Therefore, high order modulation technology M-QAM (quadrature amplitude modulation of multi-system) is owing to its high availability of frequency spectrum is widely used.Gray mappings can improve the bit error rate performance of system, also is applied in the modern wireless communication systems.In having the communication system of forward error correction coding, the soft-decision algorithm improves significantly than the performance of declaring firmly, see document: F.Tosato and P.Bisaglia, " Simplifed soft-outputdemapper for binary interleaved COFDM with application to HIPERLAN/2; " in Proc.IEEE ICC ' 02, vol.2, pp.664-668,2002.Therefore, the Gray's high order modulation of making a start, the receiving end method of carrying out soft decision demodulation (soft demodulation) is widely used in the modern wireless communication systems.

Yet soft demodulating algorithm need calculate the bit likelihood ratio information of each received signal, and therefore, its computation complexity is very high.Here, introduce the computational methods of bit likelihood ratio in traditional soft demodulating algorithm earlier, i.e. bit likelihood ratio computing module 7 among Fig. 1.

Make a start and adopt gray mappings M-QAM to modulate (3GPP standard recommendation), then the real part of received signal and imaginary part satisfy orthogonality, and all are

Therefore modulation carries out hard decision to the real part x ' in the signal after the equilibrium, obtains constellation point $s_x^{*}=\{b_{\left(x\right)0}^{*}{b}_{\left(x\right)1}^{*}\·\·\·b_{\left(x\right)k-1}^{*}\};$ Y ' declares firmly to imaginary part, obtains constellation point $s_y^{*}=\{b_{\left(y\right)0}^{*}{b}_{\left(y\right)1}^{*}\·\·\·b_{\left(y\right)k-1}^{*}\},$ A kind of hard decision shortcut calculation is seen document: and " C.K, Yuen, A fast analog togray code converter, " in Proceedings of the IEEE, vol.65, pp.1510-1511, Oct.1977.

To real part, imaginary part is carried out the calculating of bit likelihood ratio respectively:

$\mathrm{LLR}\left(b_{\left(x\right)i}\right)=\frac{1}{{\mathrm{\σ}}^2}({|x\′-s_{\left(x\right)i}^{-}|}^2-{|x\′-s_{\left(x\right)i}^{+}|}^2)---\left(1\right)$

$\mathrm{LLR}\left(b_{\left(y\right)i}\right)=\frac{1}{{\mathrm{\σ}}^2}({|y\′-s_{\left(y\right)i}^{-}|}^2-{|y\′-s_{\left(y\right)i}^{+}|}^2)---\left(2\right)$

Wherein, LLR (b _{(x) i}) expression real part the bit likelihood ratio, LLR (b _{(y) i}) expression imaginary part the bit likelihood ratio, i represents bit sequence,

Represent the i bit be 0 and with the constellation point of imaginary part x ' Euclidean distance minimum,

Represent the i bit be 1 and with the constellation point of imaginary part y ' Euclidean distance minimum; σ ²Be the power of multiple Gaussian noise; According to declaring the result firmly, can obtain to draw a conclusion:

If $b_{\left(x\right)i}^{*}=0,$ Then $s_{\left(x\right)i}^{-}=s_x^{*};$ If $b_{\left(x\right)i}^{*}=1,$ Then $s_{\left(x\right)i}^{+}=s_x^{*};$

If $b_{\left(y\right)i}^{*}=0,$ Then $s_{\left(y\right)i}^{-}=s_y^{*};$ If $b_{\left(y\right)i}^{*}=1,$ Then $s_{\left(y\right)i}^{+}=s_y^{*};$

Utilize above-mentioned formula when calculating the bit likelihood ratio, can determine according to declaring the result firmly

Or

Or

One, still, another one must search for to determine that each search all will be carried out

The computing of individual Euclidean distance, under high order modulation, the implementation complexity of this calculating is very high especially.

Summary of the invention

The object of the present invention is to provide a kind of soft demodulating method of low complexity for Gray quadrature amplitude modulation, the characteristics that it has are: utilize the rule declare firmly between symbol and all required constellation point of calculating bit likelihood ratio, directly obtain calculating the required constellation point of bit likelihood ratio by declaring symbol firmly, utilize the computing formula of bit likelihood ratio to calculate the bit likelihood ratio then, so this algorithm implementation complexity is lower.

In order to describe the content of this paper easily, at first make term definition:

Gray mappings M-QAM modulation: the M rank quadrature amplitude modulation that has gray mappings;

Gray mappings

Modulation: have gray mappings

The rank pulse amplitude modulation;

A kind of gray mappings M-QAM modulation provided by the invention is the soft demodulating method of low complex degree down, and it is as follows that it comprises the treatment step and the receiving end treatment step to received signal of making a start to transmitting, as shown in Figure 2.

Described making a start to the treatment step that transmits is:

Step 1: will import data and encode 1 successively, gray mappings M-QAM modulation 2, the signal that modulation obtains through the gray mappings M-QAM S that transmits exactly like this;

Described receiving end treatment step to received signal is:

Step 2: the S power normalization that will transmit through channel 3, obtains received signal r; Carry out channel estimating 4 by received signal r, obtain channel impulse a;

Step 3: according to channel impulse a to received signal r carry out equilibrium 5, obtain the signal r ' after the equilibrium, r '=x '+y ' i, wherein x ' is the real part of balanced back signal, y ' is the imaginary part of balanced back signal;

Step 4: the inverse that multiply by the power normalization factor through the balanced 5 signal r ' that obtain is reverted to modulation constellation points, and the real part x ' of then balanced back signal is a gray mappings

Modulation signal, the imaginary part y ' of balanced back signal also is a gray mappings

Modulation signal all carries out symbol to real part x ' and imaginary part y ' then and declares 6 firmly, obtains real part and declares symbol firmly $s_x^{*}=\{b_{\left(x\right)0}^{*}\·\·\·b_{\left(x\right)i-1}^{*}{b}_{\left(x\right)i}^{*}{b}_{\left(x\right)i+1}^{*}\·\·\·b_{\left(x\right)k-1}^{*}\}$ Declare symbol firmly with imaginary part $s_y^{*}=\{b_{\left(y\right)0}^{*}\·\·\·b_{\left(y\right)i-1}^{*}{b}_{\left(y\right)i}^{*}{b}_{\left(y\right)i+1}^{*}\·\·\·b_{\left(y\right)k-1}^{*}\},$ K is an integer, and i is a bit sequence, $1\≤k\≤{\log }_2\sqrt{M},$

Expression

I bit,

Expression

I bit;

Step 5: declare symbol firmly according to the signal real part x ' after the equilibrium and real part x '

Utilize the likelihood ratio LLR (b of each bit of formula (3) signal calculated real part _{(x) i}), as shown in Figure 3:

$\mathrm{LLR}\left(b_{\left(x\right)i}\right)=\frac{1}{{\mathrm{\σ}}^2}({|x\′-s_{\left(x\right)i}^{-}|}^2-{|x\′-s_{\left(x\right)i}^2)}^2|,0\≤i\≤k-1---\left(3\right)$

Wherein, σ ²Be the noise power of received signal, i bit is 0 and from the constellation point of signal real part x ' Euclidean distance minimum

With i bit be 1 and from the constellation point of signal real part x ' Euclidean distance minimum

The method of determining is as follows:

If $b_{\left(x\right)i}^{*}=0,$ Then $s_{\left(x\right)i}^{\text{-}}=s_x^{*,}$ And constellation point In preceding i-1 bit and real part declare symbol firmly

Identical, the i bit is that 1, the i+1 bit is 0, and other bits are 1 entirely, promptly $s_{\left(x\right)i}^{+}=\{b_{\left(x\right)0}^{*}...b_{\left(x\right)i-1}^{*}101\·\·\·1\};$

If $b_{\left(x\right)i}^{*}=1,$ Then $s_{\left(x\right)i}^{\text{+}}=s_x^{*,}$ And constellation point

In preceding i-1 bit and real part declare symbol firmly

Identical, the i bit is that 0, the i+1 bit is 0, and other bits are 1 entirely, promptly $s_{\left(x\right)i}^{-}=\{b_{\left(x\right)0}^{*}...b_{\left(x\right)i-1}^{*}001\·\·\·1\};$

Step 6: obtain declaring symbol firmly through balanced signal imaginary part y ' and imaginary part y ' according to step 4

Utilize formula (4) to calculate the likelihood ratio LLR (b of each bit of real part _{(y) i}), as shown in Figure 4:

$\mathrm{LLR}\left(b_{\left(y\right)i}\right)=\frac{1}{{\mathrm{\σ}}^2}({|y\′-s_{\left(y\right)i}^{-}|}^2-{|y\′-s_{\left(y\right)i}^2)}^2|,0\≤i\≤k-1---\left(4\right)$

Wherein, σ ²Be the noise power of received signal, i bit is 0 and from the constellation point of signal imaginary part y ' Euclidean distance minimum

With i bit be 1 and from the constellation point of signal imaginary part y ' Euclidean distance minimum

Determine that rule is as follows:

If $b_{\left(y\right)i}^{*}=0,$ Then $s_{\left(y\right)i}^{\text{-}}=s_y^{*,}$ Constellation point

In preceding i-1 bit and imaginary part declare symbol firmly

Identical, the i bit is that 1, the i+1 bit is 0, and other bits are 1 entirely, promptly $s_{\left(y\right)i}^{+}=\{b_{\left(y\right)0}^{*}...b_{\left(y\right)i-1}^{*}101\·\·\·1\};$

If $b_{\left(y\right)i}^{*}=1,$ Then $s_{\left(y\right)i}^{+}=s_y^{*,}$ Constellation point In preceding i-1 bit and imaginary part declare symbol firmly

Identical, the i bit is that 0, the i+1 bit is 0, and other bits are 1 entirely, promptly $s_{\left(y\right)i}^{-}=\{b_{\left(y\right)0}^{*}...b_{\left(y\right)i-1}^{*}001\·\·\·1\};$

Step 7: with the LLR (b that obtains in the step 5 _{(x) i}) as the bit likelihood ratio of received signal r odd bits, will obtain LLR (b in the step 6 _{(y) i}) as the bit likelihood ratio of received signal r even bit, the bit likelihood ratio of received signal r odd bits and the bit likelihood ratio of received signal r even bit are sent into decoding 8 judgement outputs together, can realize the soft demodulating method of low complexity for Gray quadrature amplitude modulation through above step.

Need to prove:

1) coding in step 1～step 71, gray mappings M-QAM modulation 2, channel 3, channel estimating 4, equilibrium 5, symbol declares 6 firmly, and conventional art (as shown in Figure 1) is all adopted in decoding 8;

2) step 5～step 6 is bit likelihood ratio computing modules 9 of low complex degree;

3) must adopt 3GPP predetermined rule in TR 25.848V4.0.0 standard in the gray mappings M-QAM module in the step 3, as shown in Figure 5; Details are seen TR25.848 V4.0.0;

4) gray mappings in the step 4

Document is seen in the modulation algorithm of declaring firmly of received signal correspondence down: and " C.K, Yuen, A fast analog to gray code converter, " in Proceedings of theIEEE, vol.65, pp.1510-1511, Oct.1977.

Essence of the present invention is:

Mirroring Mapping principle according to Gray code, utilize the rule of declaring firmly between symbol and the required constellation point of calculating bit likelihood ratio, directly obtain calculating the required constellation point of bit likelihood ratio by declaring symbol firmly, avoided determining constellation point with the method for minimum euclidean distance search in the conventional method, reduced implementation complexity, made it in high-speed mobile communication system, to be used widely;

Innovation part of the present invention is:

According to the characteristics of gray mappings M-QAM modulation signal, calculate the required constellation point of received signal bit likelihood ratio

With

Directly obtain by declaring symbol firmly, avoided determining to calculate the required constellation point of bit likelihood ratio with the method for minimum euclidean distance search in the conventional method, reduce computation complexity, made soft demodulating method in future mobile communication system, can be used widely.

The present invention compares with conventional method, has following characteristics:

When the inventive method was calculated the bit likelihood ratio, each symbol only needed basis to declare the required constellation point of likelihood ratio that symbol can obtain calculating all bits firmly, and the likelihood ratio of every each bit of symbol of traditional algorithm all needs to carry out the search of minimum euclidean distance, promptly

The calculating of inferior Euclidean distance, therefore, the inventive method has been avoided complicated search procedure, has significantly reduced computation complexity.

The invention has the beneficial effects as follows:

The required constellation point of likelihood ratio of calculating bit all can directly calculate by declaring symbol firmly, avoided the search of complicated minimum euclidean distance in the traditional algorithm, reduced implementation complexity, in hardware is realized, reduce cost, make that the soft demodulating method under the gray mappings M-QAM modulation can be used widely in high-speed mobile communication system.

Description of drawings

Fig. 1 is the system block diagram of traditional soft demodulation method

Wherein 1 is coding module, and 2 are gray mappings M-QAM modulation, and 3 is channel, and 4 is channel estimation module, and 5 is balance module, and 6 declare firmly for symbol, and 7 is bit likelihood ratio module, 8 decoding modules;

Fig. 2 is the system block diagram of soft demodulating method of the present invention

Wherein, 1 is coding module, and 2 are gray mappings M-QAM modulation, and 3 is channel, and 4 is channel estimation module, and 5 is balance module, and 6 declare firmly for symbol, and 9 calculate bit likelihood ratio module, 8 decoding modules for the present invention;

Fig. 3 be soft demodulating method real part bit likelihood ratio computing module flow chart of the present invention wherein, x ' expression is through the real part of balanced back signal,

The balanced back of expression signal real part x ' declares symbol firmly, and $s_x^{*}=\{b_{\left(x\right)0}^{*}\·\·\·b_{\left(x\right)i-1}^{*}{b}_{\left(x\right)i}^{*}{b}_{\left(x\right)i+1}^{*}\·\·\·b_{\left(x\right)k-1}^{*}\},$ I represents bit sequence, Represent i bit be 0 and after equilibrium the constellation point of signal real part x ' Euclidean distance minimum,

Represent i bit be 1 and after equilibrium the constellation point of signal real part x ' Euclidean distance minimum,

Expression

The i bit, LLR (b _{(x) i}) likelihood ratio of expression received signal real part i bit, $k={\log }_2\sqrt{M}$ The expression number of bits;

Fig. 4 is a soft demodulating method imaginary part bit likelihood ratio computing module flow chart of the present invention

Wherein, the imaginary part of the balanced back of y ' expression process signal,

The balanced back of expression signal imaginary part y ' declares symbol firmly, and $s_y^{*}=\{b_{\left(y\right)0}^{*}\·\·\·b_{\left(y\right)i-1}^{*}{b}_{\left(y\right)i}^{*}{b}_{\left(y\right)i+1}^{*}\·\·\·b_{\left(y\right)k-1}^{*}\},$ I represents bit sequence,

Represent i bit be 0 and after equilibrium signal imaginary part y ' Euclidean distance minimum constellation point and

Represent i bit be 1 and after equilibrium the constellation point of signal imaginary part y ' Euclidean distance minimum,

Expression

I bit, LLR (b _{(y) i}) likelihood ratio of expression received signal imaginary part i bit, $k={\log }_2\sqrt{M}$ The expression number of bits;

The Gray M-QAM mapping ruler of Fig. 5 for adopting 3GPP in TR25.848 V4.0.0 standard, to stipulate;

Embodiment

Provide the embodiment of a concrete this patent below, need to prove: the parameter in the following example does not influence the generality of this patent.

This implementation method has adopted emulation tool MATLAB;

One, makes a start

It is 300 that input signal carries out code length, and code check carries out gray mappings after being 1/2 LDPC coding, carries out the 16-QAM modulation then, after the signal power normalization after will modulating again through awgn channel;

Two, receiving end

Carry out channel estimating, equilibrium to received signal successively, the inverse that will the signal r '=x '+y ' i after equilibrium multiply by normalization factor reverts to modulation constellation points, then real part x ' is the 4-PAM modulation signal, imaginary part y ' is the 4-PAM modulation signal, again real part x ' is carried out symbol and declares firmly and declared symbol firmly

Imaginary part y ' is carried out symbol to be declared firmly and is declared symbol firmly Calculate the bit likelihood ratio according to step 5 and step 6 then, the bit likelihood ratio with received signal is input to decoding module at last, obtains judgement output by decoding;

By emulation relatively, this method is compared with conventional method, and the error rate of system performance is without any loss, but Adopt this method directly to obtain calculating all required constellation point of bit likelihood ratio by firmly declaring symbol, relatively To carry out the calculating of 4 Euclidean distances in each bit of conventional method and determine the star that the bit likelihood ratio is required The seat point has reduced implementation complexity.

Claims (1)

1, the soft demodulating method of low complex degree under a kind of gray mappings M-QAM modulation, it comprises the treatment step of making a start to transmitting,

Described making a start to the treatment step that transmits is:

Step 1: will import data and encode 1 successively, gray mappings M-QAM modulation 2, the signal that modulation obtains through the gray mappings M-QAM S that transmits exactly like this;

It is characterized in that it also comprises receiving end treatment step to received signal, described receiving end treatment step to received signal is:

Step 2: the S power normalization that will transmit through channel 3, obtains received signal r; Carry out channel estimating 4 by r to received signal, obtain channel impulse a;

Step 3: according to channel impulse a to received signal r carry out equilibrium 5, obtain the signal r ' after the equilibrium, r '=x '+y ' i, wherein x ' is the real part of balanced back signal, y ' is the imaginary part of balanced back signal;

Step 4: the inverse that multiply by the power normalization factor through the balanced 5 signal r ' that obtain is reverted to modulation constellation points, and the real part x ' of then balanced back signal is a gray mappings

Modulation signal, the imaginary part y ' of balanced back signal also is a gray mappings

Modulation signal all carries out symbol to real part x ' and imaginary part y ' then and declares 6 firmly, obtains real part and declares symbol firmly $s_x^{*}=\{b_{\left(x\right)0}^{*}\·\·\·b_{\left(x\right)i-1}^{*}{b}_{\left(x\right)i}^{*}{b}_{\left(x\right)i-1}^{*}\·\·\·b_{\left(x\right)k-1}^{*}\}$ Declare symbol firmly with imaginary part $s_y^{*}=\{b_{\left(y\right)0}^{*}\·\·\·b_{\left(y\right)i-1}^{*}{b}_{\left(y\right)i}^{*}{b}_{\left(y\right)i+1}^{*}\·\·\·b_{\left(y\right)k-1}^{*}\},$ K is an integer, and i is a bit sequence, $1\≤k\≤{\log }_2\sqrt{M},$

Expression

I bit,

Expression

I bit;

Step 5: declare symbol firmly according to the signal real part x ' after the equilibrium and real part x '

Utilize the likelihood ratio LLR (b of each bit of formula (3) signal calculated real part _{(x) i}):

$\mathrm{LLR}\left(b_{\left(x\right)i}\right)=\frac{1}{{\mathrm{\σ}}^2}({|x\′-s_{\left(x\right)i}^{-}|}^2-{|x\′-s_{\left(x\right)i}^{+}|}^2),0\≤i\≤k-1---\left(1\right)$

Wherein, σ ²Be the noise power of received signal, i bit is 0 and from the constellation point of signal real part x ' Euclidean distance minimum

With i bit be 1 and from the constellation point of signal real part x ' Euclidean distance minimum

The method of determining is as follows:

If $b_{\left(x\right)i}^{*}=0,$ Then $s_{\left(x\right)i}^{-}=s_x^{*},$ And constellation point

In preceding i-1 bit and real part declare symbol firmly Identical, the i bit is that 1, the i+1 bit is 0, and other bits are 1 entirely, promptly $s_{\left(x\right)i}^{+}=\{b_{\left(x\right)0}^{*}...b_{\left(x\right)i-1}^{*}101\·\·\·1\};$

If $b_{\left(x\right)i}^{*}=1,$ Then $s_{\left(x\right)i}^{+}=s_x^{*},$ And constellation point

In preceding i-1 bit and real part declare symbol firmly

Identical, the i bit is that 0, the i+1 bit is 0, and other bits are 1 entirely, promptly $s_{\left(x\right)i}^{-}=\{b_{\left(x\right)0}^{*}...b_{\left(x\right)i-1}^{*}001\·\·\·1\};$

Step 6: obtain declaring symbol firmly through balanced signal imaginary part y ' and signal imaginary part y ' according to step 4

Utilize formula (4) to calculate the likelihood ratio LLR (b of each bit of real part _{(y) i}),

$\mathrm{LLR}\left(b_{\left(y\right)i}\right)=\frac{1}{{\mathrm{\σ}}^2}({|y\′-s_{\left(y\right)i}^{-}|}^2-{|y\′-s_{\left(y\right)i}^{+}|}^2),0\≤i\≤k-1---\left(2\right)$

Wherein, σ ²Be the noise power of received signal, i bit is 0 and from the constellation point of signal imaginary part y ' Euclidean distance minimum

With i bit be 1 and from the constellation point of signal imaginary part y ' Euclidean distance minimum

Determine that rule is as follows:

If $b_{\left(y\right)i}^{*}=0,$ Then $s_{(\mathrm{yi}}^{-}=s_y^{*},$ Constellation point

In preceding i-1 bit and imaginary part declare symbol firmly

Identical, the i bit is that 1, the i+1 bit is 0, and other bits are 1 entirely, promptly $s_{\left(y\right)i}^{+}=\{b_{\left(y\right)0}^{*}...b_{\left(y\right)i-1}^{*}101\·\·\·1\};$

If $b_{\left(y\right)i}^{*}=1,$ Then $s_{\left(y\right)i}^{+}=s_y^{*},$ Constellation point

In preceding i-1 bit and imaginary part declare symbol firmly

Identical, the i bit is that 0, the i+1 bit is 0, and other bits are 1 entirely, promptly $s_{\left(y\right)i}^{-}=\{b_{\left(y\right)0}^{*}...b_{\left(y\right)i-1}^{*}001\·\·\·1\};$

Step 7: with the LLR (b that obtains in the step 5 _{(x) i}) as the bit likelihood ratio of received signal r odd bits, will obtain LLR (b in the step 6 _{(y) i}) as the bit likelihood ratio of received signal r even bit, the bit likelihood ratio of received signal r odd bits and the bit likelihood ratio of received signal r even bit are sent into decoding 8 judgement outputs together.

Images