CN102724163A - MQAM signal modulation recognition method based on state statistical advantage - Google Patents

Abstract

The invention belongs to the field of communication technology, and specially relates to a Mary Quadrature Amplitude Modulation (MQAM) signal recognition method. According to the method, a set of template is defined for each possible constellation diagram style; each constellation point-mode of to-be-recognized signal is counted; an evaluation function related to the signal-to-noise ratio of input signals is designed; and if the state of the to-be-recognized signal and the sample in the templates have matching advantages, the signal constellation diagram style can be determined so as to give the recognition result of the modulation mode. The recognition method provided by the invention needs a few of code elements number, only counts frequency of the signal points in the certain section where the components I and Q on a coordinate plane of the constellation diagram drop into, therefore, compared with the conventional method, the complexity of the recognition is greatly reduced and the recognition rate is improved. The method can be widely used in signal analysis, software-defined radio, wireless communication and other civilian and military communications systems.

Description

A kind of MQAM signal Modulation Identification method based on state advantage statistics

Technical field

The invention belongs to communication technical field, be specifically related to a kind of M-ary orthogonal amplitude modulation(PAM) MQAM signal recognition method.

Background technology

Modulation mode of communication signal identification is the important component part in signal analysis field, also is technologies of software radio, in civil and military communication, all has very high practical value.The quadrature amplitude modulation of high-order is widely used in many digital communication systems with its high spectrum utilization efficiency; Yet; But there is great difficulty in identification to the high-order QAM modulation signal, on the one hand, and along with the increase of order of modulation; The Euclidean distance of signal is more and more littler, and the division in judgement territory is also more difficult; On the other hand, computation complexity increases along with order of modulation and increases substantially.

Existing many high-order QAM modulation type recognition technologies mainly comprise based on the algorithm of feature extraction with based on two big types of the algorithms of maximum likelihood.In the algorithm based on feature extraction, at present commonly used is with planisphere as characteristic, adopts star map reconstruction algorithm such as subtractive clustering etc.; And under the situation of unknown order of modulation; Clustering algorithm is difficult to find suitable value, makes its requirement of satisfying high-order and low-order-modulated signal simultaneously, and when identification needs number of data points big; Amount of calculation can increase considerably, so this method recognition speed is slow and discrimination is low.In algorithm based on maximum likelihood; Realize classification though can utilize likelihood function to the QAM signal; But need more priori, comprise the carrier frequency, bit rate, symbol timing of signal etc., if there is unknown parameter; The sufficient statistic expression formula that will cause likelihood ratio to be classified is very complicated, amount of calculation is big, be difficult to real-time processing, therefore is difficult to be applied to real system.

Summary of the invention

The objective of the invention is to solve the existing problem that high-order QAM signal Modulation Identification method computation complexity is high, recognition speed is slow, discrimination is low, propose a kind of high-order QAM signal Modulation Identification method based on state advantage statistics based on planisphere.

Design philosophy based on the high-order QAM signal Modulation Identification method of state advantage statistics is: to all possible planisphere pattern definition one cover template; Add up the constellation point state of signal to be identified; Design a valuation functions relevant with the input signal signal to noise ratio; If the state of signal to be identified has the advantage coupling through certain sample in assessment and the template, get final product the planisphere pattern of decision signal, and then provide the Modulation Mode Recognition result.

The concrete steps of the inventive method are following:

If the complex radical band data of the MQAM signal that obtains are:

r(n)=I(n)+j×Q(n) n=1，2，…N

Wherein, n is current sampling point sequence number, and N is total sampling number.

Step 1, definition template: if the total M kind (formation set A) of possible MQAM modulation type, its corresponding exponent number is respectively

Then confirm template k=[k ₁, K ₂K _M], k _iValue confirm by (1) formula:

$k_i=\left\{\begin{array}{cc}{k}_{2m}& n_i=2m\\ \left\{\begin{array}{c}{k}_{2m+1}\\ 2^n\×k_5\end{array}\right.& \begin{array}{c}{n}_i=2m+1,n_i\≤5\\ n_i=5+2n\end{array}\end{array}\right.i=\mathrm{1,2},...M,---\left(1\right)$

Wherein, K _2mFor satisfying condition

Minimum positive integer, K _2m+1For satisfying condition

Minimum positive integer, m and n are positive integer.

Step 2, choose minimum standard frequency p _Std(0<p _Std<1) and statistics thresholding k _Th(0<k _Th<1), p _StdAnd k _ThBe the empirical value relevant with signal to noise ratio.

Step 3, r (n) is mapped to planisphere, obtains the constellation point coordinate and be:

R(n)=I _R(n)+j×Q _R(n) n=1，2，…N，

Judge I then _R(n), n=1,2 ... The value k that is complementary among N and the template k _x, concrete grammar is:

A. make initial index i _x=1;

B. get Statistics I _R(n) n=1,2 ... N falls into the interval

,

Interior number of samples does $N_R=\{{\mathrm{<figure-callout\; id="1"\; label="N"\; filenames="HDA00001869259300011.png,HDA00001869259300021.png"\; state="\{\{state\}\}">N</figure-callout>}}_1,{\mathrm{<figure-callout\; id="1"\; label="N"\; filenames="HDA00001869259300011.png,HDA00001869259300021.png"\; state="\{\{state\}\}">N</figure-callout>}}_1,...N_{i_x}\},$ Then each interval frequency does $p=\{{\mathrm{<figure-callout\; id="1"\; label="p"\; filenames="HDA00001869259300011.png,HDA00001869259300021.png"\; state="\{\{state\}\}">p</figure-callout>}}_1,...,p_{i_x}\}=\{\frac{N_1}{N},\frac{N_2}{N},...\frac{N_{i_x}}{N}\}.$ According to formula (2), an innings frequency p demands perfection _g

$p_g=\left(\begin{array}{cc}\frac{1}{k_{i_x}}\underset{i=1}{\overset{i_x}{\mathrm{\&Sigma;}}}{p}_i& k_{i_x}=2^n\\ \frac{1}{k_{i_x}}(\underset{i=1}{\overset{2^{n+1}}{\mathrm{\&Sigma;}}}{p}_i+\frac{3}{2}\underset{i=2^{n+1}+1}{\overset{i_x}{\mathrm{\&Sigma;}}}{p}_i)& k_{i_x}=2^n+2^{n+1}\end{array}\right),---\left(2\right)$

Wherein, n is 0 or positive integer;

C. again according to formula (3), calculate tmp (i _x),

For satisfying

Maximum integer,

D. if $p_g>\frac{p_{\mathrm{Std}}}{\mathrm{Tmp}\left(i_x\right)},$ Then judge $k_x=k_{i_x};$ If $p_g\&le;\frac{p_{\mathrm{Std}}}{\mathrm{Tmp}\left(i_x\right)},$ Then make i _x=i _x+ 1, the step of repetition b to d.

Step 4, judgement Q _R(n) n=1,2 ... The value k that is complementary among N and the template k _yConcrete grammar is:

A. make initial index i _y=1;

B. get

Statistics Q _R(n) n=1,2 ... N falls into the interval

,

Interior number of samples does $N_R=\{{\mathrm{<figure-callout\; id="1"\; label="N"\; filenames="HDA00001869259300011.png,HDA00001869259300021.png"\; state="\{\{state\}\}">N</figure-callout>}}_1,{\mathrm{<figure-callout\; id="1"\; label="N"\; filenames="HDA00001869259300011.png,HDA00001869259300021.png"\; state="\{\{state\}\}">N</figure-callout>}}_1,...N_{i_y}\},$ Then each interval frequency does $p=\{{\mathrm{<figure-callout\; id="1"\; label="p"\; filenames="HDA00001869259300011.png,HDA00001869259300021.png"\; state="\{\{state\}\}">p</figure-callout>}}_1,...,p_{i_y}\}=\{\frac{N_1}{N},\frac{N_2}{N},...\frac{N_{i_y}}{N}\}.$ According to formula (4), an innings frequency p demands perfection _g

$p_g=\left(\begin{array}{cc}\frac{1}{k_{i_y}}\underset{i=1}{\overset{i_y}{\mathrm{\&Sigma;}}}{p}_i& k_{i_y}=2^n\\ \frac{1}{k_{i_y}}(\underset{i=1}{\overset{2^{n+1}}{\mathrm{\&Sigma;}}}{p}_i+\frac{3}{2}\underset{i=2^{n+1}+1}{\overset{i_y}{\mathrm{\&Sigma;}}}{p}_i)& k_{i_y}=2^n+2^{n+1}\end{array}\right),---\left(4\right)$

Wherein, n is 0 or positive integer;

C. again according to formula (5), calculate tmp (i _y), For satisfying

Maximum integer,

D. if $p_g>\frac{p_{\mathrm{Std}}}{\mathrm{Tmp}\left(i_y\right)},$ Then judge $k_y=k_{i_y},$ If $p_g\&le;\frac{p_{\mathrm{Std}}}{\mathrm{Tmp}\left(i_y\right)},$ Then make i _y=i _y+ 1, the step of repetition b to d.

Step 5, the k that utilizes step 3 and step 4 to obtain _xAnd k _yAnd formula (6) is asked order of modulation M:

$M=\left\{\begin{array}{cc}4\&times;k_x\&times;k_y& k_x=k_y=2^{{\mathrm{<figure-callout\; id="0"\; label="n"\; filenames="HDA00001869259300021.png"\; state="\{\{state\}\}">n</figure-callout>}}_0}\\ 2^t& k_x=k_y=2^{{\mathrm{<figure-callout\; id="0"\; label="n"\; filenames="HDA00001869259300021.png"\; state="\{\{state\}\}">n</figure-callout>}}_0}+2^{{\mathrm{<figure-callout\; id="0"\; label="n"\; filenames="HDA00001869259300021.png"\; state="\{\{state\}\}">n</figure-callout>}}_0+1}\\ M_p& k_x\&NotEqual;k_y\end{array}\right.,---\left(6\right)$

Wherein, t is for satisfying condition 2 ^t≤4 * k _x* k _yMaximum integer, n ₀Be 0 or positive integer, M _pBe one and n _MRelevant variable.Ask satisfied

N _M, if

With

All belong to A, then M _pEach gets M with 0.5 probability ₁Perhaps M ₂Otherwise, M _p=min||k _i-4 * k _x* k _y|| ²I=1,2 ... M.

Beneficial effect

The present invention propose based on the high-order orthogonal amplitude modulation(PAM) MQAM signal Modulation Identification method of state advantage statistics with respect to prior art, discern needed number of symbols still less.Simultaneously, method of the present invention only need be added up the frequency that I, Q component on the planisphere coordinate plane fall into the signaling point in certain interval, and existing relatively method greatly reduces the complexity of identification, has improved discrimination.This method can be widely used in that signal analysis, software radio, radio communication etc. are civilian, in the military communication system.

Description of drawings

Fig. 1 flow chart based on state advantage statistical recognition method of the present invention;

Fig. 2 is based on subtractive clustering method recognition result figure;

Recognition result figure in Fig. 3 the inventive method embodiment;

Embodiment

Below in conjunction with accompanying drawing and embodiment the present invention is described further.

The flow process of MQAM modulation signal Modulation Identification method of the present invention is as shown in Figure 1.In this instance, signal modulation system set A is: and 4QAM, 16QAM, 64QAM, 256QAM, 1024QAM}, modulation system to be identified is 16QAM, the baseband signal sampling point is synchronous fully, and signal to noise ratio is 10dB.Concrete design procedure is following:

If: the base band MQAM signal sampling point of acquisition is:

r(n)=I(n)+j×Q(n) n=1，2，…6000，

5 kinds of modulation types to be selected are arranged among step 1, the A, and its exponent number is respectively { 2 ², 2 ⁴, 2 ⁶, 2 ⁸, 2 ¹⁰, then confirm template k=[1,2,4,8,16].

Step 2, make minimum standard frequency p _Std=0.75, statistics thresholding k _Th=0.6.

Step 3, r (n) sequence constellation point coordinate are R (n)=I _R(n)+j * Q _R(n) n=1,2 ... 6000, judge I _R(n) n=1,2 ... Which value among the N matching template k, making it is k _xJudge as follows:

A. make initial index i _x=1;

B. get

Statistics I _R(n) n=1,2 ... 6000 fall into the interval

Interior number of samples is N ₁=2700, then the frequency does

According to formula (2), get overall frequency p _g=0.45;

C. by formula (3), tmp (i _x)=1;

d.

p _g=0.45。Obviously

Be false, make i _x=i _x+ 1=2;

Get

Statistics I _R(n) n=1,2 ... 6000 fall into the interval

$[\frac{{\mathrm{<figure-callout\; id="1"\; label="k"\; filenames="HDA00001869259300011.png,HDA00001869259300021.png"\; state="\{\{state\}\}">k</figure-callout>}}_1}{2}(1-k_{\mathrm{Th}}),\frac{{\mathrm{<figure-callout\; id="1"\; label="k"\; filenames="HDA00001869259300011.png,HDA00001869259300021.png"\; state="\{\{state\}\}">k</figure-callout>}}_1}{2}(1+k_{\mathrm{Th}})]=\left[\mathrm{0.2,0.8}\right],$ $[\frac{k_{i_x}}{2}(1-k_{\mathrm{Th}}),\frac{k_{i_x}}{2}(1+k_{\mathrm{Th}})]=\left[\mathrm{0.4,1.6}\right]$ Interior number of samples is { N ₁=2700, N ₂=2348}, then the frequency is p={0.45,0.3913}.According to formula (2), get overall frequency p _g=0.4207.By formula (3), this moment tmp (i _x)=2,

Obviously $p_g>\frac{p_{\mathrm{Std}}}{\mathrm{Tmp}\left(i_x\right)}$ Set up, promptly $k_x=k_{i_x}=2.$

Step 4, judgement Q _R(n) n=1,2 ... Which value among the N matching template k, making it is k _yJudge as follows:

A. make initial index i _y=1;

B. get Statistics Q _R(n) n=1,2 ... The number of samples that N falls in the interval [0.2,0.8] is N ₁=2675, the frequency then According to formula (4), get overall frequency p _g=0.4458;

C. by formula (5), tmp (i _y)=1;

d.

p _g=0.4458。Obviously

Be false, make i _y=i _y+ 1=2;

Get

Statistics Q _R(n) n=1,2 ... 6000 fall into the interval

$[\frac{{\mathrm{<figure-callout\; id="1"\; label="k"\; filenames="HDA00001869259300011.png,HDA00001869259300021.png"\; state="\{\{state\}\}">k</figure-callout>}}_1}{2}(1-k_{\mathrm{Th}}),\frac{{\mathrm{<figure-callout\; id="1"\; label="k"\; filenames="HDA00001869259300011.png,HDA00001869259300021.png"\; state="\{\{state\}\}">k</figure-callout>}}_1}{2}(1+k_{\mathrm{Th}})]=\left[\mathrm{0.2,0.8}\right],$ $[\frac{k_{i_y}}{2}(1-k_{\mathrm{Th}}),\frac{k_{i_y}}{2}(1+k_{\mathrm{Th}})]=\left[\mathrm{0.4,1.6}\right]$ Interior number of samples is { N ₁=2612, N ₂=2791}, then the frequency is p={0.4353,0.4652}.According to formula (4), get overall frequency p _g=0.4503.By formula (5), this moment tmp (i _y)=2, Obviously $p_g>\frac{p_{\mathrm{Std}}}{\mathrm{Tmp}\left(i_x\right)}$ Set up, promptly $k_x=k_{i_y}=2.$

Step 6, because k _x=k _y=2, get M=16 by formula (6).

Be the validity of checking the inventive method, employing table 1 simulated environment is carried out emulation

Table 1 simulated environment

The simulation parameter that is provided with based on table 1 carries out emulation.Fig. 2, Fig. 3 are respectively under additive gaussian white noise channel, the different signal to noise ratio condition, and existing algorithm and the inventive method based on subtractive clustering is to the discrimination of different modulating mode.Fig. 3 shows that when signal to noise ratio was higher than 10dB, all signals all can be realized 100% identification among the A; Because based on subtractive clustering method too complex when 1024QAM discerns among the A, so only provide the discrimination of preceding 3 kinds of letters under different state of signal-to-noise among the A among Fig. 2.Obviously, during identical signal to noise ratio, the discrimination of each signal all will be lower than the inventive method, shows the validity of this method.

Claims (2)

1. MQAM signal Modulation Identification method based on state advantage statistics is characterized in that:

The complex radical band data of the MQAM signal that obtains are:

r(n)＝I(n)+j×Q(n)n＝1，2，...N

Wherein, n is current sampling point sequence number, and N is total sampling number; Specifically comprise the steps:

Step 1, definition template: if the total M kind of the MQAM modulation type of system constitutes set A, its corresponding exponent number is respectively

Then confirm template k=[k ₁, k ₂... k _M], k _iValue be:

$k_i=\left\{\begin{array}{cc}{k}_{2m}& n_i=2m\\ \left\{\begin{array}{c}{k}_{2m+1}\\ 2^n\&times;k_5\end{array}\right.& \begin{array}{c}{n}_i=2m+1,n_i\&le;5\\ n_i=5+2n\end{array}\end{array}\right.i=\mathrm{1,2},...M,---\left(1\right)$

Wherein, k _2mFor satisfying condition Minimum positive integer, k _2m+1For satisfying condition

Minimum positive integer, m and n are positive integer;

Step 2, choose minimum standard frequency P _StdWith statistics thresholding K _Th

Step 3, r (n) is mapped to planisphere, obtains the constellation point coordinate and be:

R(n)＝I _R(n)+j×Q _R(n)n＝1，2，...N

Judge I then _R(n) with template k in the value K that is complementary _x, concrete grammar is:

A. make initial index i _x=1;

B. get

Statistics I _R(n) n=1,2 ... N falls into the interval

$[\frac{k_2}{2}(1-k_{\mathrm{Th}}),\frac{k_2}{2}(1+k_{\mathrm{Th}})],......,[\frac{k_{i_x}}{2}(1-k_{\mathrm{Th}}),\frac{k_{i_x}}{2}(1+k_{\mathrm{Th}})]$ Interior number of samples does $N_R=\{N_1,N_2,...N_{i_x}\},$ Then each interval frequency does $p=\{p_1,...,p_{i_x}\}=\{\frac{N_1}{N},\frac{N_2}{N},...\frac{N_{i_x}}{N}\};$ An innings frequency P demands perfection _g:

$p_g=\left\{\begin{array}{cc}\frac{1}{k_{i_x}}\underset{i=1}{\overset{i_x}{\mathrm{\&Sigma;}}}{p}_i& k_{i_x}=2^n\\ \frac{1}{k_{i_x}}(\underset{i=1}{\overset{2^{n+1}}{\mathrm{\&Sigma;}}}{p}_i+\frac{3}{2}\underset{i=2^{n+1}+1}{\overset{i_x}{\mathrm{\&Sigma;}}}{p}_i)& k_{i_x}=2^n+2^{n+1}\end{array}\right.,---\left(2\right)$

Wherein, n is 0 or positive integer;

C. calculate tmp (i _x):

In order to meet

the largest integer;

D. if $p_g>\frac{p_{\mathrm{Std}}}{\mathrm{Tmp}\left(i_x\right)},$ Then judge $k_x=k_{i_x};$ If $p_g\&le;\frac{p_{\mathrm{Std}}}{\mathrm{Tmp}\left(i_x\right)},$ Then make i _x=i _x+ 1, repeat

The step of b to d;

Step 4, judgement Q _R(n) with template k in the value k that is complementary _y, concrete grammar is:

A. make initial index i _y=1;

B. get

Statistics Q _R(n) n=1,2 ... N falls into the interval

$[\frac{k_2}{2}(1-k_{\mathrm{Th}}),\frac{k_2}{2}(1+k_{\mathrm{Th}})],......,[\frac{k_{i_y}}{2}(1-k_{\mathrm{Th}}),\frac{k_{i_y}}{2}(1+k_{\mathrm{Th}})]$ Interior number of samples does $N_R=\{N_1,N_2,...N_{i_y}\},$ Then each interval frequency does $p=\{p_1,...,p_{i_y}\}=\{\frac{N_1}{N},\frac{N_2}{N},...\frac{N_{i_y}}{N}\},$ An innings frequency P demands perfection _g:

$p_g=\left\{\begin{array}{cc}\frac{1}{k_{i_y}}\underset{i=1}{\overset{i_y}{\mathrm{\&Sigma;}}}{p}_i& k_{i_y}=2^n\\ \frac{1}{k_{i_y}}(\underset{i=1}{\overset{2^{n+1}}{\mathrm{\&Sigma;}}}{p}_i+\frac{3}{2}\underset{i=2^{n+1}+1}{\overset{i_y}{\mathrm{\&Sigma;}}}{p}_i)& k_{i_y}=2^n+2^{n+1}\end{array}\right.,---\left(4\right)$

Wherein, n is 0 or positive integer;

C. calculate tmp (i _y):

In order to meet the largest integer;

D. if $p_g>\frac{p_{\mathrm{Std}}}{\mathrm{Tmp}\left(i_y\right)},$ Then judge $k_x=k_{i_y},$ If $p_g\&le;\frac{p_{\mathrm{Std}}}{\mathrm{Tmp}\left(i_y\right)},$ Then make i _y=i _y+ 1, the step of repetition b to d;

Step 5, the k that utilizes step 3 and step 4 to obtain _xAnd k _y, ask order of modulation M:

$M=\left\{\begin{array}{cc}4\&times;k_x\&times;k_y& k_x=k_y=2^{n_0}\\ 2^t& k_x=k_y=2^{n_0}+2^{n_0+1}\\ M_p& k_x\&NotEqual;k_y\end{array}\right.,---\left(6\right)$

Wherein, t is the 2t that satisfies condition≤4 * k _x* k _yMaximum integer, n ₀Be 0 or positive integer, M _pBe one and n _MRelevant variable; Ask satisfied $2^{n_M}<4\&times;k_x\&times;k_y<2^{n_M+1}$ N _M, if $M_1=2^{n_M}$ With $M_2=2^{n_M+1}$ All belong to A, then M _pEach gets M with 0.5 probability ₁Or M ₂Otherwise, M _p=min ‖ k _i-4 * k _x* k _y‖ ²I=1,2 ... M;

Thereby obtain order of modulation M.

2. a kind of MQAM signal Modulation Identification method based on state advantage statistics according to claim 1 is characterized in that: p _StdAnd K _ThBe the empirical value relevant with signal to noise ratio, 0＜p _Std＜1,0＜k _Th＜1.

Images